test: update unit tests to use Vitest framework

- Refactored test files for data source options, drawer, and pipeline settings to utilize Vitest for improved testing capabilities.
- Ensured consistent testing practices across components by importing necessary Vitest functions.

api/controllers/console/__init__.py

@@ -39,7 +39,6 @@ from . import (
     feature,
     human_input_form,
     init_validate,
-    notification,
     ping,
     setup,
     spec,
@@ -185,7 +184,6 @@ __all__ = [
     "model_config",
     "model_providers",
     "models",
-    "notification",
     "oauth",
     "oauth_server",
     "ops_trace",

api/controllers/console/admin.py

@@ -1,5 +1,3 @@
-import csv
-import io
 from collections.abc import Callable
 from functools import wraps
 from typing import ParamSpec, TypeVar
@@ -8,7 +6,7 @@ from flask import request
 from flask_restx import Resource
 from pydantic import BaseModel, Field, field_validator
 from sqlalchemy import select
-from werkzeug.exceptions import BadRequest, NotFound, Unauthorized
+from werkzeug.exceptions import NotFound, Unauthorized
 
 from configs import dify_config
 from constants.languages import supported_language
@@ -18,7 +16,6 @@ from core.db.session_factory import session_factory
 from extensions.ext_database import db
 from libs.token import extract_access_token
 from models.model import App, ExporleBanner, InstalledApp, RecommendedApp, TrialApp
-from services.billing_service import BillingService
 
 P = ParamSpec("P")
 R = TypeVar("R")
@@ -280,115 +277,3 @@ class DeleteExploreBannerApi(Resource):
         db.session.commit()
 
         return {"result": "success"}, 204
-
-
-class SaveNotificationContentPayload(BaseModel):
-    content: str = Field(...)
-
-
-class SaveNotificationUserPayload(BaseModel):
-    user_email: list[str] = Field(...)
-
-
-console_ns.schema_model(
-    SaveNotificationContentPayload.__name__,
-    SaveNotificationContentPayload.model_json_schema(ref_template=DEFAULT_REF_TEMPLATE_SWAGGER_2_0),
-)
-
-console_ns.schema_model(
-    SaveNotificationUserPayload.__name__,
-    SaveNotificationUserPayload.model_json_schema(ref_template=DEFAULT_REF_TEMPLATE_SWAGGER_2_0),
-)
-
-
-@console_ns.route("/admin/save_notification_content")
-class SaveNotificationContentApi(Resource):
-    @console_ns.doc("save_notification_content")
-    @console_ns.doc(description="Save a notification content")
-    @console_ns.expect(console_ns.models[SaveNotificationContentPayload.__name__])
-    @console_ns.response(200, "Notification content saved successfully")
-    @only_edition_cloud
-    @admin_required
-    def post(self):
-        payload = SaveNotificationContentPayload.model_validate(console_ns.payload)
-        BillingService.save_notification_content(payload.content)
-        return {"result": "success"}, 200
-
-
-@console_ns.route("/admin/save_notification_user")
-class SaveNotificationUserApi(Resource):
-    @console_ns.doc("save_notification_user")
-    @console_ns.doc(
-        description="Save notification users via JSON body or file upload. "
-        'JSON: {"user_email": ["a@example.com", ...]}. '
-        "File: multipart/form-data with a 'file' field (CSV or TXT, one email per line)."
-    )
-    @console_ns.response(200, "Notification users saved successfully")
-    @only_edition_cloud
-    @admin_required
-    def post(self):
-        # Determine input mode: file upload or JSON body
-        if "file" in request.files:
-            emails = self._parse_emails_from_file()
-        else:
-            payload = SaveNotificationUserPayload.model_validate(console_ns.payload)
-            emails = payload.user_email
-
-        if not emails:
-            raise BadRequest("No valid email addresses provided.")
-
-        # Use batch API for bulk insert (chunks of 1000 per request to billing service)
-        result = BillingService.save_notification_users_batch(emails)
-
-        return {
-            "result": "success",
-            "total": len(emails),
-            "succeeded": result["succeeded"],
-            "failed_chunks": result["failed_chunks"],
-        }, 200
-
-    @staticmethod
-    def _parse_emails_from_file() -> list[str]:
-        """Parse email addresses from an uploaded CSV or TXT file."""
-        file = request.files["file"]
-
-        if not file.filename:
-            raise BadRequest("Uploaded file has no filename.")
-
-        filename_lower = file.filename.lower()
-        if not filename_lower.endswith((".csv", ".txt")):
-            raise BadRequest("Invalid file type. Only CSV (.csv) and TXT (.txt) files are allowed.")
-
-        # Read file content
-        try:
-            content = file.read().decode("utf-8")
-        except UnicodeDecodeError:
-            try:
-                file.seek(0)
-                content = file.read().decode("gbk")
-            except UnicodeDecodeError:
-                raise BadRequest("Unable to decode the file. Please use UTF-8 or GBK encoding.")
-
-        emails: list[str] = []
-        if filename_lower.endswith(".csv"):
-            reader = csv.reader(io.StringIO(content))
-            for row in reader:
-                for cell in row:
-                    cell = cell.strip()
-                    emails.append(cell)
-        else:
-            # TXT file: one email per line
-            for line in content.splitlines():
-                line = line.strip()
-                emails.append(line)
-
-        # Deduplicate while preserving order
-        seen: set[str] = set()
-        unique_emails: list[str] = []
-        for email in emails:
-            email_lower = email.lower()
-            if email_lower not in seen:
-                seen.add(email_lower)
-                unique_emails.append(email)
-
-        return unique_emails

api/controllers/console/notification.py

@@ -1,26 +0,0 @@
-from flask_restx import Resource
-
-from controllers.console import console_ns
-from controllers.console.wraps import account_initialization_required, only_edition_cloud, setup_required
-from libs.login import current_account_with_tenant, login_required
-from services.billing_service import BillingService
-
-
-@console_ns.route("/notification")
-class NotificationApi(Resource):
-    @console_ns.doc("get_notification")
-    @console_ns.doc(description="Get notification for the current user")
-    @console_ns.doc(
-        responses={
-            200: "Success",
-            401: "Unauthorized",
-        }
-    )
-    @setup_required
-    @login_required
-    @account_initialization_required
-    @only_edition_cloud
-    def get(self):
-        current_user, _ = current_account_with_tenant()
-        notification = BillingService.read_notification(current_user.email)
-        return notification

api/services/billing_service.py

@@ -393,35 +393,3 @@ class BillingService:
         for item in data:
             tenant_whitelist.append(item["tenant_id"])
         return tenant_whitelist
-
-    @classmethod
-    def read_notification(cls, user_email: str):
-        params = {"user_email": user_email}
-        return cls._send_request("GET", "/notification/read", params=params)
-
-    @classmethod
-    def save_notification_user(cls, user_email: str):
-        json = {"user_email": user_email}
-        return cls._send_request("POST", "/notification/new-notification-user", json=json)
-
-    @classmethod
-    def save_notification_users_batch(cls, user_emails: list[str]) -> dict:
-        """Batch save notification users in chunks of 1000."""
-        chunk_size = 1000
-        total_succeeded = 0
-        failed_chunks: list[dict] = []
-
-        for i in range(0, len(user_emails), chunk_size):
-            chunk = user_emails[i : i + chunk_size]
-            try:
-                resp = cls._send_request("POST", "/notification/batch-notification-users", json={"user_emails": chunk})
-                total_succeeded += resp.get("count", len(chunk))
-            except Exception as e:
-                failed_chunks.append({"offset": i, "count": len(chunk), "error": str(e)})
-
-        return {"succeeded": total_succeeded, "failed_chunks": failed_chunks}
-
-    @classmethod
-    def save_notification_content(cls, content: str):
-        json = {"content": content}
-        return cls._send_request("POST", "/notification/new-notification", json=json)

web/__tests__/datasets/create-dataset-flow.test.tsx

@@ -0,0 +1,300 @@
+/**
+ * Integration Test: Create Dataset Flow
+ *
+ * Tests cross-module data flow: step-one data → step-two hooks → creation params → API call
+ * Validates data contracts between steps.
+ */
+
+import type { CustomFile } from '@/models/datasets'
+import type { RetrievalConfig } from '@/types/app'
+import { act, renderHook } from '@testing-library/react'
+import { ChunkingMode, DataSourceType, ProcessMode } from '@/models/datasets'
+import { RETRIEVE_METHOD } from '@/types/app'
+
+const mockCreateFirstDocument = vi.fn()
+const mockCreateDocument = vi.fn()
+vi.mock('@/service/knowledge/use-create-dataset', () => ({
+  useCreateFirstDocument: () => ({ mutateAsync: mockCreateFirstDocument, isPending: false }),
+  useCreateDocument: () => ({ mutateAsync: mockCreateDocument, isPending: false }),
+  getNotionInfo: (pages: { page_id: string }[], credentialId: string) => ({
+    workspace_id: 'ws-1',
+    pages: pages.map(p => p.page_id),
+    notion_credential_id: credentialId,
+  }),
+  getWebsiteInfo: (opts: { websitePages: { url: string }[], websiteCrawlProvider: string }) => ({
+    urls: opts.websitePages.map(p => p.url),
+    only_main_content: true,
+    provider: opts.websiteCrawlProvider,
+  }),
+}))
+
+vi.mock('@/service/knowledge/use-dataset', () => ({
+  useInvalidDatasetList: () => vi.fn(),
+}))
+
+vi.mock('@/app/components/base/toast', () => ({
+  default: { notify: vi.fn() },
+}))
+
+vi.mock('@/app/components/base/amplitude', () => ({
+  trackEvent: vi.fn(),
+}))
+
+// Import hooks after mocks
+const { useSegmentationState, DEFAULT_SEGMENT_IDENTIFIER, DEFAULT_MAXIMUM_CHUNK_LENGTH, DEFAULT_OVERLAP }
+  = await import('@/app/components/datasets/create/step-two/hooks')
+const { useDocumentCreation, IndexingType }
+  = await import('@/app/components/datasets/create/step-two/hooks')
+
+const createMockFile = (overrides?: Partial<CustomFile>): CustomFile => ({
+  id: 'file-1',
+  name: 'test.txt',
+  type: 'text/plain',
+  size: 1024,
+  extension: '.txt',
+  mime_type: 'text/plain',
+  created_at: 0,
+  created_by: '',
+  ...overrides,
+} as CustomFile)
+
+describe('Create Dataset Flow - Cross-Step Data Contract', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Step-One → Step-Two: Segmentation Defaults', () => {
+    it('should initialise with correct default segmentation values', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      expect(result.current.segmentIdentifier).toBe(DEFAULT_SEGMENT_IDENTIFIER)
+      expect(result.current.maxChunkLength).toBe(DEFAULT_MAXIMUM_CHUNK_LENGTH)
+      expect(result.current.overlap).toBe(DEFAULT_OVERLAP)
+      expect(result.current.segmentationType).toBe(ProcessMode.general) // 'custom'
+    })
+
+    it('should produce valid process rule for general chunking', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      const processRule = result.current.getProcessRule(ChunkingMode.text)
+
+      // mode should be segmentationType = ProcessMode.general = 'custom'
+      expect(processRule.mode).toBe('custom')
+      expect(processRule.rules.segmentation).toEqual({
+        separator: '\n\n', // unescaped from \\n\\n
+        max_tokens: DEFAULT_MAXIMUM_CHUNK_LENGTH,
+        chunk_overlap: DEFAULT_OVERLAP,
+      })
+      // rules is empty initially since no default config loaded
+      expect(processRule.rules.pre_processing_rules).toEqual([])
+    })
+
+    it('should produce valid process rule for parent-child chunking', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      const processRule = result.current.getProcessRule(ChunkingMode.parentChild)
+
+      expect(processRule.mode).toBe('hierarchical')
+      expect(processRule.rules.parent_mode).toBe('paragraph')
+      expect(processRule.rules.segmentation).toEqual({
+        separator: '\n\n',
+        max_tokens: 1024,
+      })
+      expect(processRule.rules.subchunk_segmentation).toEqual({
+        separator: '\n',
+        max_tokens: 512,
+      })
+    })
+  })
+
+  describe('Step-Two → Creation API: Params Building', () => {
+    it('should build valid creation params for file upload workflow', () => {
+      const files = [createMockFile()]
+      const { result: segResult } = renderHook(() => useSegmentationState())
+      const { result: creationResult } = renderHook(() =>
+        useDocumentCreation({
+          dataSourceType: DataSourceType.FILE,
+          files,
+          notionPages: [],
+          notionCredentialId: '',
+          websitePages: [],
+        }),
+      )
+
+      const processRule = segResult.current.getProcessRule(ChunkingMode.text)
+      const retrievalConfig: RetrievalConfig = {
+        search_method: RETRIEVE_METHOD.semantic,
+        reranking_enable: false,
+        reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+        top_k: 3,
+        score_threshold_enabled: false,
+        score_threshold: 0,
+      }
+
+      const params = creationResult.current.buildCreationParams(
+        ChunkingMode.text,
+        'English',
+        processRule,
+        retrievalConfig,
+        { provider: 'openai', model: 'text-embedding-ada-002' },
+        IndexingType.QUALIFIED,
+      )
+
+      expect(params).not.toBeNull()
+      // File IDs come from file.id (not file.file.id)
+      expect(params!.data_source.type).toBe(DataSourceType.FILE)
+      expect(params!.data_source.info_list.file_info_list?.file_ids).toContain('file-1')
+
+      expect(params!.indexing_technique).toBe(IndexingType.QUALIFIED)
+      expect(params!.doc_form).toBe(ChunkingMode.text)
+      expect(params!.doc_language).toBe('English')
+      expect(params!.embedding_model).toBe('text-embedding-ada-002')
+      expect(params!.embedding_model_provider).toBe('openai')
+      expect(params!.process_rule.mode).toBe('custom')
+    })
+
+    it('should validate params: overlap must not exceed maxChunkLength', () => {
+      const { result } = renderHook(() =>
+        useDocumentCreation({
+          dataSourceType: DataSourceType.FILE,
+          files: [createMockFile()],
+          notionPages: [],
+          notionCredentialId: '',
+          websitePages: [],
+        }),
+      )
+
+      // validateParams returns false (invalid) when overlap > maxChunkLength for general mode
+      const isValid = result.current.validateParams({
+        segmentationType: 'general',
+        maxChunkLength: 100,
+        limitMaxChunkLength: 4000,
+        overlap: 200, // overlap > maxChunkLength
+        indexType: IndexingType.QUALIFIED,
+        embeddingModel: { provider: 'openai', model: 'text-embedding-ada-002' },
+        rerankModelList: [],
+        retrievalConfig: {
+          search_method: RETRIEVE_METHOD.semantic,
+          reranking_enable: false,
+          reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+          top_k: 3,
+          score_threshold_enabled: false,
+          score_threshold: 0,
+        },
+      })
+      expect(isValid).toBe(false)
+    })
+
+    it('should validate params: maxChunkLength must not exceed limit', () => {
+      const { result } = renderHook(() =>
+        useDocumentCreation({
+          dataSourceType: DataSourceType.FILE,
+          files: [createMockFile()],
+          notionPages: [],
+          notionCredentialId: '',
+          websitePages: [],
+        }),
+      )
+
+      const isValid = result.current.validateParams({
+        segmentationType: 'general',
+        maxChunkLength: 5000,
+        limitMaxChunkLength: 4000, // limit < maxChunkLength
+        overlap: 50,
+        indexType: IndexingType.QUALIFIED,
+        embeddingModel: { provider: 'openai', model: 'text-embedding-ada-002' },
+        rerankModelList: [],
+        retrievalConfig: {
+          search_method: RETRIEVE_METHOD.semantic,
+          reranking_enable: false,
+          reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+          top_k: 3,
+          score_threshold_enabled: false,
+          score_threshold: 0,
+        },
+      })
+      expect(isValid).toBe(false)
+    })
+  })
+
+  describe('Full Flow: Segmentation State → Process Rule → Creation Params Consistency', () => {
+    it('should keep segmentation values consistent across getProcessRule and buildCreationParams', () => {
+      const files = [createMockFile()]
+      const { result: segResult } = renderHook(() => useSegmentationState())
+      const { result: creationResult } = renderHook(() =>
+        useDocumentCreation({
+          dataSourceType: DataSourceType.FILE,
+          files,
+          notionPages: [],
+          notionCredentialId: '',
+          websitePages: [],
+        }),
+      )
+
+      // Change segmentation settings
+      act(() => {
+        segResult.current.setMaxChunkLength(2048)
+        segResult.current.setOverlap(100)
+      })
+
+      const processRule = segResult.current.getProcessRule(ChunkingMode.text)
+      expect(processRule.rules.segmentation.max_tokens).toBe(2048)
+      expect(processRule.rules.segmentation.chunk_overlap).toBe(100)
+
+      const params = creationResult.current.buildCreationParams(
+        ChunkingMode.text,
+        'Chinese',
+        processRule,
+        {
+          search_method: RETRIEVE_METHOD.semantic,
+          reranking_enable: false,
+          reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+          top_k: 3,
+          score_threshold_enabled: false,
+          score_threshold: 0,
+        },
+        { provider: 'openai', model: 'text-embedding-ada-002' },
+        IndexingType.QUALIFIED,
+      )
+
+      expect(params).not.toBeNull()
+      expect(params!.process_rule.rules.segmentation.max_tokens).toBe(2048)
+      expect(params!.process_rule.rules.segmentation.chunk_overlap).toBe(100)
+      expect(params!.doc_language).toBe('Chinese')
+    })
+
+    it('should support parent-child mode through the full pipeline', () => {
+      const files = [createMockFile()]
+      const { result: segResult } = renderHook(() => useSegmentationState())
+      const { result: creationResult } = renderHook(() =>
+        useDocumentCreation({
+          dataSourceType: DataSourceType.FILE,
+          files,
+          notionPages: [],
+          notionCredentialId: '',
+          websitePages: [],
+        }),
+      )
+
+      const processRule = segResult.current.getProcessRule(ChunkingMode.parentChild)
+      const params = creationResult.current.buildCreationParams(
+        ChunkingMode.parentChild,
+        'English',
+        processRule,
+        {
+          search_method: RETRIEVE_METHOD.semantic,
+          reranking_enable: false,
+          reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+          top_k: 3,
+          score_threshold_enabled: false,
+          score_threshold: 0,
+        },
+        { provider: 'openai', model: 'text-embedding-ada-002' },
+        IndexingType.QUALIFIED,
+      )
+
+      expect(params).not.toBeNull()
+      expect(params!.doc_form).toBe(ChunkingMode.parentChild)
+      expect(params!.process_rule.mode).toBe('hierarchical')
+      expect(params!.process_rule.rules.parent_mode).toBe('paragraph')
+      expect(params!.process_rule.rules.subchunk_segmentation).toBeDefined()
+    })
+  })
+})

web/__tests__/datasets/dataset-settings-flow.test.tsx

@@ -0,0 +1,451 @@
+/**
+ * Integration Test: Dataset Settings Flow
+ *
+ * Tests cross-module data contracts in the dataset settings form:
+ *   useFormState hook ↔ index method config ↔ retrieval config ↔ permission state.
+ *
+ * The unit-level use-form-state.spec.ts validates the hook in isolation.
+ * This integration test verifies that changing one configuration dimension
+ * correctly cascades to dependent parts (index method → retrieval config,
+ * permission → member list visibility, embedding model → embedding available state).
+ */
+
+import type { DataSet } from '@/models/datasets'
+import type { RetrievalConfig } from '@/types/app'
+import { act, renderHook, waitFor } from '@testing-library/react'
+import { IndexingType } from '@/app/components/datasets/create/step-two'
+import { ChunkingMode, DatasetPermission, DataSourceType, WeightedScoreEnum } from '@/models/datasets'
+import { RETRIEVE_METHOD } from '@/types/app'
+
+// --- Mocks ---
+
+const mockMutateDatasets = vi.fn()
+const mockInvalidDatasetList = vi.fn()
+const mockUpdateDatasetSetting = vi.fn().mockResolvedValue({})
+
+vi.mock('@/context/app-context', () => ({
+  useSelector: () => false,
+}))
+
+vi.mock('@/service/datasets', () => ({
+  updateDatasetSetting: (...args: unknown[]) => mockUpdateDatasetSetting(...args),
+}))
+
+vi.mock('@/service/knowledge/use-dataset', () => ({
+  useInvalidDatasetList: () => mockInvalidDatasetList,
+}))
+
+vi.mock('@/service/use-common', () => ({
+  useMembers: () => ({
+    data: {
+      accounts: [
+        { id: 'user-1', name: 'Alice', email: 'alice@example.com', role: 'owner', avatar: '', avatar_url: '', last_login_at: '', created_at: '', status: 'active' },
+        { id: 'user-2', name: 'Bob', email: 'bob@example.com', role: 'admin', avatar: '', avatar_url: '', last_login_at: '', created_at: '', status: 'active' },
+        { id: 'user-3', name: 'Charlie', email: 'charlie@example.com', role: 'normal', avatar: '', avatar_url: '', last_login_at: '', created_at: '', status: 'active' },
+      ],
+    },
+  }),
+}))
+
+vi.mock('@/app/components/header/account-setting/model-provider-page/hooks', () => ({
+  useModelList: () => ({ data: [] }),
+}))
+
+vi.mock('@/app/components/datasets/common/check-rerank-model', () => ({
+  isReRankModelSelected: () => true,
+}))
+
+vi.mock('@/app/components/base/toast', () => ({
+  default: { notify: vi.fn() },
+}))
+
+// --- Dataset factory ---
+
+const createMockDataset = (overrides?: Partial<DataSet>): DataSet => ({
+  id: 'ds-settings-1',
+  name: 'Settings Test Dataset',
+  description: 'Integration test dataset',
+  permission: DatasetPermission.onlyMe,
+  icon_info: {
+    icon_type: 'emoji',
+    icon: '📙',
+    icon_background: '#FFF4ED',
+    icon_url: '',
+  },
+  indexing_technique: 'high_quality',
+  indexing_status: 'completed',
+  data_source_type: DataSourceType.FILE,
+  doc_form: ChunkingMode.text,
+  embedding_model: 'text-embedding-ada-002',
+  embedding_model_provider: 'openai',
+  embedding_available: true,
+  app_count: 2,
+  document_count: 10,
+  total_document_count: 10,
+  word_count: 5000,
+  provider: 'vendor',
+  tags: [],
+  partial_member_list: [],
+  external_knowledge_info: {
+    external_knowledge_id: '',
+    external_knowledge_api_id: '',
+    external_knowledge_api_name: '',
+    external_knowledge_api_endpoint: '',
+  },
+  external_retrieval_model: {
+    top_k: 2,
+    score_threshold: 0.5,
+    score_threshold_enabled: false,
+  },
+  retrieval_model_dict: {
+    search_method: RETRIEVE_METHOD.semantic,
+    reranking_enable: false,
+    reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+    top_k: 3,
+    score_threshold_enabled: false,
+    score_threshold: 0,
+  } as RetrievalConfig,
+  retrieval_model: {
+    search_method: RETRIEVE_METHOD.semantic,
+    reranking_enable: false,
+    reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+    top_k: 3,
+    score_threshold_enabled: false,
+    score_threshold: 0,
+  } as RetrievalConfig,
+  built_in_field_enabled: false,
+  keyword_number: 10,
+  created_by: 'user-1',
+  updated_by: 'user-1',
+  updated_at: Date.now(),
+  runtime_mode: 'general',
+  enable_api: true,
+  is_multimodal: false,
+  ...overrides,
+} as DataSet)
+
+let mockDataset: DataSet = createMockDataset()
+
+vi.mock('@/context/dataset-detail', () => ({
+  useDatasetDetailContextWithSelector: (
+    selector: (state: { dataset: DataSet | null, mutateDatasetRes: () => void }) => unknown,
+  ) => selector({ dataset: mockDataset, mutateDatasetRes: mockMutateDatasets }),
+}))
+
+// Import after mocks are registered
+const { useFormState } = await import(
+  '@/app/components/datasets/settings/form/hooks/use-form-state',
+)
+
+describe('Dataset Settings Flow - Cross-Module Configuration Cascade', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mockUpdateDatasetSetting.mockResolvedValue({})
+    mockDataset = createMockDataset()
+  })
+
+  describe('Form State Initialization from Dataset → Index Method → Retrieval Config Chain', () => {
+    it('should initialise all form dimensions from a QUALIFIED dataset', () => {
+      const { result } = renderHook(() => useFormState())
+
+      expect(result.current.name).toBe('Settings Test Dataset')
+      expect(result.current.description).toBe('Integration test dataset')
+      expect(result.current.indexMethod).toBe('high_quality')
+      expect(result.current.embeddingModel).toEqual({
+        provider: 'openai',
+        model: 'text-embedding-ada-002',
+      })
+      expect(result.current.retrievalConfig.search_method).toBe(RETRIEVE_METHOD.semantic)
+    })
+
+    it('should initialise from an ECONOMICAL dataset with keyword retrieval', () => {
+      mockDataset = createMockDataset({
+        indexing_technique: IndexingType.ECONOMICAL,
+        embedding_model: '',
+        embedding_model_provider: '',
+        retrieval_model_dict: {
+          search_method: RETRIEVE_METHOD.keywordSearch,
+          reranking_enable: false,
+          reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+          top_k: 5,
+          score_threshold_enabled: false,
+          score_threshold: 0,
+        } as RetrievalConfig,
+      })
+
+      const { result } = renderHook(() => useFormState())
+
+      expect(result.current.indexMethod).toBe(IndexingType.ECONOMICAL)
+      expect(result.current.embeddingModel).toEqual({ provider: '', model: '' })
+      expect(result.current.retrievalConfig.search_method).toBe(RETRIEVE_METHOD.keywordSearch)
+    })
+  })
+
+  describe('Index Method Change → Retrieval Config Sync', () => {
+    it('should allow switching index method from QUALIFIED to ECONOMICAL', () => {
+      const { result } = renderHook(() => useFormState())
+
+      expect(result.current.indexMethod).toBe('high_quality')
+
+      act(() => {
+        result.current.setIndexMethod(IndexingType.ECONOMICAL)
+      })
+
+      expect(result.current.indexMethod).toBe(IndexingType.ECONOMICAL)
+    })
+
+    it('should allow updating retrieval config after index method switch', () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setIndexMethod(IndexingType.ECONOMICAL)
+      })
+
+      act(() => {
+        result.current.setRetrievalConfig({
+          ...result.current.retrievalConfig,
+          search_method: RETRIEVE_METHOD.keywordSearch,
+          reranking_enable: false,
+        })
+      })
+
+      expect(result.current.indexMethod).toBe(IndexingType.ECONOMICAL)
+      expect(result.current.retrievalConfig.search_method).toBe(RETRIEVE_METHOD.keywordSearch)
+      expect(result.current.retrievalConfig.reranking_enable).toBe(false)
+    })
+
+    it('should preserve retrieval config when switching back to QUALIFIED', () => {
+      const { result } = renderHook(() => useFormState())
+
+      const originalConfig = { ...result.current.retrievalConfig }
+
+      act(() => {
+        result.current.setIndexMethod(IndexingType.ECONOMICAL)
+      })
+      act(() => {
+        result.current.setIndexMethod(IndexingType.QUALIFIED)
+      })
+
+      expect(result.current.indexMethod).toBe('high_quality')
+      expect(result.current.retrievalConfig.search_method).toBe(originalConfig.search_method)
+    })
+  })
+
+  describe('Permission Change → Member List Visibility Logic', () => {
+    it('should start with onlyMe permission and empty member selection', () => {
+      const { result } = renderHook(() => useFormState())
+
+      expect(result.current.permission).toBe(DatasetPermission.onlyMe)
+      expect(result.current.selectedMemberIDs).toEqual([])
+    })
+
+    it('should enable member selection when switching to partialMembers', () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setPermission(DatasetPermission.partialMembers)
+      })
+
+      expect(result.current.permission).toBe(DatasetPermission.partialMembers)
+      expect(result.current.memberList).toHaveLength(3)
+      expect(result.current.memberList.map(m => m.id)).toEqual(['user-1', 'user-2', 'user-3'])
+    })
+
+    it('should persist member selection through permission toggle', () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setPermission(DatasetPermission.partialMembers)
+        result.current.setSelectedMemberIDs(['user-1', 'user-3'])
+      })
+
+      act(() => {
+        result.current.setPermission(DatasetPermission.allTeamMembers)
+      })
+
+      act(() => {
+        result.current.setPermission(DatasetPermission.partialMembers)
+      })
+
+      expect(result.current.selectedMemberIDs).toEqual(['user-1', 'user-3'])
+    })
+
+    it('should include partial_member_list in save payload only for partialMembers', async () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setPermission(DatasetPermission.partialMembers)
+        result.current.setSelectedMemberIDs(['user-2'])
+      })
+
+      await act(async () => {
+        await result.current.handleSave()
+      })
+
+      expect(mockUpdateDatasetSetting).toHaveBeenCalledWith({
+        datasetId: 'ds-settings-1',
+        body: expect.objectContaining({
+          permission: DatasetPermission.partialMembers,
+          partial_member_list: [
+            expect.objectContaining({ user_id: 'user-2', role: 'admin' }),
+          ],
+        }),
+      })
+    })
+
+    it('should not include partial_member_list for allTeamMembers permission', async () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setPermission(DatasetPermission.allTeamMembers)
+      })
+
+      await act(async () => {
+        await result.current.handleSave()
+      })
+
+      const savedBody = mockUpdateDatasetSetting.mock.calls[0][0].body as Record<string, unknown>
+      expect(savedBody).not.toHaveProperty('partial_member_list')
+    })
+  })
+
+  describe('Form Submission Validation → All Fields Together', () => {
+    it('should reject empty name on save', async () => {
+      const Toast = await import('@/app/components/base/toast')
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setName('')
+      })
+
+      await act(async () => {
+        await result.current.handleSave()
+      })
+
+      expect(Toast.default.notify).toHaveBeenCalledWith({
+        type: 'error',
+        message: expect.any(String),
+      })
+      expect(mockUpdateDatasetSetting).not.toHaveBeenCalled()
+    })
+
+    it('should include all configuration dimensions in a successful save', async () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setName('Updated Name')
+        result.current.setDescription('Updated Description')
+        result.current.setIndexMethod(IndexingType.ECONOMICAL)
+        result.current.setKeywordNumber(15)
+      })
+
+      await act(async () => {
+        await result.current.handleSave()
+      })
+
+      expect(mockUpdateDatasetSetting).toHaveBeenCalledWith({
+        datasetId: 'ds-settings-1',
+        body: expect.objectContaining({
+          name: 'Updated Name',
+          description: 'Updated Description',
+          indexing_technique: 'economy',
+          keyword_number: 15,
+          embedding_model: 'text-embedding-ada-002',
+          embedding_model_provider: 'openai',
+        }),
+      })
+    })
+
+    it('should call mutateDatasets and invalidDatasetList after successful save', async () => {
+      const { result } = renderHook(() => useFormState())
+
+      await act(async () => {
+        await result.current.handleSave()
+      })
+
+      await waitFor(() => {
+        expect(mockMutateDatasets).toHaveBeenCalled()
+        expect(mockInvalidDatasetList).toHaveBeenCalled()
+      })
+    })
+  })
+
+  describe('Embedding Model Change → Retrieval Config Cascade', () => {
+    it('should update embedding model independently of retrieval config', () => {
+      const { result } = renderHook(() => useFormState())
+
+      const originalRetrievalConfig = { ...result.current.retrievalConfig }
+
+      act(() => {
+        result.current.setEmbeddingModel({ provider: 'cohere', model: 'embed-english-v3.0' })
+      })
+
+      expect(result.current.embeddingModel).toEqual({
+        provider: 'cohere',
+        model: 'embed-english-v3.0',
+      })
+      expect(result.current.retrievalConfig.search_method).toBe(originalRetrievalConfig.search_method)
+    })
+
+    it('should propagate embedding model into weighted retrieval config on save', async () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setEmbeddingModel({ provider: 'cohere', model: 'embed-v3' })
+        result.current.setRetrievalConfig({
+          ...result.current.retrievalConfig,
+          search_method: RETRIEVE_METHOD.hybrid,
+          weights: {
+            weight_type: WeightedScoreEnum.Customized,
+            vector_setting: {
+              vector_weight: 0.6,
+              embedding_provider_name: '',
+              embedding_model_name: '',
+            },
+            keyword_setting: { keyword_weight: 0.4 },
+          },
+        })
+      })
+
+      await act(async () => {
+        await result.current.handleSave()
+      })
+
+      expect(mockUpdateDatasetSetting).toHaveBeenCalledWith({
+        datasetId: 'ds-settings-1',
+        body: expect.objectContaining({
+          embedding_model: 'embed-v3',
+          embedding_model_provider: 'cohere',
+          retrieval_model: expect.objectContaining({
+            weights: expect.objectContaining({
+              vector_setting: expect.objectContaining({
+                embedding_provider_name: 'cohere',
+                embedding_model_name: 'embed-v3',
+              }),
+            }),
+          }),
+        }),
+      })
+    })
+
+    it('should handle switching from semantic to hybrid search with embedding model', () => {
+      const { result } = renderHook(() => useFormState())
+
+      act(() => {
+        result.current.setRetrievalConfig({
+          ...result.current.retrievalConfig,
+          search_method: RETRIEVE_METHOD.hybrid,
+          reranking_enable: true,
+          reranking_model: {
+            reranking_provider_name: 'cohere',
+            reranking_model_name: 'rerank-english-v3.0',
+          },
+        })
+      })
+
+      expect(result.current.retrievalConfig.search_method).toBe(RETRIEVE_METHOD.hybrid)
+      expect(result.current.retrievalConfig.reranking_enable).toBe(true)
+      expect(result.current.embeddingModel.model).toBe('text-embedding-ada-002')
+    })
+  })
+})

web/__tests__/datasets/document-management.test.tsx

@@ -0,0 +1,334 @@
+/**
+ * Integration Test: Document Management Flow
+ *
+ * Tests cross-module interactions: query state (URL-based) → document list sorting →
+ * document selection → status filter utilities.
+ * Validates the data contract between documents page hooks and list component hooks.
+ */
+
+import type { SimpleDocumentDetail } from '@/models/datasets'
+import { act, renderHook } from '@testing-library/react'
+import { DataSourceType } from '@/models/datasets'
+
+const mockPush = vi.fn()
+vi.mock('next/navigation', () => ({
+  useSearchParams: () => new URLSearchParams(''),
+  useRouter: () => ({ push: mockPush }),
+  usePathname: () => '/datasets/ds-1/documents',
+}))
+
+const { sanitizeStatusValue, normalizeStatusForQuery } = await import(
+  '@/app/components/datasets/documents/status-filter',
+)
+
+const { useDocumentSort } = await import(
+  '@/app/components/datasets/documents/components/document-list/hooks/use-document-sort',
+)
+const { useDocumentSelection } = await import(
+  '@/app/components/datasets/documents/components/document-list/hooks/use-document-selection',
+)
+const { default: useDocumentListQueryState } = await import(
+  '@/app/components/datasets/documents/hooks/use-document-list-query-state',
+)
+
+type LocalDoc = SimpleDocumentDetail & { percent?: number }
+
+const createDoc = (overrides?: Partial<LocalDoc>): LocalDoc => ({
+  id: `doc-${Math.random().toString(36).slice(2, 8)}`,
+  name: 'test-doc.txt',
+  word_count: 500,
+  hit_count: 10,
+  created_at: Date.now() / 1000,
+  data_source_type: DataSourceType.FILE,
+  display_status: 'available',
+  indexing_status: 'completed',
+  enabled: true,
+  archived: false,
+  doc_type: null,
+  doc_metadata: null,
+  position: 1,
+  dataset_process_rule_id: 'rule-1',
+  ...overrides,
+} as LocalDoc)
+
+describe('Document Management Flow', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Status Filter Utilities', () => {
+    it('should sanitize valid status values', () => {
+      expect(sanitizeStatusValue('all')).toBe('all')
+      expect(sanitizeStatusValue('available')).toBe('available')
+      expect(sanitizeStatusValue('error')).toBe('error')
+    })
+
+    it('should fallback to "all" for invalid values', () => {
+      expect(sanitizeStatusValue(null)).toBe('all')
+      expect(sanitizeStatusValue(undefined)).toBe('all')
+      expect(sanitizeStatusValue('')).toBe('all')
+      expect(sanitizeStatusValue('nonexistent')).toBe('all')
+    })
+
+    it('should handle URL aliases', () => {
+      // 'active' is aliased to 'available'
+      expect(sanitizeStatusValue('active')).toBe('available')
+    })
+
+    it('should normalize status for API query', () => {
+      expect(normalizeStatusForQuery('all')).toBe('all')
+      // 'enabled' normalized to 'available' for query
+      expect(normalizeStatusForQuery('enabled')).toBe('available')
+    })
+  })
+
+  describe('URL-based Query State', () => {
+    it('should parse default query from empty URL params', () => {
+      const { result } = renderHook(() => useDocumentListQueryState())
+
+      expect(result.current.query).toEqual({
+        page: 1,
+        limit: 10,
+        keyword: '',
+        status: 'all',
+        sort: '-created_at',
+      })
+    })
+
+    it('should update query and push to router', () => {
+      const { result } = renderHook(() => useDocumentListQueryState())
+
+      act(() => {
+        result.current.updateQuery({ keyword: 'test', page: 2 })
+      })
+
+      expect(mockPush).toHaveBeenCalled()
+      // The push call should contain the updated query params
+      const pushUrl = mockPush.mock.calls[0][0] as string
+      expect(pushUrl).toContain('keyword=test')
+      expect(pushUrl).toContain('page=2')
+    })
+
+    it('should reset query to defaults', () => {
+      const { result } = renderHook(() => useDocumentListQueryState())
+
+      act(() => {
+        result.current.resetQuery()
+      })
+
+      expect(mockPush).toHaveBeenCalled()
+      // Default query omits default values from URL
+      const pushUrl = mockPush.mock.calls[0][0] as string
+      expect(pushUrl).toBe('/datasets/ds-1/documents')
+    })
+  })
+
+  describe('Document Sort Integration', () => {
+    it('should return documents unsorted when no sort field set', () => {
+      const docs = [
+        createDoc({ id: 'doc-1', name: 'Banana.txt', word_count: 300 }),
+        createDoc({ id: 'doc-2', name: 'Apple.txt', word_count: 100 }),
+        createDoc({ id: 'doc-3', name: 'Cherry.txt', word_count: 200 }),
+      ]
+
+      const { result } = renderHook(() => useDocumentSort({
+        documents: docs,
+        statusFilterValue: '',
+        remoteSortValue: '-created_at',
+      }))
+
+      expect(result.current.sortField).toBeNull()
+      expect(result.current.sortedDocuments).toHaveLength(3)
+    })
+
+    it('should sort by name descending', () => {
+      const docs = [
+        createDoc({ id: 'doc-1', name: 'Banana.txt' }),
+        createDoc({ id: 'doc-2', name: 'Apple.txt' }),
+        createDoc({ id: 'doc-3', name: 'Cherry.txt' }),
+      ]
+
+      const { result } = renderHook(() => useDocumentSort({
+        documents: docs,
+        statusFilterValue: '',
+        remoteSortValue: '-created_at',
+      }))
+
+      act(() => {
+        result.current.handleSort('name')
+      })
+
+      expect(result.current.sortField).toBe('name')
+      expect(result.current.sortOrder).toBe('desc')
+      const names = result.current.sortedDocuments.map(d => d.name)
+      expect(names).toEqual(['Cherry.txt', 'Banana.txt', 'Apple.txt'])
+    })
+
+    it('should toggle sort order on same field click', () => {
+      const docs = [createDoc({ id: 'doc-1', name: 'A.txt' }), createDoc({ id: 'doc-2', name: 'B.txt' })]
+
+      const { result } = renderHook(() => useDocumentSort({
+        documents: docs,
+        statusFilterValue: '',
+        remoteSortValue: '-created_at',
+      }))
+
+      act(() => result.current.handleSort('name'))
+      expect(result.current.sortOrder).toBe('desc')
+
+      act(() => result.current.handleSort('name'))
+      expect(result.current.sortOrder).toBe('asc')
+    })
+
+    it('should filter by status before sorting', () => {
+      const docs = [
+        createDoc({ id: 'doc-1', name: 'A.txt', display_status: 'available' }),
+        createDoc({ id: 'doc-2', name: 'B.txt', display_status: 'error' }),
+        createDoc({ id: 'doc-3', name: 'C.txt', display_status: 'available' }),
+      ]
+
+      const { result } = renderHook(() => useDocumentSort({
+        documents: docs,
+        statusFilterValue: 'available',
+        remoteSortValue: '-created_at',
+      }))
+
+      // Only 'available' documents should remain
+      expect(result.current.sortedDocuments).toHaveLength(2)
+      expect(result.current.sortedDocuments.every(d => d.display_status === 'available')).toBe(true)
+    })
+  })
+
+  describe('Document Selection Integration', () => {
+    it('should manage selection state externally', () => {
+      const docs = [
+        createDoc({ id: 'doc-1' }),
+        createDoc({ id: 'doc-2' }),
+        createDoc({ id: 'doc-3' }),
+      ]
+      const onSelectedIdChange = vi.fn()
+
+      const { result } = renderHook(() => useDocumentSelection({
+        documents: docs,
+        selectedIds: [],
+        onSelectedIdChange,
+      }))
+
+      expect(result.current.isAllSelected).toBe(false)
+      expect(result.current.isSomeSelected).toBe(false)
+    })
+
+    it('should select all documents', () => {
+      const docs = [
+        createDoc({ id: 'doc-1' }),
+        createDoc({ id: 'doc-2' }),
+      ]
+      const onSelectedIdChange = vi.fn()
+
+      const { result } = renderHook(() => useDocumentSelection({
+        documents: docs,
+        selectedIds: [],
+        onSelectedIdChange,
+      }))
+
+      act(() => {
+        result.current.onSelectAll()
+      })
+
+      expect(onSelectedIdChange).toHaveBeenCalledWith(
+        expect.arrayContaining(['doc-1', 'doc-2']),
+      )
+    })
+
+    it('should detect all-selected state', () => {
+      const docs = [
+        createDoc({ id: 'doc-1' }),
+        createDoc({ id: 'doc-2' }),
+      ]
+
+      const { result } = renderHook(() => useDocumentSelection({
+        documents: docs,
+        selectedIds: ['doc-1', 'doc-2'],
+        onSelectedIdChange: vi.fn(),
+      }))
+
+      expect(result.current.isAllSelected).toBe(true)
+    })
+
+    it('should detect partial selection', () => {
+      const docs = [
+        createDoc({ id: 'doc-1' }),
+        createDoc({ id: 'doc-2' }),
+        createDoc({ id: 'doc-3' }),
+      ]
+
+      const { result } = renderHook(() => useDocumentSelection({
+        documents: docs,
+        selectedIds: ['doc-1'],
+        onSelectedIdChange: vi.fn(),
+      }))
+
+      expect(result.current.isSomeSelected).toBe(true)
+      expect(result.current.isAllSelected).toBe(false)
+    })
+
+    it('should identify downloadable selected documents (FILE type only)', () => {
+      const docs = [
+        createDoc({ id: 'doc-1', data_source_type: DataSourceType.FILE }),
+        createDoc({ id: 'doc-2', data_source_type: DataSourceType.NOTION }),
+      ]
+
+      const { result } = renderHook(() => useDocumentSelection({
+        documents: docs,
+        selectedIds: ['doc-1', 'doc-2'],
+        onSelectedIdChange: vi.fn(),
+      }))
+
+      expect(result.current.downloadableSelectedIds).toEqual(['doc-1'])
+    })
+
+    it('should clear selection', () => {
+      const onSelectedIdChange = vi.fn()
+      const docs = [createDoc({ id: 'doc-1' })]
+
+      const { result } = renderHook(() => useDocumentSelection({
+        documents: docs,
+        selectedIds: ['doc-1'],
+        onSelectedIdChange,
+      }))
+
+      act(() => {
+        result.current.clearSelection()
+      })
+
+      expect(onSelectedIdChange).toHaveBeenCalledWith([])
+    })
+  })
+
+  describe('Cross-Module: Query State → Sort → Selection Pipeline', () => {
+    it('should maintain consistent default state across all hooks', () => {
+      const docs = [createDoc({ id: 'doc-1' })]
+      const { result: queryResult } = renderHook(() => useDocumentListQueryState())
+      const { result: sortResult } = renderHook(() => useDocumentSort({
+        documents: docs,
+        statusFilterValue: queryResult.current.query.status,
+        remoteSortValue: queryResult.current.query.sort,
+      }))
+      const { result: selResult } = renderHook(() => useDocumentSelection({
+        documents: sortResult.current.sortedDocuments,
+        selectedIds: [],
+        onSelectedIdChange: vi.fn(),
+      }))
+
+      // Query defaults
+      expect(queryResult.current.query.sort).toBe('-created_at')
+      expect(queryResult.current.query.status).toBe('all')
+
+      // Sort inherits 'all' status → no filtering applied
+      expect(sortResult.current.sortedDocuments).toHaveLength(1)
+
+      // Selection starts empty
+      expect(selResult.current.isAllSelected).toBe(false)
+    })
+  })
+})

web/__tests__/datasets/external-knowledge-base.test.tsx

@@ -0,0 +1,214 @@
+/**
+ * Integration Test: External Knowledge Base Creation Flow
+ *
+ * Tests the data contract, validation logic, and API interaction
+ * for external knowledge base creation.
+ */
+
+import type { CreateKnowledgeBaseReq } from '@/app/components/datasets/external-knowledge-base/create/declarations'
+
+// --- Factory ---
+const createFormData = (overrides?: Partial<CreateKnowledgeBaseReq>): CreateKnowledgeBaseReq => ({
+  name: 'My External KB',
+  description: 'A test external knowledge base',
+  external_knowledge_api_id: 'api-1',
+  external_knowledge_id: 'ext-kb-123',
+  external_retrieval_model: {
+    top_k: 4,
+    score_threshold: 0.5,
+    score_threshold_enabled: false,
+  },
+  provider: 'external',
+  ...overrides,
+})
+
+describe('External Knowledge Base Creation Flow', () => {
+  describe('Data Contract: CreateKnowledgeBaseReq', () => {
+    it('should define a complete form structure', () => {
+      const form = createFormData()
+
+      expect(form).toHaveProperty('name')
+      expect(form).toHaveProperty('external_knowledge_api_id')
+      expect(form).toHaveProperty('external_knowledge_id')
+      expect(form).toHaveProperty('external_retrieval_model')
+      expect(form).toHaveProperty('provider')
+      expect(form.provider).toBe('external')
+    })
+
+    it('should include retrieval model settings', () => {
+      const form = createFormData()
+
+      expect(form.external_retrieval_model).toEqual({
+        top_k: 4,
+        score_threshold: 0.5,
+        score_threshold_enabled: false,
+      })
+    })
+
+    it('should allow partial overrides', () => {
+      const form = createFormData({
+        name: 'Custom Name',
+        external_retrieval_model: {
+          top_k: 10,
+          score_threshold: 0.8,
+          score_threshold_enabled: true,
+        },
+      })
+
+      expect(form.name).toBe('Custom Name')
+      expect(form.external_retrieval_model.top_k).toBe(10)
+      expect(form.external_retrieval_model.score_threshold_enabled).toBe(true)
+    })
+  })
+
+  describe('Form Validation Logic', () => {
+    const isFormValid = (form: CreateKnowledgeBaseReq): boolean => {
+      return (
+        form.name.trim() !== ''
+        && form.external_knowledge_api_id !== ''
+        && form.external_knowledge_id !== ''
+        && form.external_retrieval_model.top_k !== undefined
+        && form.external_retrieval_model.score_threshold !== undefined
+      )
+    }
+
+    it('should validate a complete form', () => {
+      const form = createFormData()
+      expect(isFormValid(form)).toBe(true)
+    })
+
+    it('should reject empty name', () => {
+      const form = createFormData({ name: '' })
+      expect(isFormValid(form)).toBe(false)
+    })
+
+    it('should reject whitespace-only name', () => {
+      const form = createFormData({ name: '   ' })
+      expect(isFormValid(form)).toBe(false)
+    })
+
+    it('should reject empty external_knowledge_api_id', () => {
+      const form = createFormData({ external_knowledge_api_id: '' })
+      expect(isFormValid(form)).toBe(false)
+    })
+
+    it('should reject empty external_knowledge_id', () => {
+      const form = createFormData({ external_knowledge_id: '' })
+      expect(isFormValid(form)).toBe(false)
+    })
+  })
+
+  describe('Form State Transitions', () => {
+    it('should start with empty default state', () => {
+      const defaultForm: CreateKnowledgeBaseReq = {
+        name: '',
+        description: '',
+        external_knowledge_api_id: '',
+        external_knowledge_id: '',
+        external_retrieval_model: {
+          top_k: 4,
+          score_threshold: 0.5,
+          score_threshold_enabled: false,
+        },
+        provider: 'external',
+      }
+
+      // Verify default state matches component's initial useState
+      expect(defaultForm.name).toBe('')
+      expect(defaultForm.external_knowledge_api_id).toBe('')
+      expect(defaultForm.external_knowledge_id).toBe('')
+      expect(defaultForm.provider).toBe('external')
+    })
+
+    it('should support immutable form updates', () => {
+      const form = createFormData({ name: '' })
+      const updated = { ...form, name: 'Updated Name' }
+
+      expect(form.name).toBe('')
+      expect(updated.name).toBe('Updated Name')
+      // Other fields should remain unchanged
+      expect(updated.external_knowledge_api_id).toBe(form.external_knowledge_api_id)
+    })
+
+    it('should support retrieval model updates', () => {
+      const form = createFormData()
+      const updated = {
+        ...form,
+        external_retrieval_model: {
+          ...form.external_retrieval_model,
+          top_k: 10,
+          score_threshold_enabled: true,
+        },
+      }
+
+      expect(updated.external_retrieval_model.top_k).toBe(10)
+      expect(updated.external_retrieval_model.score_threshold_enabled).toBe(true)
+      // Unchanged field
+      expect(updated.external_retrieval_model.score_threshold).toBe(0.5)
+    })
+  })
+
+  describe('API Call Data Contract', () => {
+    it('should produce a valid API payload from form data', () => {
+      const form = createFormData()
+
+      // The API expects the full CreateKnowledgeBaseReq
+      expect(form.name).toBeTruthy()
+      expect(form.external_knowledge_api_id).toBeTruthy()
+      expect(form.external_knowledge_id).toBeTruthy()
+      expect(form.provider).toBe('external')
+      expect(typeof form.external_retrieval_model.top_k).toBe('number')
+      expect(typeof form.external_retrieval_model.score_threshold).toBe('number')
+      expect(typeof form.external_retrieval_model.score_threshold_enabled).toBe('boolean')
+    })
+
+    it('should support optional description', () => {
+      const formWithDesc = createFormData({ description: 'Some description' })
+      const formWithoutDesc = createFormData({ description: '' })
+
+      expect(formWithDesc.description).toBe('Some description')
+      expect(formWithoutDesc.description).toBe('')
+    })
+
+    it('should validate retrieval model bounds', () => {
+      const form = createFormData({
+        external_retrieval_model: {
+          top_k: 0,
+          score_threshold: 0,
+          score_threshold_enabled: false,
+        },
+      })
+
+      expect(form.external_retrieval_model.top_k).toBe(0)
+      expect(form.external_retrieval_model.score_threshold).toBe(0)
+    })
+  })
+
+  describe('External API List Integration', () => {
+    it('should validate API item structure', () => {
+      const apiItem = {
+        id: 'api-1',
+        name: 'Production API',
+        settings: {
+          endpoint: 'https://api.example.com',
+          api_key: 'key-123',
+        },
+      }
+
+      expect(apiItem).toHaveProperty('id')
+      expect(apiItem).toHaveProperty('name')
+      expect(apiItem).toHaveProperty('settings')
+      expect(apiItem.settings).toHaveProperty('endpoint')
+      expect(apiItem.settings).toHaveProperty('api_key')
+    })
+
+    it('should link API selection to form data', () => {
+      const selectedApi = { id: 'api-2', name: 'Staging API' }
+      const form = createFormData({
+        external_knowledge_api_id: selectedApi.id,
+      })
+
+      expect(form.external_knowledge_api_id).toBe('api-2')
+    })
+  })
+})

web/__tests__/datasets/hit-testing-flow.test.tsx

@@ -0,0 +1,404 @@
+/**
+ * Integration Test: Hit Testing Flow
+ *
+ * Tests the query submission → API response → callback chain flow
+ * by rendering the actual QueryInput component and triggering user interactions.
+ * Validates that the production onSubmit logic correctly constructs payloads
+ * and invokes callbacks on success/failure.
+ */
+
+import type {
+  HitTestingResponse,
+  Query,
+} from '@/models/datasets'
+import type { RetrievalConfig } from '@/types/app'
+import { fireEvent, render, screen, waitFor } from '@testing-library/react'
+import QueryInput from '@/app/components/datasets/hit-testing/components/query-input'
+import { RETRIEVE_METHOD } from '@/types/app'
+
+// --- Mocks ---
+
+vi.mock('@/context/dataset-detail', () => ({
+  default: {},
+  useDatasetDetailContext: vi.fn(() => ({ dataset: undefined })),
+  useDatasetDetailContextWithSelector: vi.fn(() => false),
+}))
+
+vi.mock('use-context-selector', () => ({
+  useContext: vi.fn(() => ({})),
+  useContextSelector: vi.fn(() => false),
+  createContext: vi.fn(() => ({})),
+}))
+
+vi.mock('@/app/components/datasets/common/image-uploader/image-uploader-in-retrieval-testing', () => ({
+  default: ({ textArea, actionButton }: { textArea: React.ReactNode, actionButton: React.ReactNode }) => (
+    <div data-testid="image-uploader-mock">
+      {textArea}
+      {actionButton}
+    </div>
+  ),
+}))
+
+// --- Factories ---
+
+const createRetrievalConfig = (overrides = {}): RetrievalConfig => ({
+  search_method: RETRIEVE_METHOD.semantic,
+  reranking_enable: false,
+  reranking_mode: undefined,
+  reranking_model: {
+    reranking_provider_name: '',
+    reranking_model_name: '',
+  },
+  weights: undefined,
+  top_k: 3,
+  score_threshold_enabled: false,
+  score_threshold: 0.5,
+  ...overrides,
+} as RetrievalConfig)
+
+const createHitTestingResponse = (numResults: number): HitTestingResponse => ({
+  query: {
+    content: 'What is Dify?',
+    tsne_position: { x: 0, y: 0 },
+  },
+  records: Array.from({ length: numResults }, (_, i) => ({
+    segment: {
+      id: `seg-${i}`,
+      document: {
+        id: `doc-${i}`,
+        data_source_type: 'upload_file',
+        name: `document-${i}.txt`,
+        doc_type: null as unknown as import('@/models/datasets').DocType,
+      },
+      content: `Result content ${i}`,
+      sign_content: `Result content ${i}`,
+      position: i + 1,
+      word_count: 100 + i * 50,
+      tokens: 50 + i * 25,
+      keywords: ['test', 'dify'],
+      hit_count: i * 5,
+      index_node_hash: `hash-${i}`,
+      answer: '',
+    },
+    content: {
+      id: `seg-${i}`,
+      document: {
+        id: `doc-${i}`,
+        data_source_type: 'upload_file',
+        name: `document-${i}.txt`,
+        doc_type: null as unknown as import('@/models/datasets').DocType,
+      },
+      content: `Result content ${i}`,
+      sign_content: `Result content ${i}`,
+      position: i + 1,
+      word_count: 100 + i * 50,
+      tokens: 50 + i * 25,
+      keywords: ['test', 'dify'],
+      hit_count: i * 5,
+      index_node_hash: `hash-${i}`,
+      answer: '',
+    },
+    score: 0.95 - i * 0.1,
+    tsne_position: { x: 0, y: 0 },
+    child_chunks: null,
+    files: [],
+  })),
+})
+
+const createTextQuery = (content: string): Query[] => [
+  { content, content_type: 'text_query', file_info: null },
+]
+
+// --- Helpers ---
+
+const findSubmitButton = () => {
+  const buttons = screen.getAllByRole('button')
+  const submitButton = buttons.find(btn => btn.classList.contains('w-[88px]'))
+  expect(submitButton).toBeTruthy()
+  return submitButton!
+}
+
+// --- Tests ---
+
+describe('Hit Testing Flow', () => {
+  const mockHitTestingMutation = vi.fn()
+  const mockExternalMutation = vi.fn()
+  const mockSetHitResult = vi.fn()
+  const mockSetExternalHitResult = vi.fn()
+  const mockOnUpdateList = vi.fn()
+  const mockSetQueries = vi.fn()
+  const mockOnClickRetrievalMethod = vi.fn()
+  const mockOnSubmit = vi.fn()
+
+  const createDefaultProps = (overrides: Record<string, unknown> = {}) => ({
+    onUpdateList: mockOnUpdateList,
+    setHitResult: mockSetHitResult,
+    setExternalHitResult: mockSetExternalHitResult,
+    loading: false,
+    queries: [] as Query[],
+    setQueries: mockSetQueries,
+    isExternal: false,
+    onClickRetrievalMethod: mockOnClickRetrievalMethod,
+    retrievalConfig: createRetrievalConfig(),
+    isEconomy: false,
+    onSubmit: mockOnSubmit,
+    hitTestingMutation: mockHitTestingMutation,
+    externalKnowledgeBaseHitTestingMutation: mockExternalMutation,
+    ...overrides,
+  })
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Query Submission → API Call', () => {
+    it('should call hitTestingMutation with correct payload including retrieval model', async () => {
+      const retrievalConfig = createRetrievalConfig({
+        search_method: RETRIEVE_METHOD.semantic,
+        top_k: 3,
+        score_threshold_enabled: false,
+      })
+      mockHitTestingMutation.mockResolvedValue(createHitTestingResponse(3))
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('How does RAG work?'),
+          retrievalConfig,
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockHitTestingMutation).toHaveBeenCalledWith(
+          expect.objectContaining({
+            query: 'How does RAG work?',
+            attachment_ids: [],
+            retrieval_model: expect.objectContaining({
+              search_method: RETRIEVE_METHOD.semantic,
+              top_k: 3,
+              score_threshold_enabled: false,
+            }),
+          }),
+          expect.objectContaining({
+            onSuccess: expect.any(Function),
+          }),
+        )
+      })
+    })
+
+    it('should override search_method to keywordSearch when isEconomy is true', async () => {
+      const retrievalConfig = createRetrievalConfig({ search_method: RETRIEVE_METHOD.semantic })
+      mockHitTestingMutation.mockResolvedValue(createHitTestingResponse(1))
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('test query'),
+          retrievalConfig,
+          isEconomy: true,
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockHitTestingMutation).toHaveBeenCalledWith(
+          expect.objectContaining({
+            retrieval_model: expect.objectContaining({
+              search_method: RETRIEVE_METHOD.keywordSearch,
+            }),
+          }),
+          expect.anything(),
+        )
+      })
+    })
+
+    it('should handle empty results by calling setHitResult with empty records', async () => {
+      const emptyResponse = createHitTestingResponse(0)
+      mockHitTestingMutation.mockImplementation(async (_params: unknown, options?: { onSuccess?: (data: HitTestingResponse) => void }) => {
+        options?.onSuccess?.(emptyResponse)
+        return emptyResponse
+      })
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('nonexistent topic'),
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockSetHitResult).toHaveBeenCalledWith(
+          expect.objectContaining({ records: [] }),
+        )
+      })
+    })
+
+    it('should not call success callbacks when mutation resolves without onSuccess', async () => {
+      // Simulate a mutation that resolves but does not invoke the onSuccess callback
+      mockHitTestingMutation.mockResolvedValue(undefined)
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('test'),
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockHitTestingMutation).toHaveBeenCalled()
+      })
+      // Success callbacks should not fire when onSuccess is not invoked
+      expect(mockSetHitResult).not.toHaveBeenCalled()
+      expect(mockOnUpdateList).not.toHaveBeenCalled()
+      expect(mockOnSubmit).not.toHaveBeenCalled()
+    })
+  })
+
+  describe('API Response → Results Data Contract', () => {
+    it('should produce results with required segment fields for rendering', () => {
+      const response = createHitTestingResponse(3)
+
+      // Validate each result has the fields needed by ResultItem component
+      response.records.forEach((record) => {
+        expect(record.segment).toHaveProperty('id')
+        expect(record.segment).toHaveProperty('content')
+        expect(record.segment).toHaveProperty('position')
+        expect(record.segment).toHaveProperty('word_count')
+        expect(record.segment).toHaveProperty('document')
+        expect(record.segment.document).toHaveProperty('name')
+        expect(record.score).toBeGreaterThanOrEqual(0)
+        expect(record.score).toBeLessThanOrEqual(1)
+      })
+    })
+
+    it('should maintain correct score ordering', () => {
+      const response = createHitTestingResponse(5)
+
+      for (let i = 1; i < response.records.length; i++) {
+        expect(response.records[i - 1].score).toBeGreaterThanOrEqual(response.records[i].score)
+      }
+    })
+
+    it('should include document metadata for result item display', () => {
+      const response = createHitTestingResponse(1)
+      const record = response.records[0]
+
+      expect(record.segment.document.name).toBeTruthy()
+      expect(record.segment.document.data_source_type).toBeTruthy()
+    })
+  })
+
+  describe('Successful Submission → Callback Chain', () => {
+    it('should call setHitResult, onUpdateList, and onSubmit after successful submission', async () => {
+      const response = createHitTestingResponse(3)
+      mockHitTestingMutation.mockImplementation(async (_params: unknown, options?: { onSuccess?: (data: HitTestingResponse) => void }) => {
+        options?.onSuccess?.(response)
+        return response
+      })
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('Test query'),
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockSetHitResult).toHaveBeenCalledWith(response)
+        expect(mockOnUpdateList).toHaveBeenCalledTimes(1)
+        expect(mockOnSubmit).toHaveBeenCalledTimes(1)
+      })
+    })
+
+    it('should trigger records list refresh via onUpdateList after query', async () => {
+      const response = createHitTestingResponse(1)
+      mockHitTestingMutation.mockImplementation(async (_params: unknown, options?: { onSuccess?: (data: HitTestingResponse) => void }) => {
+        options?.onSuccess?.(response)
+        return response
+      })
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('new query'),
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockOnUpdateList).toHaveBeenCalledTimes(1)
+      })
+    })
+  })
+
+  describe('External KB Hit Testing', () => {
+    it('should use external mutation with correct payload for external datasets', async () => {
+      mockExternalMutation.mockImplementation(async (_params: unknown, options?: { onSuccess?: (data: { records: never[] }) => void }) => {
+        const response = { records: [] }
+        options?.onSuccess?.(response)
+        return response
+      })
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('test'),
+          isExternal: true,
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockExternalMutation).toHaveBeenCalledWith(
+          expect.objectContaining({
+            query: 'test',
+            external_retrieval_model: expect.objectContaining({
+              top_k: 4,
+              score_threshold: 0.5,
+              score_threshold_enabled: false,
+            }),
+          }),
+          expect.objectContaining({
+            onSuccess: expect.any(Function),
+          }),
+        )
+        // Internal mutation should NOT be called
+        expect(mockHitTestingMutation).not.toHaveBeenCalled()
+      })
+    })
+
+    it('should call setExternalHitResult and onUpdateList on successful external submission', async () => {
+      const externalResponse = { records: [] }
+      mockExternalMutation.mockImplementation(async (_params: unknown, options?: { onSuccess?: (data: { records: never[] }) => void }) => {
+        options?.onSuccess?.(externalResponse)
+        return externalResponse
+      })
+
+      render(
+        <QueryInput {...createDefaultProps({
+          queries: createTextQuery('external query'),
+          isExternal: true,
+        })}
+        />,
+      )
+
+      fireEvent.click(findSubmitButton())
+
+      await waitFor(() => {
+        expect(mockSetExternalHitResult).toHaveBeenCalledWith(externalResponse)
+        expect(mockOnUpdateList).toHaveBeenCalledTimes(1)
+      })
+    })
+  })
+})

web/__tests__/datasets/metadata-management-flow.test.tsx

@@ -0,0 +1,337 @@
+/**
+ * Integration Test: Metadata Management Flow
+ *
+ * Tests the cross-module composition of metadata name validation, type constraints,
+ * and duplicate detection across the metadata management hooks.
+ *
+ * The unit-level use-check-metadata-name.spec.ts tests the validation hook alone.
+ * This integration test verifies:
+ *   - Name validation combined with existing metadata list (duplicate detection)
+ *   - Metadata type enum constraints matching expected data model
+ *   - Full add/rename workflow: validate name → check duplicates → allow or reject
+ *   - Name uniqueness logic: existing metadata keeps its own name, cannot take another's
+ */
+
+import type { MetadataItemWithValueLength } from '@/app/components/datasets/metadata/types'
+import { renderHook } from '@testing-library/react'
+import { DataType } from '@/app/components/datasets/metadata/types'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
+}))
+
+const { default: useCheckMetadataName } = await import(
+  '@/app/components/datasets/metadata/hooks/use-check-metadata-name',
+)
+
+// --- Factory functions ---
+
+const createMetadataItem = (
+  id: string,
+  name: string,
+  type = DataType.string,
+  count = 0,
+): MetadataItemWithValueLength => ({
+  id,
+  name,
+  type,
+  count,
+})
+
+const createMetadataList = (): MetadataItemWithValueLength[] => [
+  createMetadataItem('meta-1', 'author', DataType.string, 5),
+  createMetadataItem('meta-2', 'created_date', DataType.time, 10),
+  createMetadataItem('meta-3', 'page_count', DataType.number, 3),
+  createMetadataItem('meta-4', 'source_url', DataType.string, 8),
+  createMetadataItem('meta-5', 'version', DataType.number, 2),
+]
+
+describe('Metadata Management Flow - Cross-Module Validation Composition', () => {
+  describe('Name Validation Flow: Format Rules', () => {
+    it('should accept valid lowercase names with underscores', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+
+      expect(result.current.checkName('valid_name').errorMsg).toBe('')
+      expect(result.current.checkName('author').errorMsg).toBe('')
+      expect(result.current.checkName('page_count').errorMsg).toBe('')
+      expect(result.current.checkName('v2_field').errorMsg).toBe('')
+    })
+
+    it('should reject empty names', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+
+      expect(result.current.checkName('').errorMsg).toBeTruthy()
+    })
+
+    it('should reject names with invalid characters', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+
+      expect(result.current.checkName('Author').errorMsg).toBeTruthy()
+      expect(result.current.checkName('my-field').errorMsg).toBeTruthy()
+      expect(result.current.checkName('field name').errorMsg).toBeTruthy()
+      expect(result.current.checkName('1field').errorMsg).toBeTruthy()
+      expect(result.current.checkName('_private').errorMsg).toBeTruthy()
+    })
+
+    it('should reject names exceeding 255 characters', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+
+      const longName = 'a'.repeat(256)
+      expect(result.current.checkName(longName).errorMsg).toBeTruthy()
+
+      const maxName = 'a'.repeat(255)
+      expect(result.current.checkName(maxName).errorMsg).toBe('')
+    })
+  })
+
+  describe('Metadata Type Constraints: Enum Values Match Expected Set', () => {
+    it('should define exactly three data types', () => {
+      const typeValues = Object.values(DataType)
+      expect(typeValues).toHaveLength(3)
+    })
+
+    it('should include string, number, and time types', () => {
+      expect(DataType.string).toBe('string')
+      expect(DataType.number).toBe('number')
+      expect(DataType.time).toBe('time')
+    })
+
+    it('should use consistent types in metadata items', () => {
+      const metadataList = createMetadataList()
+
+      const stringItems = metadataList.filter(m => m.type === DataType.string)
+      const numberItems = metadataList.filter(m => m.type === DataType.number)
+      const timeItems = metadataList.filter(m => m.type === DataType.time)
+
+      expect(stringItems).toHaveLength(2)
+      expect(numberItems).toHaveLength(2)
+      expect(timeItems).toHaveLength(1)
+    })
+
+    it('should enforce type-safe metadata item construction', () => {
+      const item = createMetadataItem('test-1', 'test_field', DataType.number, 0)
+
+      expect(item.id).toBe('test-1')
+      expect(item.name).toBe('test_field')
+      expect(item.type).toBe(DataType.number)
+      expect(item.count).toBe(0)
+    })
+  })
+
+  describe('Duplicate Name Detection: Add Metadata → Check Name → Detect Duplicates', () => {
+    it('should detect duplicate names against an existing metadata list', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const checkDuplicate = (newName: string): boolean => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return false
+        return existingMetadata.some(m => m.name === newName)
+      }
+
+      expect(checkDuplicate('author')).toBe(true)
+      expect(checkDuplicate('created_date')).toBe(true)
+      expect(checkDuplicate('page_count')).toBe(true)
+    })
+
+    it('should allow names that do not conflict with existing metadata', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const isNameAvailable = (newName: string): boolean => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return false
+        return !existingMetadata.some(m => m.name === newName)
+      }
+
+      expect(isNameAvailable('category')).toBe(true)
+      expect(isNameAvailable('file_size')).toBe(true)
+      expect(isNameAvailable('language')).toBe(true)
+    })
+
+    it('should reject names that fail format validation before duplicate check', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+
+      const validateAndCheckDuplicate = (newName: string): { valid: boolean, reason: string } => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return { valid: false, reason: 'format' }
+        return { valid: true, reason: '' }
+      }
+
+      expect(validateAndCheckDuplicate('Author').reason).toBe('format')
+      expect(validateAndCheckDuplicate('').reason).toBe('format')
+      expect(validateAndCheckDuplicate('valid_name').valid).toBe(true)
+    })
+  })
+
+  describe('Name Uniqueness Across Edits: Rename Workflow', () => {
+    it('should allow an existing metadata item to keep its own name', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const isRenameValid = (itemId: string, newName: string): boolean => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return false
+        // Allow keeping the same name (skip self in duplicate check)
+        return !existingMetadata.some(m => m.name === newName && m.id !== itemId)
+      }
+
+      // Author keeping its own name should be valid
+      expect(isRenameValid('meta-1', 'author')).toBe(true)
+      // page_count keeping its own name should be valid
+      expect(isRenameValid('meta-3', 'page_count')).toBe(true)
+    })
+
+    it('should reject renaming to another existing metadata name', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const isRenameValid = (itemId: string, newName: string): boolean => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return false
+        return !existingMetadata.some(m => m.name === newName && m.id !== itemId)
+      }
+
+      // Author trying to rename to "page_count" (taken by meta-3)
+      expect(isRenameValid('meta-1', 'page_count')).toBe(false)
+      // version trying to rename to "source_url" (taken by meta-4)
+      expect(isRenameValid('meta-5', 'source_url')).toBe(false)
+    })
+
+    it('should allow renaming to a completely new valid name', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const isRenameValid = (itemId: string, newName: string): boolean => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return false
+        return !existingMetadata.some(m => m.name === newName && m.id !== itemId)
+      }
+
+      expect(isRenameValid('meta-1', 'document_author')).toBe(true)
+      expect(isRenameValid('meta-2', 'publish_date')).toBe(true)
+      expect(isRenameValid('meta-3', 'total_pages')).toBe(true)
+    })
+
+    it('should reject renaming with an invalid format even if name is unique', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const isRenameValid = (itemId: string, newName: string): boolean => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return false
+        return !existingMetadata.some(m => m.name === newName && m.id !== itemId)
+      }
+
+      expect(isRenameValid('meta-1', 'New Author')).toBe(false)
+      expect(isRenameValid('meta-2', '2024_date')).toBe(false)
+      expect(isRenameValid('meta-3', '')).toBe(false)
+    })
+  })
+
+  describe('Full Metadata Management Workflow', () => {
+    it('should support a complete add-validate-check-duplicate cycle', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const addMetadataField = (
+        name: string,
+        type: DataType,
+      ): { success: boolean, error?: string } => {
+        const formatCheck = result.current.checkName(name)
+        if (formatCheck.errorMsg)
+          return { success: false, error: 'invalid_format' }
+
+        if (existingMetadata.some(m => m.name === name))
+          return { success: false, error: 'duplicate_name' }
+
+        existingMetadata.push(createMetadataItem(`meta-${existingMetadata.length + 1}`, name, type))
+        return { success: true }
+      }
+
+      // Add a valid new field
+      const result1 = addMetadataField('department', DataType.string)
+      expect(result1.success).toBe(true)
+      expect(existingMetadata).toHaveLength(6)
+
+      // Try to add a duplicate
+      const result2 = addMetadataField('author', DataType.string)
+      expect(result2.success).toBe(false)
+      expect(result2.error).toBe('duplicate_name')
+      expect(existingMetadata).toHaveLength(6)
+
+      // Try to add an invalid name
+      const result3 = addMetadataField('Invalid Name', DataType.string)
+      expect(result3.success).toBe(false)
+      expect(result3.error).toBe('invalid_format')
+      expect(existingMetadata).toHaveLength(6)
+
+      // Add another valid field
+      const result4 = addMetadataField('priority_level', DataType.number)
+      expect(result4.success).toBe(true)
+      expect(existingMetadata).toHaveLength(7)
+    })
+
+    it('should support a complete rename workflow with validation chain', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+      const existingMetadata = createMetadataList()
+
+      const renameMetadataField = (
+        itemId: string,
+        newName: string,
+      ): { success: boolean, error?: string } => {
+        const formatCheck = result.current.checkName(newName)
+        if (formatCheck.errorMsg)
+          return { success: false, error: 'invalid_format' }
+
+        if (existingMetadata.some(m => m.name === newName && m.id !== itemId))
+          return { success: false, error: 'duplicate_name' }
+
+        const item = existingMetadata.find(m => m.id === itemId)
+        if (!item)
+          return { success: false, error: 'not_found' }
+
+        // Simulate the rename in-place
+        const index = existingMetadata.indexOf(item)
+        existingMetadata[index] = { ...item, name: newName }
+        return { success: true }
+      }
+
+      // Rename author to document_author
+      expect(renameMetadataField('meta-1', 'document_author').success).toBe(true)
+      expect(existingMetadata.find(m => m.id === 'meta-1')?.name).toBe('document_author')
+
+      // Try renaming created_date to page_count (already taken)
+      expect(renameMetadataField('meta-2', 'page_count').error).toBe('duplicate_name')
+
+      // Rename to invalid format
+      expect(renameMetadataField('meta-3', 'Page Count').error).toBe('invalid_format')
+
+      // Rename non-existent item
+      expect(renameMetadataField('meta-999', 'something').error).toBe('not_found')
+    })
+
+    it('should maintain validation consistency across multiple operations', () => {
+      const { result } = renderHook(() => useCheckMetadataName())
+
+      // Validate the same name multiple times for consistency
+      const name = 'consistent_field'
+      const results = Array.from({ length: 5 }, () => result.current.checkName(name))
+
+      expect(results.every(r => r.errorMsg === '')).toBe(true)
+
+      // Validate an invalid name multiple times
+      const invalidResults = Array.from({ length: 5 }, () => result.current.checkName('Invalid'))
+      expect(invalidResults.every(r => r.errorMsg !== '')).toBe(true)
+    })
+  })
+})

web/__tests__/datasets/pipeline-datasource-flow.test.tsx

@@ -0,0 +1,477 @@
+/**
+ * Integration Test: Pipeline Data Source Store Composition
+ *
+ * Tests cross-slice interactions in the pipeline data source Zustand store.
+ * The unit-level slice specs test each slice in isolation.
+ * This integration test verifies:
+ *   - Store initialization produces correct defaults across all slices
+ *   - Cross-slice coordination (e.g. credential shared across slices)
+ *   - State isolation: changes in one slice do not affect others
+ *   - Full workflow simulation through credential → source → data path
+ */
+
+import type { NotionPage } from '@/models/common'
+import type { CrawlResultItem, FileItem } from '@/models/datasets'
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { createDataSourceStore } from '@/app/components/datasets/documents/create-from-pipeline/data-source/store'
+import { CrawlStep } from '@/models/datasets'
+import { OnlineDriveFileType } from '@/models/pipeline'
+
+// --- Factory functions ---
+
+const createFileItem = (id: string): FileItem => ({
+  fileID: id,
+  file: { id, name: `${id}.txt`, size: 1024 } as FileItem['file'],
+  progress: 100,
+})
+
+const createCrawlResultItem = (url: string, title?: string): CrawlResultItem => ({
+  title: title ?? `Page: ${url}`,
+  markdown: `# ${title ?? url}\n\nContent for ${url}`,
+  description: `Description for ${url}`,
+  source_url: url,
+})
+
+const createOnlineDriveFile = (id: string, name: string, type = OnlineDriveFileType.file): OnlineDriveFile => ({
+  id,
+  name,
+  size: 2048,
+  type,
+})
+
+const createNotionPage = (pageId: string): NotionPage => ({
+  page_id: pageId,
+  page_name: `Page ${pageId}`,
+  page_icon: null,
+  is_bound: true,
+  parent_id: 'parent-1',
+  type: 'page',
+  workspace_id: 'ws-1',
+})
+
+describe('Pipeline Data Source Store Composition - Cross-Slice Integration', () => {
+  describe('Store Initialization → All Slices Have Correct Defaults', () => {
+    it('should create a store with all five slices combined', () => {
+      const store = createDataSourceStore()
+      const state = store.getState()
+
+      // Common slice defaults
+      expect(state.currentCredentialId).toBe('')
+      expect(state.currentNodeIdRef.current).toBe('')
+
+      // Local file slice defaults
+      expect(state.localFileList).toEqual([])
+      expect(state.currentLocalFile).toBeUndefined()
+
+      // Online document slice defaults
+      expect(state.documentsData).toEqual([])
+      expect(state.onlineDocuments).toEqual([])
+      expect(state.searchValue).toBe('')
+      expect(state.selectedPagesId).toEqual(new Set())
+
+      // Website crawl slice defaults
+      expect(state.websitePages).toEqual([])
+      expect(state.step).toBe(CrawlStep.init)
+      expect(state.previewIndex).toBe(-1)
+
+      // Online drive slice defaults
+      expect(state.breadcrumbs).toEqual([])
+      expect(state.prefix).toEqual([])
+      expect(state.keywords).toBe('')
+      expect(state.selectedFileIds).toEqual([])
+      expect(state.onlineDriveFileList).toEqual([])
+      expect(state.bucket).toBe('')
+      expect(state.hasBucket).toBe(false)
+    })
+  })
+
+  describe('Cross-Slice Coordination: Shared Credential', () => {
+    it('should set credential that is accessible from the common slice', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setCurrentCredentialId('cred-abc-123')
+
+      expect(store.getState().currentCredentialId).toBe('cred-abc-123')
+    })
+
+    it('should allow credential update independently of all other slices', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setLocalFileList([createFileItem('f1')])
+      store.getState().setCurrentCredentialId('cred-xyz')
+
+      expect(store.getState().currentCredentialId).toBe('cred-xyz')
+      expect(store.getState().localFileList).toHaveLength(1)
+    })
+  })
+
+  describe('Local File Workflow: Set Files → Verify List → Clear', () => {
+    it('should set and retrieve local file list', () => {
+      const store = createDataSourceStore()
+      const files = [createFileItem('f1'), createFileItem('f2'), createFileItem('f3')]
+
+      store.getState().setLocalFileList(files)
+
+      expect(store.getState().localFileList).toHaveLength(3)
+      expect(store.getState().localFileList[0].fileID).toBe('f1')
+      expect(store.getState().localFileList[2].fileID).toBe('f3')
+    })
+
+    it('should update preview ref when setting file list', () => {
+      const store = createDataSourceStore()
+      const files = [createFileItem('f-preview')]
+
+      store.getState().setLocalFileList(files)
+
+      expect(store.getState().previewLocalFileRef.current).toBeDefined()
+    })
+
+    it('should clear files by setting empty list', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setLocalFileList([createFileItem('f1')])
+      expect(store.getState().localFileList).toHaveLength(1)
+
+      store.getState().setLocalFileList([])
+      expect(store.getState().localFileList).toHaveLength(0)
+    })
+
+    it('should set and clear current local file selection', () => {
+      const store = createDataSourceStore()
+      const file = { id: 'current-file', name: 'current.txt' } as FileItem['file']
+
+      store.getState().setCurrentLocalFile(file)
+      expect(store.getState().currentLocalFile).toBeDefined()
+      expect(store.getState().currentLocalFile?.id).toBe('current-file')
+
+      store.getState().setCurrentLocalFile(undefined)
+      expect(store.getState().currentLocalFile).toBeUndefined()
+    })
+  })
+
+  describe('Online Document Workflow: Set Documents → Select Pages → Verify', () => {
+    it('should set documents data and online documents', () => {
+      const store = createDataSourceStore()
+      const pages = [createNotionPage('page-1'), createNotionPage('page-2')]
+
+      store.getState().setOnlineDocuments(pages)
+
+      expect(store.getState().onlineDocuments).toHaveLength(2)
+      expect(store.getState().onlineDocuments[0].page_id).toBe('page-1')
+    })
+
+    it('should update preview ref when setting online documents', () => {
+      const store = createDataSourceStore()
+      const pages = [createNotionPage('page-preview')]
+
+      store.getState().setOnlineDocuments(pages)
+
+      expect(store.getState().previewOnlineDocumentRef.current).toBeDefined()
+      expect(store.getState().previewOnlineDocumentRef.current?.page_id).toBe('page-preview')
+    })
+
+    it('should track selected page IDs', () => {
+      const store = createDataSourceStore()
+      const pages = [createNotionPage('p1'), createNotionPage('p2'), createNotionPage('p3')]
+
+      store.getState().setOnlineDocuments(pages)
+      store.getState().setSelectedPagesId(new Set(['p1', 'p3']))
+
+      expect(store.getState().selectedPagesId.size).toBe(2)
+      expect(store.getState().selectedPagesId.has('p1')).toBe(true)
+      expect(store.getState().selectedPagesId.has('p2')).toBe(false)
+      expect(store.getState().selectedPagesId.has('p3')).toBe(true)
+    })
+
+    it('should manage search value for filtering documents', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setSearchValue('meeting notes')
+
+      expect(store.getState().searchValue).toBe('meeting notes')
+    })
+
+    it('should set and clear current document selection', () => {
+      const store = createDataSourceStore()
+      const page = createNotionPage('current-page')
+
+      store.getState().setCurrentDocument(page)
+      expect(store.getState().currentDocument?.page_id).toBe('current-page')
+
+      store.getState().setCurrentDocument(undefined)
+      expect(store.getState().currentDocument).toBeUndefined()
+    })
+  })
+
+  describe('Website Crawl Workflow: Set Pages → Track Step → Preview', () => {
+    it('should set website pages and update preview ref', () => {
+      const store = createDataSourceStore()
+      const pages = [
+        createCrawlResultItem('https://example.com'),
+        createCrawlResultItem('https://example.com/about'),
+      ]
+
+      store.getState().setWebsitePages(pages)
+
+      expect(store.getState().websitePages).toHaveLength(2)
+      expect(store.getState().previewWebsitePageRef.current?.source_url).toBe('https://example.com')
+    })
+
+    it('should manage crawl step transitions', () => {
+      const store = createDataSourceStore()
+
+      expect(store.getState().step).toBe(CrawlStep.init)
+
+      store.getState().setStep(CrawlStep.running)
+      expect(store.getState().step).toBe(CrawlStep.running)
+
+      store.getState().setStep(CrawlStep.finished)
+      expect(store.getState().step).toBe(CrawlStep.finished)
+    })
+
+    it('should set crawl result with data and timing', () => {
+      const store = createDataSourceStore()
+      const result = {
+        data: [createCrawlResultItem('https://test.com')],
+        time_consuming: 3.5,
+      }
+
+      store.getState().setCrawlResult(result)
+
+      expect(store.getState().crawlResult?.data).toHaveLength(1)
+      expect(store.getState().crawlResult?.time_consuming).toBe(3.5)
+    })
+
+    it('should manage preview index for page navigation', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setPreviewIndex(2)
+      expect(store.getState().previewIndex).toBe(2)
+
+      store.getState().setPreviewIndex(-1)
+      expect(store.getState().previewIndex).toBe(-1)
+    })
+
+    it('should set and clear current website selection', () => {
+      const store = createDataSourceStore()
+      const page = createCrawlResultItem('https://current.com')
+
+      store.getState().setCurrentWebsite(page)
+      expect(store.getState().currentWebsite?.source_url).toBe('https://current.com')
+
+      store.getState().setCurrentWebsite(undefined)
+      expect(store.getState().currentWebsite).toBeUndefined()
+    })
+  })
+
+  describe('Online Drive Workflow: Breadcrumbs → File Selection → Navigation', () => {
+    it('should manage breadcrumb navigation', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setBreadcrumbs(['root', 'folder-a', 'subfolder'])
+
+      expect(store.getState().breadcrumbs).toEqual(['root', 'folder-a', 'subfolder'])
+    })
+
+    it('should support breadcrumb push/pop pattern', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setBreadcrumbs(['root'])
+      store.getState().setBreadcrumbs([...store.getState().breadcrumbs, 'level-1'])
+      store.getState().setBreadcrumbs([...store.getState().breadcrumbs, 'level-2'])
+
+      expect(store.getState().breadcrumbs).toEqual(['root', 'level-1', 'level-2'])
+
+      // Pop back one level
+      store.getState().setBreadcrumbs(store.getState().breadcrumbs.slice(0, -1))
+      expect(store.getState().breadcrumbs).toEqual(['root', 'level-1'])
+    })
+
+    it('should manage file list and selection', () => {
+      const store = createDataSourceStore()
+      const files = [
+        createOnlineDriveFile('drive-1', 'report.pdf'),
+        createOnlineDriveFile('drive-2', 'data.csv'),
+        createOnlineDriveFile('drive-3', 'images', OnlineDriveFileType.folder),
+      ]
+
+      store.getState().setOnlineDriveFileList(files)
+      expect(store.getState().onlineDriveFileList).toHaveLength(3)
+
+      store.getState().setSelectedFileIds(['drive-1', 'drive-2'])
+      expect(store.getState().selectedFileIds).toEqual(['drive-1', 'drive-2'])
+    })
+
+    it('should update preview ref when selecting files', () => {
+      const store = createDataSourceStore()
+      const files = [
+        createOnlineDriveFile('drive-a', 'file-a.txt'),
+        createOnlineDriveFile('drive-b', 'file-b.txt'),
+      ]
+
+      store.getState().setOnlineDriveFileList(files)
+      store.getState().setSelectedFileIds(['drive-b'])
+
+      expect(store.getState().previewOnlineDriveFileRef.current?.id).toBe('drive-b')
+    })
+
+    it('should manage bucket and prefix for S3-like navigation', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setBucket('my-data-bucket')
+      store.getState().setPrefix(['data', '2024'])
+      store.getState().setHasBucket(true)
+
+      expect(store.getState().bucket).toBe('my-data-bucket')
+      expect(store.getState().prefix).toEqual(['data', '2024'])
+      expect(store.getState().hasBucket).toBe(true)
+    })
+
+    it('should manage keywords for search filtering', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setKeywords('quarterly report')
+      expect(store.getState().keywords).toBe('quarterly report')
+    })
+  })
+
+  describe('State Isolation: Changes to One Slice Do Not Affect Others', () => {
+    it('should keep local file state independent from online document state', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setLocalFileList([createFileItem('local-1')])
+      store.getState().setOnlineDocuments([createNotionPage('notion-1')])
+
+      expect(store.getState().localFileList).toHaveLength(1)
+      expect(store.getState().onlineDocuments).toHaveLength(1)
+
+      // Clearing local files should not affect online documents
+      store.getState().setLocalFileList([])
+      expect(store.getState().localFileList).toHaveLength(0)
+      expect(store.getState().onlineDocuments).toHaveLength(1)
+    })
+
+    it('should keep website crawl state independent from online drive state', () => {
+      const store = createDataSourceStore()
+
+      store.getState().setWebsitePages([createCrawlResultItem('https://site.com')])
+      store.getState().setOnlineDriveFileList([createOnlineDriveFile('d1', 'file.txt')])
+
+      expect(store.getState().websitePages).toHaveLength(1)
+      expect(store.getState().onlineDriveFileList).toHaveLength(1)
+
+      // Clearing website pages should not affect drive files
+      store.getState().setWebsitePages([])
+      expect(store.getState().websitePages).toHaveLength(0)
+      expect(store.getState().onlineDriveFileList).toHaveLength(1)
+    })
+
+    it('should create fully independent store instances', () => {
+      const storeA = createDataSourceStore()
+      const storeB = createDataSourceStore()
+
+      storeA.getState().setCurrentCredentialId('cred-A')
+      storeA.getState().setLocalFileList([createFileItem('fa-1')])
+
+      expect(storeA.getState().currentCredentialId).toBe('cred-A')
+      expect(storeB.getState().currentCredentialId).toBe('')
+      expect(storeB.getState().localFileList).toEqual([])
+    })
+  })
+
+  describe('Full Workflow Simulation: Credential → Source → Data → Verify', () => {
+    it('should support a complete local file upload workflow', () => {
+      const store = createDataSourceStore()
+
+      // Step 1: Set credential
+      store.getState().setCurrentCredentialId('upload-cred-1')
+
+      // Step 2: Set file list
+      const files = [createFileItem('upload-1'), createFileItem('upload-2')]
+      store.getState().setLocalFileList(files)
+
+      // Step 3: Select current file for preview
+      store.getState().setCurrentLocalFile(files[0].file)
+
+      // Verify all state is consistent
+      expect(store.getState().currentCredentialId).toBe('upload-cred-1')
+      expect(store.getState().localFileList).toHaveLength(2)
+      expect(store.getState().currentLocalFile?.id).toBe('upload-1')
+      expect(store.getState().previewLocalFileRef.current).toBeDefined()
+    })
+
+    it('should support a complete website crawl workflow', () => {
+      const store = createDataSourceStore()
+
+      // Step 1: Set credential
+      store.getState().setCurrentCredentialId('crawl-cred-1')
+
+      // Step 2: Init crawl
+      store.getState().setStep(CrawlStep.running)
+
+      // Step 3: Crawl completes with results
+      const crawledPages = [
+        createCrawlResultItem('https://docs.example.com/guide'),
+        createCrawlResultItem('https://docs.example.com/api'),
+        createCrawlResultItem('https://docs.example.com/faq'),
+      ]
+      store.getState().setCrawlResult({ data: crawledPages, time_consuming: 12.5 })
+      store.getState().setStep(CrawlStep.finished)
+
+      // Step 4: Set website pages from results
+      store.getState().setWebsitePages(crawledPages)
+
+      // Step 5: Set preview
+      store.getState().setPreviewIndex(1)
+
+      // Verify all state
+      expect(store.getState().currentCredentialId).toBe('crawl-cred-1')
+      expect(store.getState().step).toBe(CrawlStep.finished)
+      expect(store.getState().websitePages).toHaveLength(3)
+      expect(store.getState().crawlResult?.time_consuming).toBe(12.5)
+      expect(store.getState().previewIndex).toBe(1)
+      expect(store.getState().previewWebsitePageRef.current?.source_url).toBe('https://docs.example.com/guide')
+    })
+
+    it('should support a complete online drive navigation workflow', () => {
+      const store = createDataSourceStore()
+
+      // Step 1: Set credential
+      store.getState().setCurrentCredentialId('drive-cred-1')
+
+      // Step 2: Set bucket
+      store.getState().setBucket('company-docs')
+      store.getState().setHasBucket(true)
+
+      // Step 3: Navigate into folders
+      store.getState().setBreadcrumbs(['company-docs'])
+      store.getState().setPrefix(['projects'])
+      const folderFiles = [
+        createOnlineDriveFile('proj-1', 'project-alpha', OnlineDriveFileType.folder),
+        createOnlineDriveFile('proj-2', 'project-beta', OnlineDriveFileType.folder),
+        createOnlineDriveFile('readme', 'README.md', OnlineDriveFileType.file),
+      ]
+      store.getState().setOnlineDriveFileList(folderFiles)
+
+      // Step 4: Navigate deeper
+      store.getState().setBreadcrumbs([...store.getState().breadcrumbs, 'project-alpha'])
+      store.getState().setPrefix([...store.getState().prefix, 'project-alpha'])
+
+      // Step 5: Select files
+      store.getState().setOnlineDriveFileList([
+        createOnlineDriveFile('doc-1', 'spec.pdf'),
+        createOnlineDriveFile('doc-2', 'design.fig'),
+      ])
+      store.getState().setSelectedFileIds(['doc-1'])
+
+      // Verify full state
+      expect(store.getState().currentCredentialId).toBe('drive-cred-1')
+      expect(store.getState().bucket).toBe('company-docs')
+      expect(store.getState().breadcrumbs).toEqual(['company-docs', 'project-alpha'])
+      expect(store.getState().prefix).toEqual(['projects', 'project-alpha'])
+      expect(store.getState().onlineDriveFileList).toHaveLength(2)
+      expect(store.getState().selectedFileIds).toEqual(['doc-1'])
+      expect(store.getState().previewOnlineDriveFileRef.current?.name).toBe('spec.pdf')
+    })
+  })
+})

web/__tests__/datasets/segment-crud.test.tsx

@@ -0,0 +1,300 @@
+/**
+ * Integration Test: Segment CRUD Flow
+ *
+ * Tests segment selection, search/filter, and modal state management across hooks.
+ * Validates cross-hook data contracts in the completed segment module.
+ */
+
+import type { SegmentDetailModel } from '@/models/datasets'
+import { act, renderHook } from '@testing-library/react'
+import { useModalState } from '@/app/components/datasets/documents/detail/completed/hooks/use-modal-state'
+import { useSearchFilter } from '@/app/components/datasets/documents/detail/completed/hooks/use-search-filter'
+import { useSegmentSelection } from '@/app/components/datasets/documents/detail/completed/hooks/use-segment-selection'
+
+const createSegment = (id: string, content = 'Test segment content'): SegmentDetailModel => ({
+  id,
+  position: 1,
+  document_id: 'doc-1',
+  content,
+  sign_content: content,
+  answer: '',
+  word_count: 50,
+  tokens: 25,
+  keywords: ['test'],
+  index_node_id: 'idx-1',
+  index_node_hash: 'hash-1',
+  hit_count: 0,
+  enabled: true,
+  disabled_at: 0,
+  disabled_by: '',
+  status: 'completed',
+  created_by: 'user-1',
+  created_at: Date.now(),
+  indexing_at: Date.now(),
+  completed_at: Date.now(),
+  error: null,
+  stopped_at: 0,
+  updated_at: Date.now(),
+  attachments: [],
+} as SegmentDetailModel)
+
+describe('Segment CRUD Flow', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Search and Filter → Segment List Query', () => {
+    it('should manage search input with debounce', () => {
+      vi.useFakeTimers()
+      const onPageChange = vi.fn()
+      const { result } = renderHook(() => useSearchFilter({ onPageChange }))
+
+      act(() => {
+        result.current.handleInputChange('keyword')
+      })
+
+      expect(result.current.inputValue).toBe('keyword')
+      expect(result.current.searchValue).toBe('')
+
+      act(() => {
+        vi.advanceTimersByTime(500)
+      })
+      expect(result.current.searchValue).toBe('keyword')
+      expect(onPageChange).toHaveBeenCalledWith(1)
+
+      vi.useRealTimers()
+    })
+
+    it('should manage status filter state', () => {
+      const onPageChange = vi.fn()
+      const { result } = renderHook(() => useSearchFilter({ onPageChange }))
+
+      // status value 1 maps to !!1 = true (enabled)
+      act(() => {
+        result.current.onChangeStatus({ value: 1, name: 'enabled' })
+      })
+      // onChangeStatus converts: value === 'all' ? 'all' : !!value
+      expect(result.current.selectedStatus).toBe(true)
+
+      act(() => {
+        result.current.onClearFilter()
+      })
+      expect(result.current.selectedStatus).toBe('all')
+      expect(result.current.inputValue).toBe('')
+    })
+
+    it('should provide status list for filter dropdown', () => {
+      const { result } = renderHook(() => useSearchFilter({ onPageChange: vi.fn() }))
+      expect(result.current.statusList).toBeInstanceOf(Array)
+      expect(result.current.statusList.length).toBe(3) // all, disabled, enabled
+    })
+
+    it('should compute selectDefaultValue based on selectedStatus', () => {
+      const { result } = renderHook(() => useSearchFilter({ onPageChange: vi.fn() }))
+
+      // Initial state: 'all'
+      expect(result.current.selectDefaultValue).toBe('all')
+
+      // Set to enabled (true)
+      act(() => {
+        result.current.onChangeStatus({ value: 1, name: 'enabled' })
+      })
+      expect(result.current.selectDefaultValue).toBe(1)
+
+      // Set to disabled (false)
+      act(() => {
+        result.current.onChangeStatus({ value: 0, name: 'disabled' })
+      })
+      expect(result.current.selectDefaultValue).toBe(0)
+    })
+  })
+
+  describe('Segment Selection → Batch Operations', () => {
+    const segments = [
+      createSegment('seg-1'),
+      createSegment('seg-2'),
+      createSegment('seg-3'),
+    ]
+
+    it('should manage individual segment selection', () => {
+      const { result } = renderHook(() => useSegmentSelection(segments))
+
+      act(() => {
+        result.current.onSelected('seg-1')
+      })
+      expect(result.current.selectedSegmentIds).toContain('seg-1')
+
+      act(() => {
+        result.current.onSelected('seg-2')
+      })
+      expect(result.current.selectedSegmentIds).toContain('seg-1')
+      expect(result.current.selectedSegmentIds).toContain('seg-2')
+      expect(result.current.selectedSegmentIds).toHaveLength(2)
+    })
+
+    it('should toggle selection on repeated click', () => {
+      const { result } = renderHook(() => useSegmentSelection(segments))
+
+      act(() => {
+        result.current.onSelected('seg-1')
+      })
+      expect(result.current.selectedSegmentIds).toContain('seg-1')
+
+      act(() => {
+        result.current.onSelected('seg-1')
+      })
+      expect(result.current.selectedSegmentIds).not.toContain('seg-1')
+    })
+
+    it('should support select all toggle', () => {
+      const { result } = renderHook(() => useSegmentSelection(segments))
+
+      act(() => {
+        result.current.onSelectedAll()
+      })
+      expect(result.current.selectedSegmentIds).toHaveLength(3)
+      expect(result.current.isAllSelected).toBe(true)
+
+      act(() => {
+        result.current.onSelectedAll()
+      })
+      expect(result.current.selectedSegmentIds).toHaveLength(0)
+      expect(result.current.isAllSelected).toBe(false)
+    })
+
+    it('should detect partial selection via isSomeSelected', () => {
+      const { result } = renderHook(() => useSegmentSelection(segments))
+
+      act(() => {
+        result.current.onSelected('seg-1')
+      })
+
+      // After selecting one of three, isSomeSelected should be true
+      expect(result.current.selectedSegmentIds).toEqual(['seg-1'])
+      expect(result.current.isSomeSelected).toBe(true)
+      expect(result.current.isAllSelected).toBe(false)
+    })
+
+    it('should clear selection via onCancelBatchOperation', () => {
+      const { result } = renderHook(() => useSegmentSelection(segments))
+
+      act(() => {
+        result.current.onSelected('seg-1')
+        result.current.onSelected('seg-2')
+      })
+      expect(result.current.selectedSegmentIds).toHaveLength(2)
+
+      act(() => {
+        result.current.onCancelBatchOperation()
+      })
+      expect(result.current.selectedSegmentIds).toHaveLength(0)
+    })
+  })
+
+  describe('Modal State Management', () => {
+    const onNewSegmentModalChange = vi.fn()
+
+    it('should open segment detail modal on card click', () => {
+      const { result } = renderHook(() => useModalState({ onNewSegmentModalChange }))
+
+      const segment = createSegment('seg-detail-1', 'Detail content')
+      act(() => {
+        result.current.onClickCard(segment)
+      })
+      expect(result.current.currSegment.showModal).toBe(true)
+      expect(result.current.currSegment.segInfo).toBeDefined()
+      expect(result.current.currSegment.segInfo!.id).toBe('seg-detail-1')
+    })
+
+    it('should close segment detail modal', () => {
+      const { result } = renderHook(() => useModalState({ onNewSegmentModalChange }))
+
+      const segment = createSegment('seg-1')
+      act(() => {
+        result.current.onClickCard(segment)
+      })
+      expect(result.current.currSegment.showModal).toBe(true)
+
+      act(() => {
+        result.current.onCloseSegmentDetail()
+      })
+      expect(result.current.currSegment.showModal).toBe(false)
+    })
+
+    it('should manage full screen toggle', () => {
+      const { result } = renderHook(() => useModalState({ onNewSegmentModalChange }))
+
+      expect(result.current.fullScreen).toBe(false)
+      act(() => {
+        result.current.toggleFullScreen()
+      })
+      expect(result.current.fullScreen).toBe(true)
+      act(() => {
+        result.current.toggleFullScreen()
+      })
+      expect(result.current.fullScreen).toBe(false)
+    })
+
+    it('should manage collapsed state', () => {
+      const { result } = renderHook(() => useModalState({ onNewSegmentModalChange }))
+
+      expect(result.current.isCollapsed).toBe(true)
+      act(() => {
+        result.current.toggleCollapsed()
+      })
+      expect(result.current.isCollapsed).toBe(false)
+    })
+
+    it('should manage new child segment modal', () => {
+      const { result } = renderHook(() => useModalState({ onNewSegmentModalChange }))
+
+      expect(result.current.showNewChildSegmentModal).toBe(false)
+      act(() => {
+        result.current.handleAddNewChildChunk('chunk-parent-1')
+      })
+      expect(result.current.showNewChildSegmentModal).toBe(true)
+      expect(result.current.currChunkId).toBe('chunk-parent-1')
+
+      act(() => {
+        result.current.onCloseNewChildChunkModal()
+      })
+      expect(result.current.showNewChildSegmentModal).toBe(false)
+    })
+  })
+
+  describe('Cross-Hook Data Flow: Search → Selection → Modal', () => {
+    it('should maintain independent state across all three hooks', () => {
+      const segments = [createSegment('seg-1'), createSegment('seg-2')]
+
+      const { result: filterResult } = renderHook(() =>
+        useSearchFilter({ onPageChange: vi.fn() }),
+      )
+      const { result: selectionResult } = renderHook(() =>
+        useSegmentSelection(segments),
+      )
+      const { result: modalResult } = renderHook(() =>
+        useModalState({ onNewSegmentModalChange: vi.fn() }),
+      )
+
+      // Set search filter to enabled
+      act(() => {
+        filterResult.current.onChangeStatus({ value: 1, name: 'enabled' })
+      })
+
+      // Select a segment
+      act(() => {
+        selectionResult.current.onSelected('seg-1')
+      })
+
+      // Open detail modal
+      act(() => {
+        modalResult.current.onClickCard(segments[0])
+      })
+
+      // All states should be independent
+      expect(filterResult.current.selectedStatus).toBe(true) // !!1
+      expect(selectionResult.current.selectedSegmentIds).toContain('seg-1')
+      expect(modalResult.current.currSegment.showModal).toBe(true)
+    })
+  })
+})

web/app/components/base/chat/embedded-chatbot/hooks.spec.tsx

@@ -162,8 +162,10 @@ describe('useEmbeddedChatbot', () => {
       await waitFor(() => {
         expect(mockFetchChatList).toHaveBeenCalledWith('conversation-1', AppSourceType.webApp, 'app-1')
       })
-      expect(result.current.pinnedConversationList).toEqual(pinnedData.data)
-      expect(result.current.conversationList).toEqual(listData.data)
+      await waitFor(() => {
+        expect(result.current.pinnedConversationList).toEqual(pinnedData.data)
+        expect(result.current.conversationList).toEqual(listData.data)
+      })
     })
   })
 

web/app/components/datasets/chunk.spec.tsx

@@ -1,111 +1,309 @@
-import { cleanup, render, screen } from '@testing-library/react'
-import { afterEach, describe, expect, it } from 'vitest'
+import type { QA } from '@/models/datasets'
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
 import { ChunkContainer, ChunkLabel, QAPreview } from './chunk'
 
-afterEach(() => {
-  cleanup()
-})
+vi.mock('../base/icons/src/public/knowledge', () => ({
+  SelectionMod: (props: React.ComponentProps<'svg'>) => (
+    <svg data-testid="selection-mod-icon" {...props} />
+  ),
+}))
+
+function createQA(overrides: Partial<QA> = {}): QA {
+  return {
+    question: 'What is Dify?',
+    answer: 'Dify is an open-source LLM app development platform.',
+    ...overrides,
+  }
+}
 
 describe('ChunkLabel', () => {
-  it('should render label text', () => {
-    render(<ChunkLabel label="Chunk 1" characterCount={100} />)
-    expect(screen.getByText('Chunk 1')).toBeInTheDocument()
+  beforeEach(() => {
+    vi.clearAllMocks()
   })
 
-  it('should render character count', () => {
-    render(<ChunkLabel label="Chunk 1" characterCount={150} />)
-    expect(screen.getByText('150 characters')).toBeInTheDocument()
+  describe('Rendering', () => {
+    it('should render the label text', () => {
+      render(<ChunkLabel label="Chunk #1" characterCount={100} />)
+
+      expect(screen.getByText('Chunk #1')).toBeInTheDocument()
+    })
+
+    it('should render the character count with unit', () => {
+      render(<ChunkLabel label="Chunk #1" characterCount={256} />)
+
+      expect(screen.getByText('256 characters')).toBeInTheDocument()
+    })
+
+    it('should render the SelectionMod icon', () => {
+      render(<ChunkLabel label="Chunk" characterCount={10} />)
+
+      expect(screen.getByTestId('selection-mod-icon')).toBeInTheDocument()
+    })
+
+    it('should render a middle dot separator between label and count', () => {
+      render(<ChunkLabel label="Chunk" characterCount={10} />)
+
+      expect(screen.getByText('·')).toBeInTheDocument()
+    })
   })
 
-  it('should render separator dot', () => {
-    render(<ChunkLabel label="Chunk 1" characterCount={100} />)
-    expect(screen.getByText('·')).toBeInTheDocument()
+  describe('Props', () => {
+    it('should display zero character count', () => {
+      render(<ChunkLabel label="Empty Chunk" characterCount={0} />)
+
+      expect(screen.getByText('0 characters')).toBeInTheDocument()
+    })
+
+    it('should display large character counts', () => {
+      render(<ChunkLabel label="Large" characterCount={999999} />)
+
+      expect(screen.getByText('999999 characters')).toBeInTheDocument()
+    })
   })
 
-  it('should render with zero character count', () => {
-    render(<ChunkLabel label="Empty Chunk" characterCount={0} />)
-    expect(screen.getByText('0 characters')).toBeInTheDocument()
-  })
+  describe('Edge Cases', () => {
+    it('should render with empty label', () => {
+      render(<ChunkLabel label="" characterCount={50} />)
 
-  it('should render with large character count', () => {
-    render(<ChunkLabel label="Large Chunk" characterCount={999999} />)
-    expect(screen.getByText('999999 characters')).toBeInTheDocument()
+      expect(screen.getByText('50 characters')).toBeInTheDocument()
+    })
+
+    it('should render with special characters in label', () => {
+      render(<ChunkLabel label="Chunk <#1> & 'test'" characterCount={10} />)
+
+      expect(screen.getByText('Chunk <#1> & \'test\'')).toBeInTheDocument()
+    })
   })
 })
 
+// Tests for ChunkContainer - wraps ChunkLabel with children content area
 describe('ChunkContainer', () => {
-  it('should render label and character count', () => {
-    render(<ChunkContainer label="Container 1" characterCount={200}>Content</ChunkContainer>)
-    expect(screen.getByText('Container 1')).toBeInTheDocument()
-    expect(screen.getByText('200 characters')).toBeInTheDocument()
+  beforeEach(() => {
+    vi.clearAllMocks()
   })
 
-  it('should render children content', () => {
-    render(<ChunkContainer label="Container 1" characterCount={200}>Test Content</ChunkContainer>)
-    expect(screen.getByText('Test Content')).toBeInTheDocument()
+  describe('Rendering', () => {
+    it('should render ChunkLabel with correct props', () => {
+      render(
+        <ChunkContainer label="Chunk #1" characterCount={200}>
+          Content here
+        </ChunkContainer>,
+      )
+
+      expect(screen.getByText('Chunk #1')).toBeInTheDocument()
+      expect(screen.getByText('200 characters')).toBeInTheDocument()
+    })
+
+    it('should render children in the content area', () => {
+      render(
+        <ChunkContainer label="Chunk" characterCount={50}>
+          <p>Paragraph content</p>
+        </ChunkContainer>,
+      )
+
+      expect(screen.getByText('Paragraph content')).toBeInTheDocument()
+    })
+
+    it('should render the SelectionMod icon via ChunkLabel', () => {
+      render(
+        <ChunkContainer label="Chunk" characterCount={10}>
+          Content
+        </ChunkContainer>,
+      )
+
+      expect(screen.getByTestId('selection-mod-icon')).toBeInTheDocument()
+    })
   })
 
-  it('should render with complex children', () => {
-    render(
-      <ChunkContainer label="Container" characterCount={100}>
-        <div data-testid="child-div">
-          <span>Nested content</span>
-        </div>
-      </ChunkContainer>,
-    )
-    expect(screen.getByTestId('child-div')).toBeInTheDocument()
-    expect(screen.getByText('Nested content')).toBeInTheDocument()
+  describe('Structure', () => {
+    it('should have space-y-2 on the outer container', () => {
+      const { container } = render(
+        <ChunkContainer label="Chunk" characterCount={10}>Content</ChunkContainer>,
+      )
+
+      expect(container.firstElementChild).toHaveClass('space-y-2')
+    })
+
+    it('should render children inside a styled content div', () => {
+      render(
+        <ChunkContainer label="Chunk" characterCount={10}>
+          <span>Test child</span>
+        </ChunkContainer>,
+      )
+
+      const contentDiv = screen.getByText('Test child').parentElement
+      expect(contentDiv).toHaveClass('body-md-regular', 'text-text-secondary')
+    })
   })
 
-  it('should render empty children', () => {
-    render(<ChunkContainer label="Empty" characterCount={0}>{null}</ChunkContainer>)
-    expect(screen.getByText('Empty')).toBeInTheDocument()
+  describe('Edge Cases', () => {
+    it('should render without children', () => {
+      const { container } = render(
+        <ChunkContainer label="Empty" characterCount={0} />,
+      )
+
+      expect(container.firstElementChild).toBeInTheDocument()
+      expect(screen.getByText('Empty')).toBeInTheDocument()
+    })
+
+    it('should render multiple children', () => {
+      render(
+        <ChunkContainer label="Multi" characterCount={100}>
+          <span>First</span>
+          <span>Second</span>
+        </ChunkContainer>,
+      )
+
+      expect(screen.getByText('First')).toBeInTheDocument()
+      expect(screen.getByText('Second')).toBeInTheDocument()
+    })
+
+    it('should render with string children', () => {
+      render(
+        <ChunkContainer label="Text" characterCount={5}>
+          Plain text content
+        </ChunkContainer>,
+      )
+
+      expect(screen.getByText('Plain text content')).toBeInTheDocument()
+    })
   })
 })
 
+// Tests for QAPreview - displays question and answer pair
 describe('QAPreview', () => {
-  const mockQA = {
-    question: 'What is the meaning of life?',
-    answer: 'The meaning of life is 42.',
-  }
-
-  it('should render question text', () => {
-    render(<QAPreview qa={mockQA} />)
-    expect(screen.getByText('What is the meaning of life?')).toBeInTheDocument()
+  beforeEach(() => {
+    vi.clearAllMocks()
   })
 
-  it('should render answer text', () => {
-    render(<QAPreview qa={mockQA} />)
-    expect(screen.getByText('The meaning of life is 42.')).toBeInTheDocument()
+  describe('Rendering', () => {
+    it('should render the question text', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      expect(screen.getByText('What is Dify?')).toBeInTheDocument()
+    })
+
+    it('should render the answer text', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      expect(screen.getByText('Dify is an open-source LLM app development platform.')).toBeInTheDocument()
+    })
+
+    it('should render Q and A labels', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      expect(screen.getByText('Q')).toBeInTheDocument()
+      expect(screen.getByText('A')).toBeInTheDocument()
+    })
   })
 
-  it('should render Q label', () => {
-    render(<QAPreview qa={mockQA} />)
-    expect(screen.getByText('Q')).toBeInTheDocument()
+  describe('Structure', () => {
+    it('should render Q label as a label element', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      const qLabel = screen.getByText('Q')
+      expect(qLabel.tagName).toBe('LABEL')
+    })
+
+    it('should render A label as a label element', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      const aLabel = screen.getByText('A')
+      expect(aLabel.tagName).toBe('LABEL')
+    })
+
+    it('should render question in a p element', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      const questionEl = screen.getByText(qa.question)
+      expect(questionEl.tagName).toBe('P')
+    })
+
+    it('should render answer in a p element', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      const answerEl = screen.getByText(qa.answer)
+      expect(answerEl.tagName).toBe('P')
+    })
+
+    it('should have the outer container with flex column layout', () => {
+      const qa = createQA()
+      const { container } = render(<QAPreview qa={qa} />)
+
+      expect(container.firstElementChild).toHaveClass('flex', 'flex-col', 'gap-y-2')
+    })
+
+    it('should apply text styling classes to question paragraph', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      const questionEl = screen.getByText(qa.question)
+      expect(questionEl).toHaveClass('body-md-regular', 'text-text-secondary')
+    })
+
+    it('should apply text styling classes to answer paragraph', () => {
+      const qa = createQA()
+      render(<QAPreview qa={qa} />)
+
+      const answerEl = screen.getByText(qa.answer)
+      expect(answerEl).toHaveClass('body-md-regular', 'text-text-secondary')
+    })
   })
 
-  it('should render A label', () => {
-    render(<QAPreview qa={mockQA} />)
-    expect(screen.getByText('A')).toBeInTheDocument()
-  })
+  describe('Edge Cases', () => {
+    it('should render with empty question', () => {
+      const qa = createQA({ question: '' })
+      render(<QAPreview qa={qa} />)
 
-  it('should render with empty strings', () => {
-    render(<QAPreview qa={{ question: '', answer: '' }} />)
-    expect(screen.getByText('Q')).toBeInTheDocument()
-    expect(screen.getByText('A')).toBeInTheDocument()
-  })
+      expect(screen.getByText('Q')).toBeInTheDocument()
+      expect(screen.getByText('A')).toBeInTheDocument()
+    })
 
-  it('should render with long text', () => {
-    const longQuestion = 'Q'.repeat(500)
-    const longAnswer = 'A'.repeat(500)
-    render(<QAPreview qa={{ question: longQuestion, answer: longAnswer }} />)
-    expect(screen.getByText(longQuestion)).toBeInTheDocument()
-    expect(screen.getByText(longAnswer)).toBeInTheDocument()
-  })
+    it('should render with empty answer', () => {
+      const qa = createQA({ answer: '' })
+      render(<QAPreview qa={qa} />)
 
-  it('should render with special characters', () => {
-    render(<QAPreview qa={{ question: 'What about <script>?', answer: '& special chars!' }} />)
-    expect(screen.getByText('What about <script>?')).toBeInTheDocument()
-    expect(screen.getByText('& special chars!')).toBeInTheDocument()
+      expect(screen.getByText('Q')).toBeInTheDocument()
+      expect(screen.getByText(qa.question)).toBeInTheDocument()
+    })
+
+    it('should render with long text', () => {
+      const longText = 'x'.repeat(1000)
+      const qa = createQA({ question: longText, answer: longText })
+      render(<QAPreview qa={qa} />)
+
+      const elements = screen.getAllByText(longText)
+      expect(elements).toHaveLength(2)
+    })
+
+    it('should render with special characters in question and answer', () => {
+      const qa = createQA({
+        question: 'What about <html> & "quotes"?',
+        answer: 'It handles \'single\' & "double" quotes.',
+      })
+      render(<QAPreview qa={qa} />)
+
+      expect(screen.getByText('What about <html> & "quotes"?')).toBeInTheDocument()
+      expect(screen.getByText('It handles \'single\' & "double" quotes.')).toBeInTheDocument()
+    })
+
+    it('should render with multiline text', () => {
+      const qa = createQA({
+        question: 'Line1\nLine2',
+        answer: 'Answer1\nAnswer2',
+      })
+      render(<QAPreview qa={qa} />)
+
+      expect(screen.getByText(/Line1/)).toBeInTheDocument()
+      expect(screen.getByText(/Answer1/)).toBeInTheDocument()
+    })
   })
 })

web/app/components/datasets/common/document-picker/document-list.spec.tsx

@@ -0,0 +1,49 @@
+import type { DocumentItem } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import DocumentList from './document-list'
+
+vi.mock('../document-file-icon', () => ({
+  default: ({ name, extension }: { name?: string, extension?: string }) => (
+    <span data-testid="file-icon">
+      {name}
+      .
+      {extension}
+    </span>
+  ),
+}))
+
+describe('DocumentList', () => {
+  const mockList = [
+    { id: 'doc-1', name: 'report', extension: 'pdf' },
+    { id: 'doc-2', name: 'data', extension: 'csv' },
+  ] as DocumentItem[]
+
+  const onChange = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render all documents', () => {
+    render(<DocumentList list={mockList} onChange={onChange} />)
+    expect(screen.getByText('report')).toBeInTheDocument()
+    expect(screen.getByText('data')).toBeInTheDocument()
+  })
+
+  it('should render file icons', () => {
+    render(<DocumentList list={mockList} onChange={onChange} />)
+    expect(screen.getAllByTestId('file-icon')).toHaveLength(2)
+  })
+
+  it('should call onChange with document on click', () => {
+    render(<DocumentList list={mockList} onChange={onChange} />)
+    fireEvent.click(screen.getByText('report'))
+    expect(onChange).toHaveBeenCalledWith(mockList[0])
+  })
+
+  it('should render empty list without errors', () => {
+    const { container } = render(<DocumentList list={[]} onChange={onChange} />)
+    expect(container.firstChild).toBeInTheDocument()
+  })
+})

web/app/components/datasets/create/embedding-process/indexing-progress-item.spec.tsx

@@ -0,0 +1,145 @@
+/* eslint-disable next/no-img-element */
+import type { ReactNode } from 'react'
+import type { IndexingStatusResponse } from '@/models/datasets'
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DataSourceType } from '@/models/datasets'
+import IndexingProgressItem from './indexing-progress-item'
+
+vi.mock('next/image', () => ({
+  default: (props: React.ImgHTMLAttributes<HTMLImageElement>) => <img {...props} />,
+}))
+vi.mock('@/app/components/billing/priority-label', () => ({
+  default: () => <span data-testid="priority-label">Priority</span>,
+}))
+vi.mock('../../common/document-file-icon', () => ({
+  default: ({ name }: { name?: string }) => <span data-testid="file-icon">{name}</span>,
+}))
+vi.mock('@/app/components/base/notion-icon', () => ({
+  default: ({ src }: { src?: string }) => <span data-testid="notion-icon">{src}</span>,
+}))
+vi.mock('@/app/components/base/tooltip', () => ({
+  default: ({ children, popupContent }: { children?: ReactNode, popupContent?: ReactNode }) => (
+    <div data-testid="tooltip" data-content={popupContent}>{children}</div>
+  ),
+}))
+
+describe('IndexingProgressItem', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  const makeDetail = (overrides: Partial<IndexingStatusResponse> = {}): IndexingStatusResponse => ({
+    id: 'doc-1',
+    indexing_status: 'indexing',
+    processing_started_at: 0,
+    parsing_completed_at: 0,
+    cleaning_completed_at: 0,
+    splitting_completed_at: 0,
+    completed_at: null,
+    paused_at: null,
+    error: null,
+    stopped_at: null,
+    completed_segments: 50,
+    total_segments: 100,
+    ...overrides,
+  })
+
+  it('should render name and progress for embedding status', () => {
+    render(
+      <IndexingProgressItem
+        detail={makeDetail()}
+        name="test.pdf"
+        sourceType={DataSourceType.FILE}
+      />,
+    )
+
+    // Name appears in both the file-icon mock and the display div; verify at least one
+    expect(screen.getAllByText('test.pdf').length).toBeGreaterThanOrEqual(1)
+    expect(screen.getByText('50%')).toBeInTheDocument()
+  })
+
+  it('should render file icon for FILE source type', () => {
+    render(
+      <IndexingProgressItem
+        detail={makeDetail()}
+        name="report.docx"
+        sourceType={DataSourceType.FILE}
+      />,
+    )
+
+    expect(screen.getByTestId('file-icon')).toBeInTheDocument()
+  })
+
+  it('should render notion icon for NOTION source type', () => {
+    render(
+      <IndexingProgressItem
+        detail={makeDetail()}
+        name="My Page"
+        sourceType={DataSourceType.NOTION}
+        notionIcon="notion-icon-url"
+      />,
+    )
+
+    expect(screen.getByTestId('notion-icon')).toBeInTheDocument()
+  })
+
+  it('should render success icon for completed status', () => {
+    render(
+      <IndexingProgressItem
+        detail={makeDetail({ indexing_status: 'completed' })}
+        name="done.pdf"
+      />,
+    )
+
+    // No progress percentage should be shown for completed
+    expect(screen.queryByText('%')).not.toBeInTheDocument()
+  })
+
+  it('should render error icon with tooltip for error status', () => {
+    render(
+      <IndexingProgressItem
+        detail={makeDetail({ indexing_status: 'error', error: 'Parse failed' })}
+        name="broken.pdf"
+      />,
+    )
+
+    expect(screen.getByTestId('tooltip')).toHaveAttribute('data-content', 'Parse failed')
+  })
+
+  it('should show priority label when billing is enabled', () => {
+    render(
+      <IndexingProgressItem
+        detail={makeDetail()}
+        name="test.pdf"
+        enableBilling={true}
+      />,
+    )
+
+    expect(screen.getByTestId('priority-label')).toBeInTheDocument()
+  })
+
+  it('should not show priority label when billing is disabled', () => {
+    render(
+      <IndexingProgressItem
+        detail={makeDetail()}
+        name="test.pdf"
+        enableBilling={false}
+      />,
+    )
+
+    expect(screen.queryByTestId('priority-label')).not.toBeInTheDocument()
+  })
+
+  it('should apply error styling for error status', () => {
+    const { container } = render(
+      <IndexingProgressItem
+        detail={makeDetail({ indexing_status: 'error' })}
+        name="error.pdf"
+      />,
+    )
+
+    const wrapper = container.firstChild as HTMLElement
+    expect(wrapper.className).toContain('bg-state-destructive-hover-alt')
+  })
+})

web/app/components/datasets/create/embedding-process/rule-detail.spec.tsx

@@ -0,0 +1,154 @@
+/* eslint-disable next/no-img-element */
+import type { ProcessRuleResponse } from '@/models/datasets'
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { ProcessMode } from '@/models/datasets'
+import { RETRIEVE_METHOD } from '@/types/app'
+import RuleDetail from './rule-detail'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string, opts?: Record<string, unknown>) => `${opts?.ns ? `${opts.ns}.` : ''}${key}`,
+  }),
+}))
+vi.mock('next/image', () => ({
+  default: (props: React.ImgHTMLAttributes<HTMLImageElement>) => <img {...props} />,
+}))
+vi.mock('@/app/components/datasets/documents/detail/metadata', () => ({
+  FieldInfo: ({ label, displayedValue }: { label: string, displayedValue: string }) => (
+    <div data-testid="field-info">
+      <span data-testid="field-label">{label}</span>
+      <span data-testid="field-value">{displayedValue}</span>
+    </div>
+  ),
+}))
+vi.mock('../icons', () => ({
+  indexMethodIcon: { economical: '/icons/economical.svg', high_quality: '/icons/hq.svg' },
+  retrievalIcon: { fullText: '/icons/ft.svg', hybrid: '/icons/hy.svg', vector: '/icons/vec.svg' },
+}))
+
+describe('RuleDetail', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  const makeSourceData = (overrides: Partial<ProcessRuleResponse> = {}): ProcessRuleResponse => ({
+    mode: ProcessMode.general,
+    rules: {
+      segmentation: { separator: '\n', max_tokens: 500, chunk_overlap: 50 },
+      pre_processing_rules: [
+        { id: 'remove_extra_spaces', enabled: true },
+        { id: 'remove_urls_emails', enabled: false },
+      ],
+    },
+    ...overrides,
+  } as ProcessRuleResponse)
+
+  it('should render mode, segment length, text cleaning, index mode, and retrieval fields', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData()}
+        indexingType="high_quality"
+        retrievalMethod={RETRIEVE_METHOD.semantic}
+      />,
+    )
+
+    const fieldInfos = screen.getAllByTestId('field-info')
+    // mode, segmentLength, textCleaning, indexMode, retrievalSetting = 5
+    expect(fieldInfos.length).toBe(5)
+  })
+
+  it('should display "custom" for general mode', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData({ mode: ProcessMode.general })}
+        indexingType="high_quality"
+      />,
+    )
+
+    const values = screen.getAllByTestId('field-value')
+    expect(values[0].textContent).toContain('embedding.custom')
+  })
+
+  it('should display hierarchical mode with parent mode label', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData({
+          mode: ProcessMode.parentChild,
+          rules: {
+            parent_mode: 'paragraph',
+            segmentation: { separator: '\n', max_tokens: 1000, chunk_overlap: 50 },
+            subchunk_segmentation: { max_tokens: 200 },
+            pre_processing_rules: [],
+          } as unknown as ProcessRuleResponse['rules'],
+        })}
+        indexingType="high_quality"
+      />,
+    )
+
+    const values = screen.getAllByTestId('field-value')
+    expect(values[0].textContent).toContain('embedding.hierarchical')
+  })
+
+  it('should display "-" when no sourceData mode', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData({ mode: undefined as unknown as ProcessMode })}
+        indexingType="high_quality"
+      />,
+    )
+
+    const values = screen.getAllByTestId('field-value')
+    expect(values[0].textContent).toBe('-')
+  })
+
+  it('should display segment length for general mode', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData()}
+        indexingType="high_quality"
+      />,
+    )
+
+    const values = screen.getAllByTestId('field-value')
+    expect(values[1].textContent).toBe('500')
+  })
+
+  it('should display enabled pre-processing rules', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData()}
+        indexingType="high_quality"
+      />,
+    )
+
+    const values = screen.getAllByTestId('field-value')
+    // Only remove_extra_spaces is enabled
+    expect(values[2].textContent).toContain('stepTwo.removeExtraSpaces')
+  })
+
+  it('should display economical index mode', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData()}
+        indexingType="economy"
+      />,
+    )
+
+    const values = screen.getAllByTestId('field-value')
+    // Index mode field is 4th (index 3)
+    expect(values[3].textContent).toContain('stepTwo.economical')
+  })
+
+  it('should display qualified index mode for high_quality', () => {
+    render(
+      <RuleDetail
+        sourceData={makeSourceData()}
+        indexingType="high_quality"
+      />,
+    )
+
+    const values = screen.getAllByTestId('field-value')
+    expect(values[3].textContent).toContain('stepTwo.qualified')
+  })
+})

web/app/components/datasets/create/embedding-process/upgrade-banner.spec.tsx

@@ -0,0 +1,34 @@
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import UpgradeBanner from './upgrade-banner'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
+}))
+vi.mock('@/app/components/base/icons/src/vender/solid/general', () => ({
+  ZapFast: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="zap-icon" {...props} />,
+}))
+vi.mock('@/app/components/billing/upgrade-btn', () => ({
+  default: ({ loc }: { loc: string }) => <button data-testid="upgrade-btn" data-loc={loc}>Upgrade</button>,
+}))
+
+describe('UpgradeBanner', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render the banner with icon, text, and upgrade button', () => {
+    render(<UpgradeBanner />)
+
+    expect(screen.getByTestId('zap-icon')).toBeInTheDocument()
+    expect(screen.getByText('plansCommon.documentProcessingPriorityUpgrade')).toBeInTheDocument()
+    expect(screen.getByTestId('upgrade-btn')).toBeInTheDocument()
+  })
+
+  it('should pass correct loc to UpgradeBtn', () => {
+    render(<UpgradeBanner />)
+    expect(screen.getByTestId('upgrade-btn')).toHaveAttribute('data-loc', 'knowledge-speed-up')
+  })
+})

web/app/components/datasets/create/embedding-process/use-indexing-status-polling.spec.ts

@@ -0,0 +1,179 @@
+import { act, renderHook } from '@testing-library/react'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import { useIndexingStatusPolling } from './use-indexing-status-polling'
+
+const mockFetchIndexingStatusBatch = vi.fn()
+
+vi.mock('@/service/datasets', () => ({
+  fetchIndexingStatusBatch: (...args: unknown[]) => mockFetchIndexingStatusBatch(...args),
+}))
+
+describe('useIndexingStatusPolling', () => {
+  beforeEach(() => {
+    vi.useFakeTimers()
+    vi.clearAllMocks()
+  })
+
+  afterEach(() => {
+    vi.useRealTimers()
+  })
+
+  const defaultParams = { datasetId: 'ds-1', batchId: 'batch-1' }
+
+  it('should initialize with empty status list', async () => {
+    mockFetchIndexingStatusBatch.mockReturnValue(new Promise(() => {}))
+    const { result } = renderHook(() => useIndexingStatusPolling(defaultParams))
+
+    expect(result.current.statusList).toEqual([])
+    expect(result.current.isEmbedding).toBe(false)
+    expect(result.current.isEmbeddingCompleted).toBe(false)
+  })
+
+  it('should fetch status on mount and update state', async () => {
+    mockFetchIndexingStatusBatch.mockResolvedValue({
+      data: [{ indexing_status: 'indexing', completed_segments: 5, total_segments: 10 }],
+    })
+
+    const { result } = renderHook(() => useIndexingStatusPolling(defaultParams))
+    // Flush the resolved promise
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+
+    expect(mockFetchIndexingStatusBatch).toHaveBeenCalledWith({
+      datasetId: 'ds-1',
+      batchId: 'batch-1',
+    })
+    expect(result.current.statusList).toHaveLength(1)
+    expect(result.current.isEmbedding).toBe(true)
+    expect(result.current.isEmbeddingCompleted).toBe(false)
+  })
+
+  it('should stop polling when all completed', async () => {
+    mockFetchIndexingStatusBatch.mockResolvedValue({
+      data: [{ indexing_status: 'completed' }],
+    })
+
+    const { result } = renderHook(() => useIndexingStatusPolling(defaultParams))
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+
+    expect(result.current.isEmbeddingCompleted).toBe(true)
+    expect(result.current.isEmbedding).toBe(false)
+
+    // Should not schedule another poll
+    const callCount = mockFetchIndexingStatusBatch.mock.calls.length
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(5000)
+    })
+    expect(mockFetchIndexingStatusBatch).toHaveBeenCalledTimes(callCount)
+  })
+
+  it('should continue polling on fetch error', async () => {
+    mockFetchIndexingStatusBatch
+      .mockRejectedValueOnce(new Error('network'))
+      .mockResolvedValueOnce({
+        data: [{ indexing_status: 'completed' }],
+      })
+
+    const { result } = renderHook(() => useIndexingStatusPolling(defaultParams))
+    // First call: rejects
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+    // Advance past polling interval for retry
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(2500)
+    })
+
+    expect(result.current.isEmbeddingCompleted).toBe(true)
+  })
+
+  it('should detect embedding statuses', async () => {
+    mockFetchIndexingStatusBatch.mockResolvedValue({
+      data: [
+        { indexing_status: 'splitting' },
+        { indexing_status: 'parsing' },
+      ],
+    })
+
+    const { result } = renderHook(() => useIndexingStatusPolling(defaultParams))
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+
+    expect(result.current.isEmbedding).toBe(true)
+    expect(result.current.isEmbeddingCompleted).toBe(false)
+  })
+
+  it('should detect mixed statuses (some completed, some embedding)', async () => {
+    mockFetchIndexingStatusBatch.mockResolvedValue({
+      data: [
+        { indexing_status: 'completed' },
+        { indexing_status: 'indexing' },
+      ],
+    })
+
+    const { result } = renderHook(() => useIndexingStatusPolling(defaultParams))
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+
+    expect(result.current.statusList).toHaveLength(2)
+    expect(result.current.isEmbedding).toBe(true)
+    expect(result.current.isEmbeddingCompleted).toBe(false)
+  })
+
+  it('should cleanup on unmount', async () => {
+    mockFetchIndexingStatusBatch.mockResolvedValue({
+      data: [{ indexing_status: 'indexing' }],
+    })
+
+    const { unmount } = renderHook(() => useIndexingStatusPolling(defaultParams))
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+
+    const callCount = mockFetchIndexingStatusBatch.mock.calls.length
+    unmount()
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(5000)
+    })
+    expect(mockFetchIndexingStatusBatch).toHaveBeenCalledTimes(callCount)
+  })
+
+  it('should treat error and paused as completed statuses', async () => {
+    mockFetchIndexingStatusBatch.mockResolvedValue({
+      data: [
+        { indexing_status: 'error' },
+        { indexing_status: 'paused' },
+      ],
+    })
+
+    const { result } = renderHook(() => useIndexingStatusPolling(defaultParams))
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+
+    expect(result.current.isEmbeddingCompleted).toBe(true)
+    expect(result.current.isEmbedding).toBe(false)
+  })
+
+  it('should poll at 2500ms intervals', async () => {
+    mockFetchIndexingStatusBatch.mockResolvedValue({
+      data: [{ indexing_status: 'indexing' }],
+    })
+
+    renderHook(() => useIndexingStatusPolling(defaultParams))
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(0)
+    })
+    expect(mockFetchIndexingStatusBatch).toHaveBeenCalledTimes(1)
+
+    await act(async () => {
+      await vi.advanceTimersByTimeAsync(2500)
+    })
+    expect(mockFetchIndexingStatusBatch).toHaveBeenCalledTimes(2)
+  })
+})

web/app/components/datasets/create/embedding-process/utils.spec.ts

@@ -0,0 +1,140 @@
+import type { DataSourceInfo, FullDocumentDetail, IndexingStatusResponse } from '@/models/datasets'
+import { describe, expect, it } from 'vitest'
+import { createDocumentLookup, getFileType, getSourcePercent, isLegacyDataSourceInfo, isSourceEmbedding } from './utils'
+
+describe('isLegacyDataSourceInfo', () => {
+  it('should return true when upload_file object exists', () => {
+    const info = { upload_file: { id: '1', name: 'test.pdf' } } as unknown as DataSourceInfo
+    expect(isLegacyDataSourceInfo(info)).toBe(true)
+  })
+
+  it('should return false when upload_file is absent', () => {
+    const info = { notion_page_icon: 'icon' } as unknown as DataSourceInfo
+    expect(isLegacyDataSourceInfo(info)).toBe(false)
+  })
+
+  it('should return false for null', () => {
+    expect(isLegacyDataSourceInfo(null as unknown as DataSourceInfo)).toBe(false)
+  })
+
+  it('should return false when upload_file is a string', () => {
+    const info = { upload_file: 'not-an-object' } as unknown as DataSourceInfo
+    expect(isLegacyDataSourceInfo(info)).toBe(false)
+  })
+})
+
+describe('isSourceEmbedding', () => {
+  const embeddingStatuses = ['indexing', 'splitting', 'parsing', 'cleaning', 'waiting']
+  const nonEmbeddingStatuses = ['completed', 'error', 'paused', 'unknown']
+
+  it.each(embeddingStatuses)('should return true for status "%s"', (status) => {
+    expect(isSourceEmbedding({ indexing_status: status } as IndexingStatusResponse)).toBe(true)
+  })
+
+  it.each(nonEmbeddingStatuses)('should return false for status "%s"', (status) => {
+    expect(isSourceEmbedding({ indexing_status: status } as IndexingStatusResponse)).toBe(false)
+  })
+})
+
+describe('getSourcePercent', () => {
+  it('should calculate correct percentage', () => {
+    expect(getSourcePercent({ completed_segments: 50, total_segments: 100 } as IndexingStatusResponse)).toBe(50)
+  })
+
+  it('should return 0 when total is 0', () => {
+    expect(getSourcePercent({ completed_segments: 0, total_segments: 0 } as IndexingStatusResponse)).toBe(0)
+  })
+
+  it('should cap at 100', () => {
+    expect(getSourcePercent({ completed_segments: 150, total_segments: 100 } as IndexingStatusResponse)).toBe(100)
+  })
+
+  it('should round to nearest integer', () => {
+    expect(getSourcePercent({ completed_segments: 1, total_segments: 3 } as IndexingStatusResponse)).toBe(33)
+  })
+
+  it('should handle undefined segments as 0', () => {
+    expect(getSourcePercent({} as IndexingStatusResponse)).toBe(0)
+  })
+})
+
+describe('getFileType', () => {
+  it('should extract extension from filename', () => {
+    expect(getFileType('document.pdf')).toBe('pdf')
+  })
+
+  it('should return last extension for multi-dot names', () => {
+    expect(getFileType('archive.tar.gz')).toBe('gz')
+  })
+
+  it('should default to "txt" for undefined', () => {
+    expect(getFileType(undefined)).toBe('txt')
+  })
+
+  it('should default to "txt" for empty string', () => {
+    expect(getFileType('')).toBe('txt')
+  })
+})
+
+describe('createDocumentLookup', () => {
+  const documents = [
+    {
+      id: 'doc-1',
+      name: 'test.pdf',
+      data_source_type: 'upload_file',
+      data_source_info: {
+        upload_file: { id: 'f1', name: 'test.pdf' },
+        notion_page_icon: undefined,
+      },
+    },
+    {
+      id: 'doc-2',
+      name: 'notion-page',
+      data_source_type: 'notion_import',
+      data_source_info: {
+        upload_file: { id: 'f2', name: '' },
+        notion_page_icon: 'https://icon.url',
+      },
+    },
+  ] as unknown as FullDocumentDetail[]
+
+  it('should get document by id', () => {
+    const lookup = createDocumentLookup(documents)
+    expect(lookup.getDocument('doc-1')).toBe(documents[0])
+  })
+
+  it('should return undefined for non-existent id', () => {
+    const lookup = createDocumentLookup(documents)
+    expect(lookup.getDocument('non-existent')).toBeUndefined()
+  })
+
+  it('should get name by id', () => {
+    const lookup = createDocumentLookup(documents)
+    expect(lookup.getName('doc-1')).toBe('test.pdf')
+  })
+
+  it('should get source type by id', () => {
+    const lookup = createDocumentLookup(documents)
+    expect(lookup.getSourceType('doc-1')).toBe('upload_file')
+  })
+
+  it('should get notion icon for legacy data source', () => {
+    const lookup = createDocumentLookup(documents)
+    expect(lookup.getNotionIcon('doc-2')).toBe('https://icon.url')
+  })
+
+  it('should return undefined notion icon for non-legacy info', () => {
+    const docs = [{
+      id: 'doc-3',
+      data_source_info: { some_other: 'field' },
+    }] as unknown as FullDocumentDetail[]
+    const lookup = createDocumentLookup(docs)
+    expect(lookup.getNotionIcon('doc-3')).toBeUndefined()
+  })
+
+  it('should handle empty documents list', () => {
+    const lookup = createDocumentLookup([])
+    expect(lookup.getDocument('any')).toBeUndefined()
+    expect(lookup.getName('any')).toBeUndefined()
+  })
+})

web/app/components/datasets/create/step-one/components/data-source-type-selector.spec.tsx

@@ -0,0 +1,66 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DataSourceType } from '@/models/datasets'
+
+// Mock config to control web crawl feature flags
+vi.mock('@/config', () => ({
+  ENABLE_WEBSITE_FIRECRAWL: true,
+  ENABLE_WEBSITE_JINAREADER: true,
+  ENABLE_WEBSITE_WATERCRAWL: false,
+}))
+
+// Mock CSS module
+vi.mock('../../index.module.css', () => ({
+  default: {
+    dataSourceItem: 'ds-item',
+    active: 'active',
+    disabled: 'disabled',
+    datasetIcon: 'icon',
+    notion: 'notion-icon',
+    web: 'web-icon',
+  },
+}))
+
+const { default: DataSourceTypeSelector } = await import('./data-source-type-selector')
+
+describe('DataSourceTypeSelector', () => {
+  const defaultProps = {
+    currentType: DataSourceType.FILE,
+    disabled: false,
+    onChange: vi.fn(),
+    onClearPreviews: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('rendering', () => {
+    it('should render file, notion, and web options', () => {
+      render(<DataSourceTypeSelector {...defaultProps} />)
+      expect(screen.getByText('datasetCreation.stepOne.dataSourceType.file')).toBeInTheDocument()
+      expect(screen.getByText('datasetCreation.stepOne.dataSourceType.notion')).toBeInTheDocument()
+      expect(screen.getByText('datasetCreation.stepOne.dataSourceType.web')).toBeInTheDocument()
+    })
+
+    it('should render as a 3-column grid', () => {
+      const { container } = render(<DataSourceTypeSelector {...defaultProps} />)
+      expect(container.firstElementChild).toHaveClass('grid-cols-3')
+    })
+  })
+
+  describe('interactions', () => {
+    it('should call onChange and onClearPreviews on type click', () => {
+      render(<DataSourceTypeSelector {...defaultProps} />)
+      fireEvent.click(screen.getByText('datasetCreation.stepOne.dataSourceType.notion'))
+      expect(defaultProps.onChange).toHaveBeenCalledWith(DataSourceType.NOTION)
+      expect(defaultProps.onClearPreviews).toHaveBeenCalledWith(DataSourceType.NOTION)
+    })
+
+    it('should not call onChange when disabled', () => {
+      render(<DataSourceTypeSelector {...defaultProps} disabled />)
+      fireEvent.click(screen.getByText('datasetCreation.stepOne.dataSourceType.notion'))
+      expect(defaultProps.onChange).not.toHaveBeenCalled()
+    })
+  })
+})

web/app/components/datasets/create/step-one/components/next-step-button.spec.tsx

@@ -0,0 +1,48 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import NextStepButton from './next-step-button'
+
+describe('NextStepButton', () => {
+  const defaultProps = {
+    disabled: false,
+    onClick: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render button text', () => {
+    render(<NextStepButton {...defaultProps} />)
+    expect(screen.getByText('datasetCreation.stepOne.button')).toBeInTheDocument()
+  })
+
+  it('should render a primary variant button', () => {
+    render(<NextStepButton {...defaultProps} />)
+    const btn = screen.getByRole('button')
+    expect(btn).toBeInTheDocument()
+  })
+
+  it('should call onClick when clicked', () => {
+    render(<NextStepButton {...defaultProps} />)
+    fireEvent.click(screen.getByRole('button'))
+    expect(defaultProps.onClick).toHaveBeenCalledOnce()
+  })
+
+  it('should be disabled when disabled prop is true', () => {
+    render(<NextStepButton disabled onClick={defaultProps.onClick} />)
+    expect(screen.getByRole('button')).toBeDisabled()
+  })
+
+  it('should not call onClick when disabled', () => {
+    render(<NextStepButton disabled onClick={defaultProps.onClick} />)
+    fireEvent.click(screen.getByRole('button'))
+    expect(defaultProps.onClick).not.toHaveBeenCalled()
+  })
+
+  it('should render arrow icon', () => {
+    const { container } = render(<NextStepButton {...defaultProps} />)
+    const svg = container.querySelector('svg')
+    expect(svg).toBeInTheDocument()
+  })
+})

web/app/components/datasets/create/step-one/components/preview-panel.spec.tsx

@@ -0,0 +1,119 @@
+import type { NotionPage } from '@/models/common'
+import type { CrawlResultItem } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+
+// Mock child components - paths must match source file's imports (relative to source)
+vi.mock('../../file-preview', () => ({
+  default: ({ file, hidePreview }: { file: { name: string }, hidePreview: () => void }) => (
+    <div data-testid="file-preview">
+      <span>{file.name}</span>
+      <button data-testid="close-file" onClick={hidePreview}>close-file</button>
+    </div>
+  ),
+}))
+
+vi.mock('../../notion-page-preview', () => ({
+  default: ({ currentPage, hidePreview }: { currentPage: { page_name: string }, hidePreview: () => void }) => (
+    <div data-testid="notion-preview">
+      <span>{currentPage.page_name}</span>
+      <button data-testid="close-notion" onClick={hidePreview}>close-notion</button>
+    </div>
+  ),
+}))
+
+vi.mock('../../website/preview', () => ({
+  default: ({ payload, hidePreview }: { payload: { title: string }, hidePreview: () => void }) => (
+    <div data-testid="website-preview">
+      <span>{payload.title}</span>
+      <button data-testid="close-website" onClick={hidePreview}>close-website</button>
+    </div>
+  ),
+}))
+
+vi.mock('@/app/components/billing/plan-upgrade-modal', () => ({
+  default: ({ show, onClose, title }: { show: boolean, onClose: () => void, title: string }) => show
+    ? (
+        <div data-testid="plan-upgrade-modal">
+          <span>{title}</span>
+          <button data-testid="close-modal" onClick={onClose}>close-modal</button>
+        </div>
+      )
+    : null,
+}))
+
+const { default: PreviewPanel } = await import('./preview-panel')
+
+describe('PreviewPanel', () => {
+  const defaultProps = {
+    currentFile: undefined,
+    currentNotionPage: undefined,
+    currentWebsite: undefined,
+    notionCredentialId: 'cred-1',
+    isShowPlanUpgradeModal: false,
+    hideFilePreview: vi.fn(),
+    hideNotionPagePreview: vi.fn(),
+    hideWebsitePreview: vi.fn(),
+    hidePlanUpgradeModal: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('rendering', () => {
+    it('should render nothing when no preview is active', () => {
+      const { container } = render(<PreviewPanel {...defaultProps} />)
+      expect(container.querySelector('[data-testid]')).toBeNull()
+    })
+
+    it('should render file preview when currentFile is set', () => {
+      render(<PreviewPanel {...defaultProps} currentFile={{ name: 'test.pdf' } as unknown as File} />)
+      expect(screen.getByTestId('file-preview')).toBeInTheDocument()
+      expect(screen.getByText('test.pdf')).toBeInTheDocument()
+    })
+
+    it('should render notion preview when currentNotionPage is set', () => {
+      render(<PreviewPanel {...defaultProps} currentNotionPage={{ page_name: 'My Page' } as unknown as NotionPage} />)
+      expect(screen.getByTestId('notion-preview')).toBeInTheDocument()
+      expect(screen.getByText('My Page')).toBeInTheDocument()
+    })
+
+    it('should render website preview when currentWebsite is set', () => {
+      render(<PreviewPanel {...defaultProps} currentWebsite={{ title: 'My Site' } as unknown as CrawlResultItem} />)
+      expect(screen.getByTestId('website-preview')).toBeInTheDocument()
+      expect(screen.getByText('My Site')).toBeInTheDocument()
+    })
+
+    it('should render plan upgrade modal when isShowPlanUpgradeModal is true', () => {
+      render(<PreviewPanel {...defaultProps} isShowPlanUpgradeModal />)
+      expect(screen.getByTestId('plan-upgrade-modal')).toBeInTheDocument()
+    })
+  })
+
+  describe('interactions', () => {
+    it('should call hideFilePreview when file preview close clicked', () => {
+      render(<PreviewPanel {...defaultProps} currentFile={{ name: 'test.pdf' } as unknown as File} />)
+      fireEvent.click(screen.getByTestId('close-file'))
+      expect(defaultProps.hideFilePreview).toHaveBeenCalledOnce()
+    })
+
+    it('should call hidePlanUpgradeModal when modal close clicked', () => {
+      render(<PreviewPanel {...defaultProps} isShowPlanUpgradeModal />)
+      fireEvent.click(screen.getByTestId('close-modal'))
+      expect(defaultProps.hidePlanUpgradeModal).toHaveBeenCalledOnce()
+    })
+
+    it('should call hideNotionPagePreview when notion preview close clicked', () => {
+      render(<PreviewPanel {...defaultProps} currentNotionPage={{ page_name: 'My Page' } as unknown as NotionPage} />)
+      fireEvent.click(screen.getByTestId('close-notion'))
+      expect(defaultProps.hideNotionPagePreview).toHaveBeenCalledOnce()
+    })
+
+    it('should call hideWebsitePreview when website preview close clicked', () => {
+      render(<PreviewPanel {...defaultProps} currentWebsite={{ title: 'My Site' } as unknown as CrawlResultItem} />)
+      fireEvent.click(screen.getByTestId('close-website'))
+      expect(defaultProps.hideWebsitePreview).toHaveBeenCalledOnce()
+    })
+  })
+})

web/app/components/datasets/create/step-one/hooks/use-preview-state.spec.ts

@@ -0,0 +1,60 @@
+import type { NotionPage } from '@/models/common'
+import type { CrawlResultItem } from '@/models/datasets'
+import { act, renderHook } from '@testing-library/react'
+import { describe, expect, it } from 'vitest'
+import usePreviewState from './use-preview-state'
+
+describe('usePreviewState', () => {
+  it('should initialize with all previews undefined', () => {
+    const { result } = renderHook(() => usePreviewState())
+
+    expect(result.current.currentFile).toBeUndefined()
+    expect(result.current.currentNotionPage).toBeUndefined()
+    expect(result.current.currentWebsite).toBeUndefined()
+  })
+
+  it('should show and hide file preview', () => {
+    const { result } = renderHook(() => usePreviewState())
+    const file = new File(['content'], 'test.pdf')
+
+    act(() => {
+      result.current.showFilePreview(file)
+    })
+    expect(result.current.currentFile).toBe(file)
+
+    act(() => {
+      result.current.hideFilePreview()
+    })
+    expect(result.current.currentFile).toBeUndefined()
+  })
+
+  it('should show and hide notion page preview', () => {
+    const { result } = renderHook(() => usePreviewState())
+    const page = { page_id: 'p1', page_name: 'Test' } as unknown as NotionPage
+
+    act(() => {
+      result.current.showNotionPagePreview(page)
+    })
+    expect(result.current.currentNotionPage).toBe(page)
+
+    act(() => {
+      result.current.hideNotionPagePreview()
+    })
+    expect(result.current.currentNotionPage).toBeUndefined()
+  })
+
+  it('should show and hide website preview', () => {
+    const { result } = renderHook(() => usePreviewState())
+    const website = { title: 'Example', source_url: 'https://example.com' } as unknown as CrawlResultItem
+
+    act(() => {
+      result.current.showWebsitePreview(website)
+    })
+    expect(result.current.currentWebsite).toBe(website)
+
+    act(() => {
+      result.current.hideWebsitePreview()
+    })
+    expect(result.current.currentWebsite).toBeUndefined()
+  })
+})

web/app/components/datasets/create/step-one/index.spec.tsx

@@ -1,11 +1,9 @@
 import type { DataSourceAuth } from '@/app/components/header/account-setting/data-source-page-new/types'
 import type { NotionPage } from '@/models/common'
 import type { CrawlOptions, CrawlResultItem, DataSet, FileItem } from '@/models/datasets'
-import { act, fireEvent, render, renderHook, screen } from '@testing-library/react'
+import { fireEvent, render, screen } from '@testing-library/react'
 import { Plan } from '@/app/components/billing/type'
 import { DataSourceType } from '@/models/datasets'
-import { DataSourceTypeSelector, NextStepButton, PreviewPanel } from './components'
-import { usePreviewState } from './hooks'
 import StepOne from './index'
 
 // ==========================================
@@ -211,541 +209,14 @@ const defaultProps = {
 }
 
 // ==========================================
-// usePreviewState Hook Tests
+// NOTE: Child component unit tests (usePreviewState, DataSourceTypeSelector,
+// NextStepButton, PreviewPanel) have been moved to their own dedicated spec files:
+//   - ./hooks/use-preview-state.spec.ts
+//   - ./components/data-source-type-selector.spec.tsx
+//   - ./components/next-step-button.spec.tsx
+//   - ./components/preview-panel.spec.tsx
+// This file now focuses exclusively on StepOne parent component tests.
 // ==========================================
-describe('usePreviewState Hook', () => {
-  // --------------------------------------------------------------------------
-  // Initial State Tests
-  // --------------------------------------------------------------------------
-  describe('Initial State', () => {
-    it('should initialize with all preview states undefined', () => {
-      // Arrange & Act
-      const { result } = renderHook(() => usePreviewState())
-
-      // Assert
-      expect(result.current.currentFile).toBeUndefined()
-      expect(result.current.currentNotionPage).toBeUndefined()
-      expect(result.current.currentWebsite).toBeUndefined()
-    })
-  })
-
-  // --------------------------------------------------------------------------
-  // File Preview Tests
-  // --------------------------------------------------------------------------
-  describe('File Preview', () => {
-    it('should show file preview when showFilePreview is called', () => {
-      // Arrange
-      const { result } = renderHook(() => usePreviewState())
-      const mockFile = new File(['test'], 'test.txt')
-
-      // Act
-      act(() => {
-        result.current.showFilePreview(mockFile)
-      })
-
-      // Assert
-      expect(result.current.currentFile).toBe(mockFile)
-    })
-
-    it('should hide file preview when hideFilePreview is called', () => {
-      // Arrange
-      const { result } = renderHook(() => usePreviewState())
-      const mockFile = new File(['test'], 'test.txt')
-
-      act(() => {
-        result.current.showFilePreview(mockFile)
-      })
-
-      // Act
-      act(() => {
-        result.current.hideFilePreview()
-      })
-
-      // Assert
-      expect(result.current.currentFile).toBeUndefined()
-    })
-  })
-
-  // --------------------------------------------------------------------------
-  // Notion Page Preview Tests
-  // --------------------------------------------------------------------------
-  describe('Notion Page Preview', () => {
-    it('should show notion page preview when showNotionPagePreview is called', () => {
-      // Arrange
-      const { result } = renderHook(() => usePreviewState())
-      const mockPage = createMockNotionPage()
-
-      // Act
-      act(() => {
-        result.current.showNotionPagePreview(mockPage)
-      })
-
-      // Assert
-      expect(result.current.currentNotionPage).toBe(mockPage)
-    })
-
-    it('should hide notion page preview when hideNotionPagePreview is called', () => {
-      // Arrange
-      const { result } = renderHook(() => usePreviewState())
-      const mockPage = createMockNotionPage()
-
-      act(() => {
-        result.current.showNotionPagePreview(mockPage)
-      })
-
-      // Act
-      act(() => {
-        result.current.hideNotionPagePreview()
-      })
-
-      // Assert
-      expect(result.current.currentNotionPage).toBeUndefined()
-    })
-  })
-
-  // --------------------------------------------------------------------------
-  // Website Preview Tests
-  // --------------------------------------------------------------------------
-  describe('Website Preview', () => {
-    it('should show website preview when showWebsitePreview is called', () => {
-      // Arrange
-      const { result } = renderHook(() => usePreviewState())
-      const mockWebsite = createMockCrawlResult()
-
-      // Act
-      act(() => {
-        result.current.showWebsitePreview(mockWebsite)
-      })
-
-      // Assert
-      expect(result.current.currentWebsite).toBe(mockWebsite)
-    })
-
-    it('should hide website preview when hideWebsitePreview is called', () => {
-      // Arrange
-      const { result } = renderHook(() => usePreviewState())
-      const mockWebsite = createMockCrawlResult()
-
-      act(() => {
-        result.current.showWebsitePreview(mockWebsite)
-      })
-
-      // Act
-      act(() => {
-        result.current.hideWebsitePreview()
-      })
-
-      // Assert
-      expect(result.current.currentWebsite).toBeUndefined()
-    })
-  })
-
-  // --------------------------------------------------------------------------
-  // Callback Stability Tests (Memoization)
-  // --------------------------------------------------------------------------
-  describe('Callback Stability', () => {
-    it('should maintain stable showFilePreview callback reference', () => {
-      // Arrange
-      const { result, rerender } = renderHook(() => usePreviewState())
-      const initialCallback = result.current.showFilePreview
-
-      // Act
-      rerender()
-
-      // Assert
-      expect(result.current.showFilePreview).toBe(initialCallback)
-    })
-
-    it('should maintain stable hideFilePreview callback reference', () => {
-      // Arrange
-      const { result, rerender } = renderHook(() => usePreviewState())
-      const initialCallback = result.current.hideFilePreview
-
-      // Act
-      rerender()
-
-      // Assert
-      expect(result.current.hideFilePreview).toBe(initialCallback)
-    })
-
-    it('should maintain stable showNotionPagePreview callback reference', () => {
-      // Arrange
-      const { result, rerender } = renderHook(() => usePreviewState())
-      const initialCallback = result.current.showNotionPagePreview
-
-      // Act
-      rerender()
-
-      // Assert
-      expect(result.current.showNotionPagePreview).toBe(initialCallback)
-    })
-
-    it('should maintain stable hideNotionPagePreview callback reference', () => {
-      // Arrange
-      const { result, rerender } = renderHook(() => usePreviewState())
-      const initialCallback = result.current.hideNotionPagePreview
-
-      // Act
-      rerender()
-
-      // Assert
-      expect(result.current.hideNotionPagePreview).toBe(initialCallback)
-    })
-
-    it('should maintain stable showWebsitePreview callback reference', () => {
-      // Arrange
-      const { result, rerender } = renderHook(() => usePreviewState())
-      const initialCallback = result.current.showWebsitePreview
-
-      // Act
-      rerender()
-
-      // Assert
-      expect(result.current.showWebsitePreview).toBe(initialCallback)
-    })
-
-    it('should maintain stable hideWebsitePreview callback reference', () => {
-      // Arrange
-      const { result, rerender } = renderHook(() => usePreviewState())
-      const initialCallback = result.current.hideWebsitePreview
-
-      // Act
-      rerender()
-
-      // Assert
-      expect(result.current.hideWebsitePreview).toBe(initialCallback)
-    })
-  })
-})
-
-// ==========================================
-// DataSourceTypeSelector Component Tests
-// ==========================================
-describe('DataSourceTypeSelector', () => {
-  const defaultSelectorProps = {
-    currentType: DataSourceType.FILE,
-    disabled: false,
-    onChange: vi.fn(),
-    onClearPreviews: vi.fn(),
-  }
-
-  beforeEach(() => {
-    vi.clearAllMocks()
-  })
-
-  // --------------------------------------------------------------------------
-  // Rendering Tests
-  // --------------------------------------------------------------------------
-  describe('Rendering', () => {
-    it('should render all data source options when web is enabled', () => {
-      // Arrange & Act
-      render(<DataSourceTypeSelector {...defaultSelectorProps} />)
-
-      // Assert
-      expect(screen.getByText('datasetCreation.stepOne.dataSourceType.file')).toBeInTheDocument()
-      expect(screen.getByText('datasetCreation.stepOne.dataSourceType.notion')).toBeInTheDocument()
-      expect(screen.getByText('datasetCreation.stepOne.dataSourceType.web')).toBeInTheDocument()
-    })
-
-    it('should highlight active type', () => {
-      // Arrange & Act
-      const { container } = render(
-        <DataSourceTypeSelector {...defaultSelectorProps} currentType={DataSourceType.NOTION} />,
-      )
-
-      // Assert - The active item should have the active class
-      const items = container.querySelectorAll('[class*="dataSourceItem"]')
-      expect(items.length).toBeGreaterThan(0)
-    })
-  })
-
-  // --------------------------------------------------------------------------
-  // User Interactions Tests
-  // --------------------------------------------------------------------------
-  describe('User Interactions', () => {
-    it('should call onChange when a type is clicked', () => {
-      // Arrange
-      const onChange = vi.fn()
-      render(<DataSourceTypeSelector {...defaultSelectorProps} onChange={onChange} />)
-
-      // Act
-      fireEvent.click(screen.getByText('datasetCreation.stepOne.dataSourceType.notion'))
-
-      // Assert
-      expect(onChange).toHaveBeenCalledWith(DataSourceType.NOTION)
-    })
-
-    it('should call onClearPreviews when a type is clicked', () => {
-      // Arrange
-      const onClearPreviews = vi.fn()
-      render(<DataSourceTypeSelector {...defaultSelectorProps} onClearPreviews={onClearPreviews} />)
-
-      // Act
-      fireEvent.click(screen.getByText('datasetCreation.stepOne.dataSourceType.web'))
-
-      // Assert
-      expect(onClearPreviews).toHaveBeenCalledWith(DataSourceType.WEB)
-    })
-
-    it('should not call onChange when disabled', () => {
-      // Arrange
-      const onChange = vi.fn()
-      render(<DataSourceTypeSelector {...defaultSelectorProps} disabled onChange={onChange} />)
-
-      // Act
-      fireEvent.click(screen.getByText('datasetCreation.stepOne.dataSourceType.notion'))
-
-      // Assert
-      expect(onChange).not.toHaveBeenCalled()
-    })
-
-    it('should not call onClearPreviews when disabled', () => {
-      // Arrange
-      const onClearPreviews = vi.fn()
-      render(<DataSourceTypeSelector {...defaultSelectorProps} disabled onClearPreviews={onClearPreviews} />)
-
-      // Act
-      fireEvent.click(screen.getByText('datasetCreation.stepOne.dataSourceType.notion'))
-
-      // Assert
-      expect(onClearPreviews).not.toHaveBeenCalled()
-    })
-  })
-})
-
-// ==========================================
-// NextStepButton Component Tests
-// ==========================================
-describe('NextStepButton', () => {
-  beforeEach(() => {
-    vi.clearAllMocks()
-  })
-
-  // --------------------------------------------------------------------------
-  // Rendering Tests
-  // --------------------------------------------------------------------------
-  describe('Rendering', () => {
-    it('should render with correct label', () => {
-      // Arrange & Act
-      render(<NextStepButton disabled={false} onClick={vi.fn()} />)
-
-      // Assert
-      expect(screen.getByText('datasetCreation.stepOne.button')).toBeInTheDocument()
-    })
-
-    it('should render with arrow icon', () => {
-      // Arrange & Act
-      const { container } = render(<NextStepButton disabled={false} onClick={vi.fn()} />)
-
-      // Assert
-      const svgIcon = container.querySelector('svg')
-      expect(svgIcon).toBeInTheDocument()
-    })
-  })
-
-  // --------------------------------------------------------------------------
-  // Props Tests
-  // --------------------------------------------------------------------------
-  describe('Props', () => {
-    it('should be disabled when disabled prop is true', () => {
-      // Arrange & Act
-      render(<NextStepButton disabled onClick={vi.fn()} />)
-
-      // Assert
-      expect(screen.getByRole('button')).toBeDisabled()
-    })
-
-    it('should be enabled when disabled prop is false', () => {
-      // Arrange & Act
-      render(<NextStepButton disabled={false} onClick={vi.fn()} />)
-
-      // Assert
-      expect(screen.getByRole('button')).not.toBeDisabled()
-    })
-
-    it('should call onClick when clicked and not disabled', () => {
-      // Arrange
-      const onClick = vi.fn()
-      render(<NextStepButton disabled={false} onClick={onClick} />)
-
-      // Act
-      fireEvent.click(screen.getByRole('button'))
-
-      // Assert
-      expect(onClick).toHaveBeenCalledTimes(1)
-    })
-
-    it('should not call onClick when clicked and disabled', () => {
-      // Arrange
-      const onClick = vi.fn()
-      render(<NextStepButton disabled onClick={onClick} />)
-
-      // Act
-      fireEvent.click(screen.getByRole('button'))
-
-      // Assert
-      expect(onClick).not.toHaveBeenCalled()
-    })
-  })
-})
-
-// ==========================================
-// PreviewPanel Component Tests
-// ==========================================
-describe('PreviewPanel', () => {
-  const defaultPreviewProps = {
-    currentFile: undefined as File | undefined,
-    currentNotionPage: undefined as NotionPage | undefined,
-    currentWebsite: undefined as CrawlResultItem | undefined,
-    notionCredentialId: 'cred-1',
-    isShowPlanUpgradeModal: false,
-    hideFilePreview: vi.fn(),
-    hideNotionPagePreview: vi.fn(),
-    hideWebsitePreview: vi.fn(),
-    hidePlanUpgradeModal: vi.fn(),
-  }
-
-  beforeEach(() => {
-    vi.clearAllMocks()
-  })
-
-  // --------------------------------------------------------------------------
-  // Conditional Rendering Tests
-  // --------------------------------------------------------------------------
-  describe('Conditional Rendering', () => {
-    it('should not render FilePreview when currentFile is undefined', () => {
-      // Arrange & Act
-      render(<PreviewPanel {...defaultPreviewProps} />)
-
-      // Assert
-      expect(screen.queryByTestId('file-preview')).not.toBeInTheDocument()
-    })
-
-    it('should render FilePreview when currentFile is defined', () => {
-      // Arrange
-      const file = new File(['test'], 'test.txt')
-
-      // Act
-      render(<PreviewPanel {...defaultPreviewProps} currentFile={file} />)
-
-      // Assert
-      expect(screen.getByTestId('file-preview')).toBeInTheDocument()
-    })
-
-    it('should not render NotionPagePreview when currentNotionPage is undefined', () => {
-      // Arrange & Act
-      render(<PreviewPanel {...defaultPreviewProps} />)
-
-      // Assert
-      expect(screen.queryByTestId('notion-page-preview')).not.toBeInTheDocument()
-    })
-
-    it('should render NotionPagePreview when currentNotionPage is defined', () => {
-      // Arrange
-      const page = createMockNotionPage()
-
-      // Act
-      render(<PreviewPanel {...defaultPreviewProps} currentNotionPage={page} />)
-
-      // Assert
-      expect(screen.getByTestId('notion-page-preview')).toBeInTheDocument()
-    })
-
-    it('should not render WebsitePreview when currentWebsite is undefined', () => {
-      // Arrange & Act
-      render(<PreviewPanel {...defaultPreviewProps} />)
-
-      // Assert - pagePreview is the title shown in WebsitePreview
-      expect(screen.queryByText('datasetCreation.stepOne.pagePreview')).not.toBeInTheDocument()
-    })
-
-    it('should render WebsitePreview when currentWebsite is defined', () => {
-      // Arrange
-      const website = createMockCrawlResult()
-
-      // Act
-      render(<PreviewPanel {...defaultPreviewProps} currentWebsite={website} />)
-
-      // Assert - Check for the preview title and source URL
-      expect(screen.getByText('datasetCreation.stepOne.pagePreview')).toBeInTheDocument()
-      expect(screen.getByText(website.source_url)).toBeInTheDocument()
-    })
-
-    it('should not render PlanUpgradeModal when isShowPlanUpgradeModal is false', () => {
-      // Arrange & Act
-      render(<PreviewPanel {...defaultPreviewProps} isShowPlanUpgradeModal={false} />)
-
-      // Assert
-      expect(screen.queryByTestId('plan-upgrade-modal')).not.toBeInTheDocument()
-    })
-
-    it('should render PlanUpgradeModal when isShowPlanUpgradeModal is true', () => {
-      // Arrange & Act
-      render(<PreviewPanel {...defaultPreviewProps} isShowPlanUpgradeModal />)
-
-      // Assert
-      expect(screen.getByTestId('plan-upgrade-modal')).toBeInTheDocument()
-    })
-  })
-
-  // --------------------------------------------------------------------------
-  // Event Handler Tests
-  // --------------------------------------------------------------------------
-  describe('Event Handlers', () => {
-    it('should call hideFilePreview when file preview close is clicked', () => {
-      // Arrange
-      const hideFilePreview = vi.fn()
-      const file = new File(['test'], 'test.txt')
-      render(<PreviewPanel {...defaultPreviewProps} currentFile={file} hideFilePreview={hideFilePreview} />)
-
-      // Act
-      fireEvent.click(screen.getByTestId('hide-file-preview'))
-
-      // Assert
-      expect(hideFilePreview).toHaveBeenCalledTimes(1)
-    })
-
-    it('should call hideNotionPagePreview when notion preview close is clicked', () => {
-      // Arrange
-      const hideNotionPagePreview = vi.fn()
-      const page = createMockNotionPage()
-      render(<PreviewPanel {...defaultPreviewProps} currentNotionPage={page} hideNotionPagePreview={hideNotionPagePreview} />)
-
-      // Act
-      fireEvent.click(screen.getByTestId('hide-notion-preview'))
-
-      // Assert
-      expect(hideNotionPagePreview).toHaveBeenCalledTimes(1)
-    })
-
-    it('should call hideWebsitePreview when website preview close is clicked', () => {
-      // Arrange
-      const hideWebsitePreview = vi.fn()
-      const website = createMockCrawlResult()
-      const { container } = render(<PreviewPanel {...defaultPreviewProps} currentWebsite={website} hideWebsitePreview={hideWebsitePreview} />)
-
-      // Act - Find the close button (div with cursor-pointer class containing the XMarkIcon)
-      const closeButton = container.querySelector('.cursor-pointer')
-      expect(closeButton).toBeInTheDocument()
-      fireEvent.click(closeButton!)
-
-      // Assert
-      expect(hideWebsitePreview).toHaveBeenCalledTimes(1)
-    })
-
-    it('should call hidePlanUpgradeModal when modal close is clicked', () => {
-      // Arrange
-      const hidePlanUpgradeModal = vi.fn()
-      render(<PreviewPanel {...defaultPreviewProps} isShowPlanUpgradeModal hidePlanUpgradeModal={hidePlanUpgradeModal} />)
-
-      // Act
-      fireEvent.click(screen.getByTestId('close-upgrade-modal'))
-
-      // Assert
-      expect(hidePlanUpgradeModal).toHaveBeenCalledTimes(1)
-    })
-  })
-})
 
 // ==========================================
 // StepOne Component Tests

web/app/components/datasets/create/step-one/upgrade-card.spec.tsx

@@ -0,0 +1,107 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import UpgradeCard from './upgrade-card'
+
+const mockSetShowPricingModal = vi.fn()
+
+vi.mock('@/context/modal-context', () => ({
+  useModalContext: () => ({
+    setShowPricingModal: mockSetShowPricingModal,
+  }),
+}))
+
+vi.mock('@/app/components/billing/upgrade-btn', () => ({
+  default: ({ onClick, className }: { onClick?: () => void, className?: string }) => (
+    <button type="button" className={className} onClick={onClick} data-testid="upgrade-btn">
+      upgrade
+    </button>
+  ),
+}))
+
+describe('UpgradeCard', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Rendering', () => {
+    it('should render without crashing', () => {
+      // Arrange & Act
+      render(<UpgradeCard />)
+
+      // Assert - title and description i18n keys are rendered
+      expect(screen.getByText(/uploadMultipleFiles\.title/i)).toBeInTheDocument()
+    })
+
+    it('should render the upgrade title text', () => {
+      // Arrange & Act
+      render(<UpgradeCard />)
+
+      // Assert
+      expect(screen.getByText(/uploadMultipleFiles\.title/i)).toBeInTheDocument()
+    })
+
+    it('should render the upgrade description text', () => {
+      // Arrange & Act
+      render(<UpgradeCard />)
+
+      // Assert
+      expect(screen.getByText(/uploadMultipleFiles\.description/i)).toBeInTheDocument()
+    })
+
+    it('should render the upgrade button', () => {
+      // Arrange & Act
+      render(<UpgradeCard />)
+
+      // Assert
+      expect(screen.getByRole('button')).toBeInTheDocument()
+    })
+  })
+
+  describe('User Interactions', () => {
+    it('should call setShowPricingModal when upgrade button is clicked', () => {
+      // Arrange
+      render(<UpgradeCard />)
+
+      // Act
+      fireEvent.click(screen.getByRole('button'))
+
+      // Assert
+      expect(mockSetShowPricingModal).toHaveBeenCalledTimes(1)
+    })
+
+    it('should not call setShowPricingModal without user interaction', () => {
+      // Arrange & Act
+      render(<UpgradeCard />)
+
+      // Assert
+      expect(mockSetShowPricingModal).not.toHaveBeenCalled()
+    })
+
+    it('should call setShowPricingModal on each button click', () => {
+      // Arrange
+      render(<UpgradeCard />)
+
+      // Act
+      const button = screen.getByRole('button')
+      fireEvent.click(button)
+      fireEvent.click(button)
+
+      // Assert
+      expect(mockSetShowPricingModal).toHaveBeenCalledTimes(2)
+    })
+  })
+
+  describe('Memoization', () => {
+    it('should maintain rendering after rerender with same props', () => {
+      // Arrange
+      const { rerender } = render(<UpgradeCard />)
+
+      // Act
+      rerender(<UpgradeCard />)
+
+      // Assert
+      expect(screen.getByText(/uploadMultipleFiles\.title/i)).toBeInTheDocument()
+      expect(screen.getByRole('button')).toBeInTheDocument()
+    })
+  })
+})

web/app/components/datasets/create/step-two/components/general-chunking-options.spec.tsx

@@ -0,0 +1,174 @@
+import type { PreProcessingRule } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { ChunkingMode } from '@/models/datasets'
+import { GeneralChunkingOptions } from './general-chunking-options'
+
+vi.mock('next/image', () => ({
+  default: ({ alt, ...props }: { alt?: string, src?: string, width?: number, height?: number }) => (
+    <img alt={alt} {...props} />
+  ),
+}))
+
+vi.mock('@/app/components/datasets/settings/summary-index-setting', () => ({
+  default: ({ onSummaryIndexSettingChange }: { onSummaryIndexSettingChange?: (val: Record<string, unknown>) => void }) => (
+    <div data-testid="summary-index-setting">
+      <button data-testid="summary-toggle" onClick={() => onSummaryIndexSettingChange?.({ enable: true })}>Toggle</button>
+    </div>
+  ),
+}))
+
+vi.mock('@/config', () => ({
+  IS_CE_EDITION: true,
+}))
+
+const ns = 'datasetCreation'
+
+const createRules = (): PreProcessingRule[] => [
+  { id: 'remove_extra_spaces', enabled: true },
+  { id: 'remove_urls_emails', enabled: false },
+]
+
+const defaultProps = {
+  segmentIdentifier: '\\n',
+  maxChunkLength: 500,
+  overlap: 50,
+  rules: createRules(),
+  currentDocForm: ChunkingMode.text,
+  docLanguage: 'English',
+  isActive: true,
+  isInUpload: false,
+  isNotUploadInEmptyDataset: false,
+  hasCurrentDatasetDocForm: false,
+  onSegmentIdentifierChange: vi.fn(),
+  onMaxChunkLengthChange: vi.fn(),
+  onOverlapChange: vi.fn(),
+  onRuleToggle: vi.fn(),
+  onDocFormChange: vi.fn(),
+  onDocLanguageChange: vi.fn(),
+  onPreview: vi.fn(),
+  onReset: vi.fn(),
+  locale: 'en',
+}
+
+describe('GeneralChunkingOptions', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Rendering', () => {
+    it('should render general chunking title', () => {
+      render(<GeneralChunkingOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.general`)).toBeInTheDocument()
+    })
+
+    it('should render delimiter, max length and overlap inputs when active', () => {
+      render(<GeneralChunkingOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.separator`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.maxLength`)).toBeInTheDocument()
+      expect(screen.getAllByText(new RegExp(`${ns}.stepTwo.overlap`)).length).toBeGreaterThan(0)
+    })
+
+    it('should render preprocessing rules as checkboxes', () => {
+      render(<GeneralChunkingOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.removeExtraSpaces`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.removeUrlEmails`)).toBeInTheDocument()
+    })
+
+    it('should render preview and reset buttons when active', () => {
+      render(<GeneralChunkingOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.previewChunk`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.reset`)).toBeInTheDocument()
+    })
+
+    it('should not render body when not active', () => {
+      render(<GeneralChunkingOptions {...defaultProps} isActive={false} />)
+      expect(screen.queryByText(`${ns}.stepTwo.separator`)).not.toBeInTheDocument()
+    })
+  })
+
+  describe('User Interactions', () => {
+    it('should call onPreview when preview button clicked', () => {
+      const onPreview = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} onPreview={onPreview} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.previewChunk`))
+      expect(onPreview).toHaveBeenCalledOnce()
+    })
+
+    it('should call onReset when reset button clicked', () => {
+      const onReset = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} onReset={onReset} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.reset`))
+      expect(onReset).toHaveBeenCalledOnce()
+    })
+
+    it('should call onRuleToggle when rule clicked', () => {
+      const onRuleToggle = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} onRuleToggle={onRuleToggle} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.removeUrlEmails`))
+      expect(onRuleToggle).toHaveBeenCalledWith('remove_urls_emails')
+    })
+
+    it('should call onDocFormChange with text mode when card switched', () => {
+      const onDocFormChange = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} isActive={false} onDocFormChange={onDocFormChange} />)
+      // OptionCard fires onSwitched which calls onDocFormChange(ChunkingMode.text)
+      // Since isActive=false, clicking the card triggers the switch
+      const titleEl = screen.getByText(`${ns}.stepTwo.general`)
+      fireEvent.click(titleEl.closest('[class*="rounded-xl"]')!)
+      expect(onDocFormChange).toHaveBeenCalledWith(ChunkingMode.text)
+    })
+  })
+
+  describe('QA Mode (CE Edition)', () => {
+    it('should render QA language checkbox', () => {
+      render(<GeneralChunkingOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.useQALanguage`)).toBeInTheDocument()
+    })
+
+    it('should toggle QA mode when checkbox clicked', () => {
+      const onDocFormChange = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} onDocFormChange={onDocFormChange} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.useQALanguage`))
+      expect(onDocFormChange).toHaveBeenCalledWith(ChunkingMode.qa)
+    })
+
+    it('should toggle back to text mode from QA mode', () => {
+      const onDocFormChange = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} currentDocForm={ChunkingMode.qa} onDocFormChange={onDocFormChange} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.useQALanguage`))
+      expect(onDocFormChange).toHaveBeenCalledWith(ChunkingMode.text)
+    })
+
+    it('should not toggle QA mode when hasCurrentDatasetDocForm is true', () => {
+      const onDocFormChange = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} hasCurrentDatasetDocForm onDocFormChange={onDocFormChange} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.useQALanguage`))
+      expect(onDocFormChange).not.toHaveBeenCalled()
+    })
+
+    it('should show QA warning tip when in QA mode', () => {
+      render(<GeneralChunkingOptions {...defaultProps} currentDocForm={ChunkingMode.qa} />)
+      expect(screen.getAllByText(`${ns}.stepTwo.QATip`).length).toBeGreaterThan(0)
+    })
+  })
+
+  describe('Summary Index Setting', () => {
+    it('should render SummaryIndexSetting when showSummaryIndexSetting is true', () => {
+      render(<GeneralChunkingOptions {...defaultProps} showSummaryIndexSetting />)
+      expect(screen.getByTestId('summary-index-setting')).toBeInTheDocument()
+    })
+
+    it('should not render SummaryIndexSetting when showSummaryIndexSetting is false', () => {
+      render(<GeneralChunkingOptions {...defaultProps} showSummaryIndexSetting={false} />)
+      expect(screen.queryByTestId('summary-index-setting')).not.toBeInTheDocument()
+    })
+
+    it('should call onSummaryIndexSettingChange', () => {
+      const onSummaryIndexSettingChange = vi.fn()
+      render(<GeneralChunkingOptions {...defaultProps} showSummaryIndexSetting onSummaryIndexSettingChange={onSummaryIndexSettingChange} />)
+      fireEvent.click(screen.getByTestId('summary-toggle'))
+      expect(onSummaryIndexSettingChange).toHaveBeenCalledWith({ enable: true })
+    })
+  })
+})

web/app/components/datasets/create/step-two/components/indexing-mode-section.spec.tsx

@@ -0,0 +1,219 @@
+import type { DefaultModel } from '@/app/components/header/account-setting/model-provider-page/declarations'
+import type { RetrievalConfig } from '@/types/app'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { ChunkingMode } from '@/models/datasets'
+import { IndexingType } from '../hooks'
+import { IndexingModeSection } from './indexing-mode-section'
+
+vi.mock('next/image', () => ({
+  default: ({ alt, ...props }: { alt?: string, src?: string, width?: number, height?: number }) => (
+    <img alt={alt} {...props} />
+  ),
+}))
+
+vi.mock('next/link', () => ({
+  default: ({ children, href, ...props }: { children?: React.ReactNode, href?: string, className?: string }) => <a href={href} {...props}>{children}</a>,
+}))
+
+// Mock external domain components
+vi.mock('@/app/components/datasets/common/retrieval-method-config', () => ({
+  default: ({ onChange, disabled }: { value?: RetrievalConfig, onChange?: (val: Record<string, unknown>) => void, disabled?: boolean }) => (
+    <div data-testid="retrieval-method-config" data-disabled={disabled}>
+      <button onClick={() => onChange?.({ search_method: 'updated' })}>Change Retrieval</button>
+    </div>
+  ),
+}))
+
+vi.mock('@/app/components/datasets/common/economical-retrieval-method-config', () => ({
+  default: ({ disabled }: { value?: RetrievalConfig, onChange?: (val: Record<string, unknown>) => void, disabled?: boolean }) => (
+    <div data-testid="economical-retrieval-config" data-disabled={disabled}>
+      Economical Config
+    </div>
+  ),
+}))
+
+vi.mock('@/app/components/header/account-setting/model-provider-page/model-selector', () => ({
+  default: ({ onSelect, readonly }: { onSelect?: (val: Record<string, string>) => void, readonly?: boolean }) => (
+    <div data-testid="model-selector" data-readonly={readonly}>
+      <button onClick={() => onSelect?.({ provider: 'openai', model: 'text-embedding-3-small' })}>Select Model</button>
+    </div>
+  ),
+}))
+
+vi.mock('@/context/i18n', () => ({
+  useDocLink: () => (path: string) => `https://docs.dify.ai${path}`,
+}))
+
+const ns = 'datasetCreation'
+
+const createDefaultModel = (overrides?: Partial<DefaultModel>): DefaultModel => ({
+  provider: 'openai',
+  model: 'text-embedding-ada-002',
+  ...overrides,
+})
+
+const createRetrievalConfig = (): RetrievalConfig => ({
+  search_method: 'semantic_search' as RetrievalConfig['search_method'],
+  reranking_enable: false,
+  reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+  top_k: 3,
+  score_threshold_enabled: false,
+  score_threshold: 0,
+})
+
+const defaultProps = {
+  indexType: IndexingType.QUALIFIED,
+  hasSetIndexType: false,
+  docForm: ChunkingMode.text,
+  embeddingModel: createDefaultModel(),
+  embeddingModelList: [],
+  retrievalConfig: createRetrievalConfig(),
+  showMultiModalTip: false,
+  isModelAndRetrievalConfigDisabled: false,
+  isQAConfirmDialogOpen: false,
+  onIndexTypeChange: vi.fn(),
+  onEmbeddingModelChange: vi.fn(),
+  onRetrievalConfigChange: vi.fn(),
+  onQAConfirmDialogClose: vi.fn(),
+  onQAConfirmDialogConfirm: vi.fn(),
+}
+
+describe('IndexingModeSection', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Rendering', () => {
+    it('should render index mode title', () => {
+      render(<IndexingModeSection {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.indexMode`)).toBeInTheDocument()
+    })
+
+    it('should render qualified option when not locked to economical', () => {
+      render(<IndexingModeSection {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.qualified`)).toBeInTheDocument()
+    })
+
+    it('should render economical option when not locked to qualified', () => {
+      render(<IndexingModeSection {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.economical`)).toBeInTheDocument()
+    })
+
+    it('should only show qualified option when hasSetIndexType and type is qualified', () => {
+      render(<IndexingModeSection {...defaultProps} hasSetIndexType indexType={IndexingType.QUALIFIED} />)
+      expect(screen.getByText(`${ns}.stepTwo.qualified`)).toBeInTheDocument()
+      expect(screen.queryByText(`${ns}.stepTwo.economical`)).not.toBeInTheDocument()
+    })
+
+    it('should only show economical option when hasSetIndexType and type is economical', () => {
+      render(<IndexingModeSection {...defaultProps} hasSetIndexType indexType={IndexingType.ECONOMICAL} />)
+      expect(screen.getByText(`${ns}.stepTwo.economical`)).toBeInTheDocument()
+      expect(screen.queryByText(`${ns}.stepTwo.qualified`)).not.toBeInTheDocument()
+    })
+  })
+
+  describe('Embedding Model', () => {
+    it('should show model selector when indexType is qualified', () => {
+      render(<IndexingModeSection {...defaultProps} indexType={IndexingType.QUALIFIED} />)
+      expect(screen.getByTestId('model-selector')).toBeInTheDocument()
+    })
+
+    it('should not show model selector when indexType is economical', () => {
+      render(<IndexingModeSection {...defaultProps} indexType={IndexingType.ECONOMICAL} />)
+      expect(screen.queryByTestId('model-selector')).not.toBeInTheDocument()
+    })
+
+    it('should mark model selector as readonly when disabled', () => {
+      render(<IndexingModeSection {...defaultProps} isModelAndRetrievalConfigDisabled />)
+      expect(screen.getByTestId('model-selector')).toHaveAttribute('data-readonly', 'true')
+    })
+
+    it('should call onEmbeddingModelChange when model selected', () => {
+      const onEmbeddingModelChange = vi.fn()
+      render(<IndexingModeSection {...defaultProps} onEmbeddingModelChange={onEmbeddingModelChange} />)
+      fireEvent.click(screen.getByText('Select Model'))
+      expect(onEmbeddingModelChange).toHaveBeenCalledWith({ provider: 'openai', model: 'text-embedding-3-small' })
+    })
+  })
+
+  describe('Retrieval Config', () => {
+    it('should show RetrievalMethodConfig when qualified', () => {
+      render(<IndexingModeSection {...defaultProps} indexType={IndexingType.QUALIFIED} />)
+      expect(screen.getByTestId('retrieval-method-config')).toBeInTheDocument()
+    })
+
+    it('should show EconomicalRetrievalMethodConfig when economical', () => {
+      render(<IndexingModeSection {...defaultProps} indexType={IndexingType.ECONOMICAL} />)
+      expect(screen.getByTestId('economical-retrieval-config')).toBeInTheDocument()
+    })
+
+    it('should call onRetrievalConfigChange from qualified config', () => {
+      const onRetrievalConfigChange = vi.fn()
+      render(<IndexingModeSection {...defaultProps} onRetrievalConfigChange={onRetrievalConfigChange} />)
+      fireEvent.click(screen.getByText('Change Retrieval'))
+      expect(onRetrievalConfigChange).toHaveBeenCalledWith({ search_method: 'updated' })
+    })
+  })
+
+  describe('Index Type Switching', () => {
+    it('should call onIndexTypeChange when switching to qualified', () => {
+      const onIndexTypeChange = vi.fn()
+      render(<IndexingModeSection {...defaultProps} indexType={IndexingType.ECONOMICAL} onIndexTypeChange={onIndexTypeChange} />)
+      const qualifiedCard = screen.getByText(`${ns}.stepTwo.qualified`).closest('[class*="rounded-xl"]')!
+      fireEvent.click(qualifiedCard)
+      expect(onIndexTypeChange).toHaveBeenCalledWith(IndexingType.QUALIFIED)
+    })
+
+    it('should disable economical when docForm is QA', () => {
+      render(<IndexingModeSection {...defaultProps} docForm={ChunkingMode.qa} />)
+      // The economical option card should have disabled styling
+      const economicalText = screen.getByText(`${ns}.stepTwo.economical`)
+      const card = economicalText.closest('[class*="rounded-xl"]')
+      expect(card).toHaveClass('pointer-events-none')
+    })
+  })
+
+  describe('High Quality Tip', () => {
+    it('should show high quality tip when qualified is selected and not locked', () => {
+      render(<IndexingModeSection {...defaultProps} indexType={IndexingType.QUALIFIED} hasSetIndexType={false} />)
+      expect(screen.getByText(`${ns}.stepTwo.highQualityTip`)).toBeInTheDocument()
+    })
+
+    it('should not show high quality tip when index type is locked', () => {
+      render(<IndexingModeSection {...defaultProps} indexType={IndexingType.QUALIFIED} hasSetIndexType />)
+      expect(screen.queryByText(`${ns}.stepTwo.highQualityTip`)).not.toBeInTheDocument()
+    })
+  })
+
+  describe('QA Confirm Dialog', () => {
+    it('should call onQAConfirmDialogClose when cancel clicked', () => {
+      const onClose = vi.fn()
+      render(<IndexingModeSection {...defaultProps} isQAConfirmDialogOpen onQAConfirmDialogClose={onClose} />)
+      const cancelBtns = screen.getAllByText(`${ns}.stepTwo.cancel`)
+      fireEvent.click(cancelBtns[0])
+      expect(onClose).toHaveBeenCalled()
+    })
+
+    it('should call onQAConfirmDialogConfirm when confirm clicked', () => {
+      const onConfirm = vi.fn()
+      render(<IndexingModeSection {...defaultProps} isQAConfirmDialogOpen onQAConfirmDialogConfirm={onConfirm} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.switch`))
+      expect(onConfirm).toHaveBeenCalled()
+    })
+  })
+
+  describe('Dataset Settings Link', () => {
+    it('should show settings link when economical and hasSetIndexType', () => {
+      render(<IndexingModeSection {...defaultProps} hasSetIndexType indexType={IndexingType.ECONOMICAL} datasetId="ds-123" />)
+      expect(screen.getByText(`${ns}.stepTwo.datasetSettingLink`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.datasetSettingLink`).closest('a')).toHaveAttribute('href', '/datasets/ds-123/settings')
+    })
+
+    it('should show settings link under model selector when disabled', () => {
+      render(<IndexingModeSection {...defaultProps} isModelAndRetrievalConfigDisabled datasetId="ds-456" />)
+      const links = screen.getAllByText(`${ns}.stepTwo.datasetSettingLink`)
+      expect(links.length).toBeGreaterThan(0)
+    })
+  })
+})

web/app/components/datasets/create/step-two/components/inputs.spec.tsx

@@ -0,0 +1,92 @@
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DelimiterInput, MaxLengthInput, OverlapInput } from './inputs'
+
+// i18n mock returns namespaced keys like "datasetCreation.stepTwo.separator"
+const ns = 'datasetCreation'
+
+describe('DelimiterInput', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render separator label', () => {
+    render(<DelimiterInput />)
+    expect(screen.getByText(`${ns}.stepTwo.separator`)).toBeInTheDocument()
+  })
+
+  it('should render text input with placeholder', () => {
+    render(<DelimiterInput />)
+    const input = screen.getByPlaceholderText(`${ns}.stepTwo.separatorPlaceholder`)
+    expect(input).toBeInTheDocument()
+    expect(input).toHaveAttribute('type', 'text')
+  })
+
+  it('should pass through value and onChange props', () => {
+    const onChange = vi.fn()
+    render(<DelimiterInput value="test-val" onChange={onChange} />)
+    expect(screen.getByDisplayValue('test-val')).toBeInTheDocument()
+  })
+
+  it('should render tooltip content', () => {
+    render(<DelimiterInput />)
+    // Tooltip triggers render; component mounts without error
+    expect(screen.getByText(`${ns}.stepTwo.separator`)).toBeInTheDocument()
+  })
+})
+
+describe('MaxLengthInput', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render max length label', () => {
+    render(<MaxLengthInput onChange={vi.fn()} />)
+    expect(screen.getByText(`${ns}.stepTwo.maxLength`)).toBeInTheDocument()
+  })
+
+  it('should render number input', () => {
+    render(<MaxLengthInput onChange={vi.fn()} />)
+    const input = screen.getByRole('spinbutton')
+    expect(input).toBeInTheDocument()
+  })
+
+  it('should accept value prop', () => {
+    render(<MaxLengthInput value={500} onChange={vi.fn()} />)
+    expect(screen.getByDisplayValue('500')).toBeInTheDocument()
+  })
+
+  it('should have min of 1', () => {
+    render(<MaxLengthInput onChange={vi.fn()} />)
+    const input = screen.getByRole('spinbutton')
+    expect(input).toHaveAttribute('min', '1')
+  })
+})
+
+describe('OverlapInput', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render overlap label', () => {
+    render(<OverlapInput onChange={vi.fn()} />)
+    expect(screen.getAllByText(new RegExp(`${ns}.stepTwo.overlap`)).length).toBeGreaterThan(0)
+  })
+
+  it('should render number input', () => {
+    render(<OverlapInput onChange={vi.fn()} />)
+    const input = screen.getByRole('spinbutton')
+    expect(input).toBeInTheDocument()
+  })
+
+  it('should accept value prop', () => {
+    render(<OverlapInput value={50} onChange={vi.fn()} />)
+    expect(screen.getByDisplayValue('50')).toBeInTheDocument()
+  })
+
+  it('should have min of 1', () => {
+    render(<OverlapInput onChange={vi.fn()} />)
+    const input = screen.getByRole('spinbutton')
+    expect(input).toHaveAttribute('min', '1')
+  })
+})

web/app/components/datasets/create/step-two/components/option-card.spec.tsx

@@ -0,0 +1,159 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { OptionCard, OptionCardHeader } from './option-card'
+
+vi.mock('next/image', () => ({
+  default: ({ src, alt, ...props }: { src?: string, alt?: string, width?: number, height?: number }) => (
+    <img src={src} alt={alt} {...props} />
+  ),
+}))
+
+describe('OptionCardHeader', () => {
+  const defaultProps = {
+    icon: <span data-testid="icon">icon</span>,
+    title: <span>Test Title</span>,
+    description: 'Test description',
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render icon, title and description', () => {
+    render(<OptionCardHeader {...defaultProps} />)
+    expect(screen.getByTestId('icon')).toBeInTheDocument()
+    expect(screen.getByText('Test Title')).toBeInTheDocument()
+    expect(screen.getByText('Test description')).toBeInTheDocument()
+  })
+
+  it('should show effect image when active and effectImg provided', () => {
+    const { container } = render(
+      <OptionCardHeader {...defaultProps} isActive effectImg="/effect.png" />,
+    )
+    const img = container.querySelector('img')
+    expect(img).toBeInTheDocument()
+  })
+
+  it('should not show effect image when not active', () => {
+    const { container } = render(
+      <OptionCardHeader {...defaultProps} isActive={false} effectImg="/effect.png" />,
+    )
+    expect(container.querySelector('img')).not.toBeInTheDocument()
+  })
+
+  it('should apply cursor-pointer when not disabled', () => {
+    const { container } = render(<OptionCardHeader {...defaultProps} />)
+    expect(container.firstChild).toHaveClass('cursor-pointer')
+  })
+
+  it('should not apply cursor-pointer when disabled', () => {
+    const { container } = render(<OptionCardHeader {...defaultProps} disabled />)
+    expect(container.firstChild).not.toHaveClass('cursor-pointer')
+  })
+
+  it('should apply activeClassName when active', () => {
+    const { container } = render(
+      <OptionCardHeader {...defaultProps} isActive activeClassName="custom-active" />,
+    )
+    expect(container.firstChild).toHaveClass('custom-active')
+  })
+
+  it('should not apply activeClassName when not active', () => {
+    const { container } = render(
+      <OptionCardHeader {...defaultProps} isActive={false} activeClassName="custom-active" />,
+    )
+    expect(container.firstChild).not.toHaveClass('custom-active')
+  })
+})
+
+describe('OptionCard', () => {
+  const defaultProps = {
+    icon: <span data-testid="icon">icon</span>,
+    title: <span>Card Title</span> as React.ReactNode,
+    description: 'Card description',
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render header content', () => {
+    render(<OptionCard {...defaultProps} />)
+    expect(screen.getByText('Card Title')).toBeInTheDocument()
+    expect(screen.getByText('Card description')).toBeInTheDocument()
+  })
+
+  it('should call onSwitched when clicked while not active and not disabled', () => {
+    const onSwitched = vi.fn()
+    const { container } = render(
+      <OptionCard {...defaultProps} isActive={false} onSwitched={onSwitched} />,
+    )
+    fireEvent.click(container.firstChild!)
+    expect(onSwitched).toHaveBeenCalledOnce()
+  })
+
+  it('should not call onSwitched when already active', () => {
+    const onSwitched = vi.fn()
+    const { container } = render(
+      <OptionCard {...defaultProps} isActive onSwitched={onSwitched} />,
+    )
+    fireEvent.click(container.firstChild!)
+    expect(onSwitched).not.toHaveBeenCalled()
+  })
+
+  it('should not call onSwitched when disabled', () => {
+    const onSwitched = vi.fn()
+    const { container } = render(
+      <OptionCard {...defaultProps} disabled onSwitched={onSwitched} />,
+    )
+    fireEvent.click(container.firstChild!)
+    expect(onSwitched).not.toHaveBeenCalled()
+  })
+
+  it('should show children and actions when active', () => {
+    render(
+      <OptionCard {...defaultProps} isActive actions={<button>Action</button>}>
+        <div>Body Content</div>
+      </OptionCard>,
+    )
+    expect(screen.getByText('Body Content')).toBeInTheDocument()
+    expect(screen.getByText('Action')).toBeInTheDocument()
+  })
+
+  it('should not show children when not active', () => {
+    render(
+      <OptionCard {...defaultProps} isActive={false}>
+        <div>Body Content</div>
+      </OptionCard>,
+    )
+    expect(screen.queryByText('Body Content')).not.toBeInTheDocument()
+  })
+
+  it('should apply selected border style when active and not noHighlight', () => {
+    const { container } = render(<OptionCard {...defaultProps} isActive />)
+    expect(container.firstChild).toHaveClass('border-components-option-card-option-selected-border')
+  })
+
+  it('should not apply selected border when noHighlight is true', () => {
+    const { container } = render(<OptionCard {...defaultProps} isActive noHighlight />)
+    expect(container.firstChild).not.toHaveClass('border-components-option-card-option-selected-border')
+  })
+
+  it('should apply disabled opacity and pointer-events styles', () => {
+    const { container } = render(<OptionCard {...defaultProps} disabled />)
+    expect(container.firstChild).toHaveClass('pointer-events-none')
+    expect(container.firstChild).toHaveClass('opacity-50')
+  })
+
+  it('should forward custom className', () => {
+    const { container } = render(<OptionCard {...defaultProps} className="custom-class" />)
+    expect(container.firstChild).toHaveClass('custom-class')
+  })
+
+  it('should forward custom style', () => {
+    const { container } = render(
+      <OptionCard {...defaultProps} style={{ maxWidth: '300px' }} />,
+    )
+    expect((container.firstChild as HTMLElement).style.maxWidth).toBe('300px')
+  })
+})

web/app/components/datasets/create/step-two/components/parent-child-options.spec.tsx

@@ -0,0 +1,156 @@
+import type { ParentChildConfig } from '../hooks'
+import type { PreProcessingRule } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { ChunkingMode } from '@/models/datasets'
+import { ParentChildOptions } from './parent-child-options'
+
+vi.mock('next/image', () => ({
+  default: ({ alt, ...props }: { alt?: string, src?: string, width?: number, height?: number }) => (
+    <img alt={alt} {...props} />
+  ),
+}))
+
+vi.mock('@/app/components/datasets/settings/summary-index-setting', () => ({
+  default: ({ onSummaryIndexSettingChange }: { onSummaryIndexSettingChange?: (val: Record<string, unknown>) => void }) => (
+    <div data-testid="summary-index-setting">
+      <button data-testid="summary-toggle" onClick={() => onSummaryIndexSettingChange?.({ enable: true })}>Toggle</button>
+    </div>
+  ),
+}))
+
+vi.mock('@/config', () => ({
+  IS_CE_EDITION: true,
+}))
+
+const ns = 'datasetCreation'
+
+const createRules = (): PreProcessingRule[] => [
+  { id: 'remove_extra_spaces', enabled: true },
+  { id: 'remove_urls_emails', enabled: false },
+]
+
+const createParentChildConfig = (overrides?: Partial<ParentChildConfig>): ParentChildConfig => ({
+  chunkForContext: 'paragraph',
+  parent: { delimiter: '\\n\\n', maxLength: 2000 },
+  child: { delimiter: '\\n', maxLength: 500 },
+  ...overrides,
+})
+
+const defaultProps = {
+  parentChildConfig: createParentChildConfig(),
+  rules: createRules(),
+  currentDocForm: ChunkingMode.parentChild,
+  isActive: true,
+  isInUpload: false,
+  isNotUploadInEmptyDataset: false,
+  onDocFormChange: vi.fn(),
+  onChunkForContextChange: vi.fn(),
+  onParentDelimiterChange: vi.fn(),
+  onParentMaxLengthChange: vi.fn(),
+  onChildDelimiterChange: vi.fn(),
+  onChildMaxLengthChange: vi.fn(),
+  onRuleToggle: vi.fn(),
+  onPreview: vi.fn(),
+  onReset: vi.fn(),
+}
+
+describe('ParentChildOptions', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Rendering', () => {
+    it('should render parent-child title', () => {
+      render(<ParentChildOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.parentChild`)).toBeInTheDocument()
+    })
+
+    it('should render parent chunk context section when active', () => {
+      render(<ParentChildOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.parentChunkForContext`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.paragraph`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.fullDoc`)).toBeInTheDocument()
+    })
+
+    it('should render child chunk retrieval section when active', () => {
+      render(<ParentChildOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.childChunkForRetrieval`)).toBeInTheDocument()
+    })
+
+    it('should render rules section when active', () => {
+      render(<ParentChildOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.removeExtraSpaces`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.removeUrlEmails`)).toBeInTheDocument()
+    })
+
+    it('should render preview and reset buttons when active', () => {
+      render(<ParentChildOptions {...defaultProps} />)
+      expect(screen.getByText(`${ns}.stepTwo.previewChunk`)).toBeInTheDocument()
+      expect(screen.getByText(`${ns}.stepTwo.reset`)).toBeInTheDocument()
+    })
+
+    it('should not render body when not active', () => {
+      render(<ParentChildOptions {...defaultProps} isActive={false} />)
+      expect(screen.queryByText(`${ns}.stepTwo.parentChunkForContext`)).not.toBeInTheDocument()
+    })
+  })
+
+  describe('User Interactions', () => {
+    it('should call onPreview when preview button clicked', () => {
+      const onPreview = vi.fn()
+      render(<ParentChildOptions {...defaultProps} onPreview={onPreview} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.previewChunk`))
+      expect(onPreview).toHaveBeenCalledOnce()
+    })
+
+    it('should call onReset when reset button clicked', () => {
+      const onReset = vi.fn()
+      render(<ParentChildOptions {...defaultProps} onReset={onReset} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.reset`))
+      expect(onReset).toHaveBeenCalledOnce()
+    })
+
+    it('should call onRuleToggle when rule clicked', () => {
+      const onRuleToggle = vi.fn()
+      render(<ParentChildOptions {...defaultProps} onRuleToggle={onRuleToggle} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.removeUrlEmails`))
+      expect(onRuleToggle).toHaveBeenCalledWith('remove_urls_emails')
+    })
+
+    it('should call onDocFormChange with parentChild when card switched', () => {
+      const onDocFormChange = vi.fn()
+      render(<ParentChildOptions {...defaultProps} isActive={false} onDocFormChange={onDocFormChange} />)
+      const titleEl = screen.getByText(`${ns}.stepTwo.parentChild`)
+      fireEvent.click(titleEl.closest('[class*="rounded-xl"]')!)
+      expect(onDocFormChange).toHaveBeenCalledWith(ChunkingMode.parentChild)
+    })
+
+    it('should call onChunkForContextChange when full-doc chosen', () => {
+      const onChunkForContextChange = vi.fn()
+      render(<ParentChildOptions {...defaultProps} onChunkForContextChange={onChunkForContextChange} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.fullDoc`))
+      expect(onChunkForContextChange).toHaveBeenCalledWith('full-doc')
+    })
+
+    it('should call onChunkForContextChange when paragraph chosen', () => {
+      const onChunkForContextChange = vi.fn()
+      const config = createParentChildConfig({ chunkForContext: 'full-doc' })
+      render(<ParentChildOptions {...defaultProps} parentChildConfig={config} onChunkForContextChange={onChunkForContextChange} />)
+      fireEvent.click(screen.getByText(`${ns}.stepTwo.paragraph`))
+      expect(onChunkForContextChange).toHaveBeenCalledWith('paragraph')
+    })
+  })
+
+  describe('Summary Index Setting', () => {
+    it('should render SummaryIndexSetting when showSummaryIndexSetting is true', () => {
+      render(<ParentChildOptions {...defaultProps} showSummaryIndexSetting />)
+      expect(screen.getByTestId('summary-index-setting')).toBeInTheDocument()
+    })
+
+    it('should not render SummaryIndexSetting when showSummaryIndexSetting is false', () => {
+      render(<ParentChildOptions {...defaultProps} showSummaryIndexSetting={false} />)
+      expect(screen.queryByTestId('summary-index-setting')).not.toBeInTheDocument()
+    })
+  })
+})

web/app/components/datasets/create/step-two/components/preview-panel.spec.tsx

@@ -0,0 +1,176 @@
+import type { ParentChildConfig } from '../hooks'
+import type { FileIndexingEstimateResponse } from '@/models/datasets'
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { ChunkingMode, DataSourceType } from '@/models/datasets'
+import { PreviewPanel } from './preview-panel'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string, opts?: { count?: number }) => opts?.count !== undefined ? `${key}-${opts.count}` : key,
+  }),
+}))
+
+vi.mock('@remixicon/react', () => ({
+  RiSearchEyeLine: () => <span data-testid="search-icon" />,
+}))
+
+vi.mock('@/app/components/base/float-right-container', () => ({
+  default: ({ children }: { children: React.ReactNode }) => <div data-testid="float-container">{children}</div>,
+}))
+
+vi.mock('@/app/components/base/badge', () => ({
+  default: ({ text }: { text: string }) => <span data-testid="badge">{text}</span>,
+}))
+
+vi.mock('@/app/components/base/skeleton', () => ({
+  SkeletonContainer: ({ children }: { children: React.ReactNode }) => <div data-testid="skeleton">{children}</div>,
+  SkeletonPoint: () => <span />,
+  SkeletonRectangle: () => <span />,
+  SkeletonRow: ({ children }: { children: React.ReactNode }) => <div>{children}</div>,
+}))
+
+vi.mock('../../../chunk', () => ({
+  ChunkContainer: ({ children, label }: { children: React.ReactNode, label: string }) => (
+    <div data-testid="chunk-container">
+      {label}
+      :
+      {' '}
+      {children}
+    </div>
+  ),
+  QAPreview: ({ qa }: { qa: { question: string } }) => <div data-testid="qa-preview">{qa.question}</div>,
+}))
+
+vi.mock('../../../common/document-picker/preview-document-picker', () => ({
+  default: () => <div data-testid="doc-picker" />,
+}))
+
+vi.mock('../../../documents/detail/completed/common/summary-label', () => ({
+  default: ({ summary }: { summary: string }) => <span data-testid="summary">{summary}</span>,
+}))
+
+vi.mock('../../../formatted-text/flavours/preview-slice', () => ({
+  PreviewSlice: ({ label, text }: { label: string, text: string }) => (
+    <span data-testid="preview-slice">
+      {label}
+      :
+      {' '}
+      {text}
+    </span>
+  ),
+}))
+
+vi.mock('../../../formatted-text/formatted', () => ({
+  FormattedText: ({ children }: { children: React.ReactNode }) => <p data-testid="formatted-text">{children}</p>,
+}))
+
+vi.mock('../../../preview/container', () => ({
+  default: ({ children, header }: { children: React.ReactNode, header: React.ReactNode }) => (
+    <div data-testid="preview-container">
+      {header}
+      {children}
+    </div>
+  ),
+}))
+
+vi.mock('../../../preview/header', () => ({
+  PreviewHeader: ({ children, title }: { children: React.ReactNode, title: string }) => (
+    <div data-testid="preview-header">
+      {title}
+      {children}
+    </div>
+  ),
+}))
+
+vi.mock('@/config', () => ({
+  FULL_DOC_PREVIEW_LENGTH: 3,
+}))
+
+describe('PreviewPanel', () => {
+  const defaultProps = {
+    isMobile: false,
+    dataSourceType: DataSourceType.FILE,
+    currentDocForm: ChunkingMode.text,
+    parentChildConfig: { chunkForContext: 'paragraph' } as ParentChildConfig,
+    pickerFiles: [{ id: '1', name: 'file.pdf', extension: 'pdf' }],
+    pickerValue: { id: '1', name: 'file.pdf', extension: 'pdf' },
+    isIdle: false,
+    isPending: false,
+    onPickerChange: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render preview header with title', () => {
+    render(<PreviewPanel {...defaultProps} />)
+    expect(screen.getByTestId('preview-header')).toHaveTextContent('stepTwo.preview')
+  })
+
+  it('should render document picker', () => {
+    render(<PreviewPanel {...defaultProps} />)
+    expect(screen.getByTestId('doc-picker')).toBeInTheDocument()
+  })
+
+  it('should show idle state when isIdle is true', () => {
+    render(<PreviewPanel {...defaultProps} isIdle={true} />)
+    expect(screen.getByText('stepTwo.previewChunkTip')).toBeInTheDocument()
+  })
+
+  it('should show loading skeletons when isPending', () => {
+    render(<PreviewPanel {...defaultProps} isPending={true} />)
+    expect(screen.getAllByTestId('skeleton')).toHaveLength(10)
+  })
+
+  it('should render text preview chunks', () => {
+    const estimate: Partial<FileIndexingEstimateResponse> = {
+      total_segments: 2,
+      preview: [
+        { content: 'chunk 1 text', child_chunks: [], summary: '' },
+        { content: 'chunk 2 text', child_chunks: [], summary: 'summary text' },
+      ],
+    }
+    render(<PreviewPanel {...defaultProps} estimate={estimate as FileIndexingEstimateResponse} />)
+    expect(screen.getAllByTestId('chunk-container')).toHaveLength(2)
+  })
+
+  it('should render QA preview', () => {
+    const estimate: Partial<FileIndexingEstimateResponse> = {
+      qa_preview: [
+        { question: 'Q1', answer: 'A1' },
+      ],
+    }
+    render(
+      <PreviewPanel
+        {...defaultProps}
+        currentDocForm={ChunkingMode.qa}
+        estimate={estimate as FileIndexingEstimateResponse}
+      />,
+    )
+    expect(screen.getByTestId('qa-preview')).toHaveTextContent('Q1')
+  })
+
+  it('should render parent-child preview', () => {
+    const estimate: Partial<FileIndexingEstimateResponse> = {
+      preview: [
+        { content: 'parent chunk', child_chunks: ['child1', 'child2'], summary: '' },
+      ],
+    }
+    render(
+      <PreviewPanel
+        {...defaultProps}
+        currentDocForm={ChunkingMode.parentChild}
+        estimate={estimate as FileIndexingEstimateResponse}
+      />,
+    )
+    expect(screen.getAllByTestId('preview-slice')).toHaveLength(2)
+  })
+
+  it('should show badge with chunk count for non-QA mode', () => {
+    const estimate: Partial<FileIndexingEstimateResponse> = { total_segments: 5, preview: [] }
+    render(<PreviewPanel {...defaultProps} estimate={estimate as FileIndexingEstimateResponse} />)
+    expect(screen.getByTestId('badge')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/create/step-two/components/step-two-footer.spec.tsx

@@ -0,0 +1,53 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { StepTwoFooter } from './step-two-footer'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({ t: (key: string) => key }),
+}))
+vi.mock('@remixicon/react', () => ({
+  RiArrowLeftLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="arrow-left" {...props} />,
+}))
+
+describe('StepTwoFooter', () => {
+  const defaultProps = {
+    isCreating: false,
+    onPrevious: vi.fn(),
+    onCreate: vi.fn(),
+    onCancel: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render previous and next buttons when not isSetting', () => {
+    render(<StepTwoFooter {...defaultProps} />)
+    expect(screen.getByText('stepTwo.previousStep')).toBeInTheDocument()
+    expect(screen.getByText('stepTwo.nextStep')).toBeInTheDocument()
+  })
+
+  it('should render save and cancel buttons when isSetting', () => {
+    render(<StepTwoFooter {...defaultProps} isSetting />)
+    expect(screen.getByText('stepTwo.save')).toBeInTheDocument()
+    expect(screen.getByText('stepTwo.cancel')).toBeInTheDocument()
+  })
+
+  it('should call onPrevious on previous button click', () => {
+    render(<StepTwoFooter {...defaultProps} />)
+    fireEvent.click(screen.getByText('stepTwo.previousStep'))
+    expect(defaultProps.onPrevious).toHaveBeenCalledOnce()
+  })
+
+  it('should call onCreate on next button click', () => {
+    render(<StepTwoFooter {...defaultProps} />)
+    fireEvent.click(screen.getByText('stepTwo.nextStep'))
+    expect(defaultProps.onCreate).toHaveBeenCalledOnce()
+  })
+
+  it('should call onCancel on cancel button click in settings mode', () => {
+    render(<StepTwoFooter {...defaultProps} isSetting />)
+    fireEvent.click(screen.getByText('stepTwo.cancel'))
+    expect(defaultProps.onCancel).toHaveBeenCalledOnce()
+  })
+})

web/app/components/datasets/create/step-two/hooks/escape.spec.ts

@@ -0,0 +1,76 @@
+import { describe, expect, it } from 'vitest'
+import escape from './escape'
+
+describe('escape', () => {
+  // Basic special character escaping
+  it('should escape null character', () => {
+    expect(escape('\0')).toBe('\\0')
+  })
+
+  it('should escape backspace', () => {
+    expect(escape('\b')).toBe('\\b')
+  })
+
+  it('should escape form feed', () => {
+    expect(escape('\f')).toBe('\\f')
+  })
+
+  it('should escape newline', () => {
+    expect(escape('\n')).toBe('\\n')
+  })
+
+  it('should escape carriage return', () => {
+    expect(escape('\r')).toBe('\\r')
+  })
+
+  it('should escape tab', () => {
+    expect(escape('\t')).toBe('\\t')
+  })
+
+  it('should escape vertical tab', () => {
+    expect(escape('\v')).toBe('\\v')
+  })
+
+  it('should escape single quote', () => {
+    expect(escape('\'')).toBe('\\\'')
+  })
+
+  // Multiple special characters in one string
+  it('should escape multiple special characters', () => {
+    expect(escape('line1\nline2\ttab')).toBe('line1\\nline2\\ttab')
+  })
+
+  it('should escape mixed special characters', () => {
+    expect(escape('\n\r\t')).toBe('\\n\\r\\t')
+  })
+
+  // Edge cases
+  it('should return empty string for null input', () => {
+    expect(escape(null as unknown as string)).toBe('')
+  })
+
+  it('should return empty string for undefined input', () => {
+    expect(escape(undefined as unknown as string)).toBe('')
+  })
+
+  it('should return empty string for empty string input', () => {
+    expect(escape('')).toBe('')
+  })
+
+  it('should return empty string for non-string input', () => {
+    expect(escape(123 as unknown as string)).toBe('')
+  })
+
+  // Pass-through for normal strings
+  it('should leave normal text unchanged', () => {
+    expect(escape('hello world')).toBe('hello world')
+  })
+
+  it('should leave special regex characters unchanged', () => {
+    expect(escape('a.b*c+d')).toBe('a.b*c+d')
+  })
+
+  it('should handle strings with no special characters', () => {
+    expect(escape('abc123')).toBe('abc123')
+  })
+})

web/app/components/datasets/create/step-two/hooks/unescape.spec.ts

@@ -0,0 +1,97 @@
+import { describe, expect, it } from 'vitest'
+import unescape from './unescape'
+
+describe('unescape', () => {
+  // Basic escape sequences
+  it('should unescape \\n to newline', () => {
+    expect(unescape('\\n')).toBe('\n')
+  })
+
+  it('should unescape \\t to tab', () => {
+    expect(unescape('\\t')).toBe('\t')
+  })
+
+  it('should unescape \\r to carriage return', () => {
+    expect(unescape('\\r')).toBe('\r')
+  })
+
+  it('should unescape \\b to backspace', () => {
+    expect(unescape('\\b')).toBe('\b')
+  })
+
+  it('should unescape \\f to form feed', () => {
+    expect(unescape('\\f')).toBe('\f')
+  })
+
+  it('should unescape \\v to vertical tab', () => {
+    expect(unescape('\\v')).toBe('\v')
+  })
+
+  it('should unescape \\0 to null character', () => {
+    expect(unescape('\\0')).toBe('\0')
+  })
+
+  it('should unescape \\\\ to backslash', () => {
+    expect(unescape('\\\\')).toBe('\\')
+  })
+
+  it('should unescape \\\'  to single quote', () => {
+    expect(unescape('\\\'')).toBe('\'')
+  })
+
+  it('should unescape \\" to double quote', () => {
+    expect(unescape('\\"')).toBe('"')
+  })
+
+  // Hex escape sequences (\\xNN)
+  it('should unescape 2-digit hex sequences', () => {
+    expect(unescape('\\x41')).toBe('A')
+    expect(unescape('\\x61')).toBe('a')
+  })
+
+  // Unicode escape sequences (\\uNNNN)
+  it('should unescape 4-digit unicode sequences', () => {
+    expect(unescape('\\u0041')).toBe('A')
+    expect(unescape('\\u4e2d')).toBe('中')
+  })
+
+  // Variable-length unicode (\\u{NNNN})
+  it('should unescape variable-length unicode sequences', () => {
+    expect(unescape('\\u{41}')).toBe('A')
+    expect(unescape('\\u{1F600}')).toBe('😀')
+  })
+
+  // Octal escape sequences
+  it('should unescape octal sequences', () => {
+    expect(unescape('\\101')).toBe('A') // 0o101 = 65 = 'A'
+    expect(unescape('\\12')).toBe('\n') // 0o12 = 10 = '\n'
+  })
+
+  // Python-style 8-digit unicode (\\UNNNNNNNN)
+  it('should unescape Python-style 8-digit unicode', () => {
+    expect(unescape('\\U0001F3B5')).toBe('🎵')
+  })
+
+  // Multiple escape sequences
+  it('should unescape multiple sequences in one string', () => {
+    expect(unescape('line1\\nline2\\ttab')).toBe('line1\nline2\ttab')
+  })
+
+  // Mixed content
+  it('should leave non-escape content unchanged', () => {
+    expect(unescape('hello world')).toBe('hello world')
+  })
+
+  it('should handle mixed escaped and non-escaped content', () => {
+    expect(unescape('before\\nafter')).toBe('before\nafter')
+  })
+
+  // Edge cases
+  it('should handle empty string', () => {
+    expect(unescape('')).toBe('')
+  })
+
+  it('should handle string with no escape sequences', () => {
+    expect(unescape('abc123')).toBe('abc123')
+  })
+})

web/app/components/datasets/create/step-two/hooks/use-document-creation.spec.ts

@@ -0,0 +1,186 @@
+import type { CustomFile, FullDocumentDetail, ProcessRule } from '@/models/datasets'
+import type { RetrievalConfig } from '@/types/app'
+import { renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { ChunkingMode, DataSourceType } from '@/models/datasets'
+import { RETRIEVE_METHOD } from '@/types/app'
+
+// Hoisted mocks
+const mocks = vi.hoisted(() => ({
+  toastNotify: vi.fn(),
+  mutateAsync: vi.fn(),
+  isReRankModelSelected: vi.fn(() => true),
+  trackEvent: vi.fn(),
+  invalidDatasetList: vi.fn(),
+}))
+
+vi.mock('@/app/components/base/toast', () => ({
+  default: { notify: mocks.toastNotify },
+}))
+
+vi.mock('@/app/components/base/amplitude', () => ({
+  trackEvent: mocks.trackEvent,
+}))
+
+vi.mock('@/app/components/datasets/common/check-rerank-model', () => ({
+  isReRankModelSelected: mocks.isReRankModelSelected,
+}))
+
+vi.mock('@/service/knowledge/use-create-dataset', () => ({
+  useCreateFirstDocument: () => ({ mutateAsync: mocks.mutateAsync, isPending: false }),
+  useCreateDocument: () => ({ mutateAsync: mocks.mutateAsync, isPending: false }),
+  getNotionInfo: vi.fn(() => []),
+  getWebsiteInfo: vi.fn(() => ({})),
+}))
+
+vi.mock('@/service/knowledge/use-dataset', () => ({
+  useInvalidDatasetList: () => mocks.invalidDatasetList,
+}))
+
+const { useDocumentCreation } = await import('./use-document-creation')
+const { IndexingType } = await import('./use-indexing-config')
+
+describe('useDocumentCreation', () => {
+  const defaultOptions = {
+    dataSourceType: DataSourceType.FILE,
+    files: [{ id: 'f-1', name: 'test.txt' }] as CustomFile[],
+    notionPages: [],
+    notionCredentialId: '',
+    websitePages: [],
+  }
+
+  const defaultValidationParams = {
+    segmentationType: 'general',
+    maxChunkLength: 1024,
+    limitMaxChunkLength: 4000,
+    overlap: 50,
+    indexType: IndexingType.QUALIFIED,
+    embeddingModel: { provider: 'openai', model: 'text-embedding-3-small' },
+    rerankModelList: [],
+    retrievalConfig: {
+      search_method: RETRIEVE_METHOD.semantic,
+      reranking_enable: false,
+      reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+      top_k: 3,
+      score_threshold_enabled: false,
+      score_threshold: 0.5,
+    } as RetrievalConfig,
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mocks.isReRankModelSelected.mockReturnValue(true)
+  })
+
+  describe('validateParams', () => {
+    it('should return true for valid params', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      expect(result.current.validateParams(defaultValidationParams)).toBe(true)
+    })
+
+    it('should return false when overlap > maxChunkLength', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      const invalid = { ...defaultValidationParams, overlap: 2000, maxChunkLength: 1000 }
+      expect(result.current.validateParams(invalid)).toBe(false)
+      expect(mocks.toastNotify).toHaveBeenCalledWith(
+        expect.objectContaining({ type: 'error' }),
+      )
+    })
+
+    it('should return false when maxChunkLength > limitMaxChunkLength', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      const invalid = { ...defaultValidationParams, maxChunkLength: 5000, limitMaxChunkLength: 4000 }
+      expect(result.current.validateParams(invalid)).toBe(false)
+    })
+
+    it('should return false when qualified but no embedding model', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      const invalid = {
+        ...defaultValidationParams,
+        indexType: IndexingType.QUALIFIED,
+        embeddingModel: { provider: '', model: '' },
+      }
+      expect(result.current.validateParams(invalid)).toBe(false)
+    })
+
+    it('should return false when rerank model not selected', () => {
+      mocks.isReRankModelSelected.mockReturnValue(false)
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      expect(result.current.validateParams(defaultValidationParams)).toBe(false)
+    })
+
+    it('should skip embedding/rerank checks when isSetting is true', () => {
+      mocks.isReRankModelSelected.mockReturnValue(false)
+      const { result } = renderHook(() =>
+        useDocumentCreation({ ...defaultOptions, isSetting: true }),
+      )
+      const params = {
+        ...defaultValidationParams,
+        embeddingModel: { provider: '', model: '' },
+      }
+      expect(result.current.validateParams(params)).toBe(true)
+    })
+  })
+
+  describe('buildCreationParams', () => {
+    it('should build params for FILE data source', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      const processRule = { mode: 'custom', rules: {} } as unknown as ProcessRule
+      const retrievalConfig = defaultValidationParams.retrievalConfig
+      const embeddingModel = { provider: 'openai', model: 'text-embedding-3-small' }
+
+      const params = result.current.buildCreationParams(
+        ChunkingMode.text,
+        'English',
+        processRule,
+        retrievalConfig,
+        embeddingModel,
+        'high_quality',
+      )
+
+      expect(params).not.toBeNull()
+      expect(params!.data_source!.type).toBe(DataSourceType.FILE)
+      expect(params!.data_source!.info_list.file_info_list?.file_ids).toContain('f-1')
+      expect(params!.embedding_model).toBe('text-embedding-3-small')
+      expect(params!.embedding_model_provider).toBe('openai')
+    })
+
+    it('should build params for isSetting mode', () => {
+      const detail = { id: 'doc-1' } as FullDocumentDetail
+      const { result } = renderHook(() =>
+        useDocumentCreation({ ...defaultOptions, isSetting: true, documentDetail: detail }),
+      )
+      const params = result.current.buildCreationParams(
+        ChunkingMode.text,
+        'English',
+        { mode: 'custom', rules: {} } as unknown as ProcessRule,
+        defaultValidationParams.retrievalConfig,
+        { provider: 'openai', model: 'text-embedding-3-small' },
+        'high_quality',
+      )
+
+      expect(params!.original_document_id).toBe('doc-1')
+      expect(params!.data_source).toBeUndefined()
+    })
+  })
+
+  describe('validatePreviewParams', () => {
+    it('should return true when maxChunkLength is within limit', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      expect(result.current.validatePreviewParams(1024)).toBe(true)
+    })
+
+    it('should return false when maxChunkLength exceeds limit', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      expect(result.current.validatePreviewParams(999999)).toBe(false)
+      expect(mocks.toastNotify).toHaveBeenCalled()
+    })
+  })
+
+  describe('isCreating', () => {
+    it('should reflect mutation pending state', () => {
+      const { result } = renderHook(() => useDocumentCreation(defaultOptions))
+      expect(result.current.isCreating).toBe(false)
+    })
+  })
+})

web/app/components/datasets/create/step-two/hooks/use-indexing-config.spec.ts

@@ -0,0 +1,161 @@
+import { act, renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { RETRIEVE_METHOD } from '@/types/app'
+
+// Hoisted mock state
+const mocks = vi.hoisted(() => ({
+  rerankModelList: [] as Array<{ provider: { provider: string }, model: string }>,
+  rerankDefaultModel: null as { provider: { provider: string }, model: string } | null,
+  isRerankDefaultModelValid: null as { provider: { provider: string }, model: string } | null,
+  embeddingModelList: [] as Array<{ provider: { provider: string }, model: string }>,
+  defaultEmbeddingModel: null as { provider: { provider: string }, model: string } | null,
+}))
+
+vi.mock('@/app/components/header/account-setting/model-provider-page/hooks', () => ({
+  useModelListAndDefaultModelAndCurrentProviderAndModel: () => ({
+    modelList: mocks.rerankModelList,
+    defaultModel: mocks.rerankDefaultModel,
+    currentModel: mocks.isRerankDefaultModelValid,
+  }),
+  useModelList: () => ({ data: mocks.embeddingModelList }),
+  useDefaultModel: () => ({ data: mocks.defaultEmbeddingModel }),
+}))
+
+vi.mock('@/app/components/datasets/settings/utils', () => ({
+  checkShowMultiModalTip: vi.fn(() => false),
+}))
+
+const { IndexingType, useIndexingConfig } = await import('./use-indexing-config')
+
+describe('useIndexingConfig', () => {
+  const defaultOptions = {
+    isAPIKeySet: true,
+    hasSetIndexType: false,
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mocks.rerankModelList = []
+    mocks.rerankDefaultModel = null
+    mocks.isRerankDefaultModelValid = null
+    mocks.embeddingModelList = []
+    mocks.defaultEmbeddingModel = null
+  })
+
+  describe('initial state', () => {
+    it('should default to QUALIFIED when API key is set', () => {
+      const { result } = renderHook(() => useIndexingConfig(defaultOptions))
+      expect(result.current.indexType).toBe(IndexingType.QUALIFIED)
+    })
+
+    it('should default to ECONOMICAL when API key is not set', () => {
+      const { result } = renderHook(() =>
+        useIndexingConfig({ ...defaultOptions, isAPIKeySet: false }),
+      )
+      expect(result.current.indexType).toBe(IndexingType.ECONOMICAL)
+    })
+
+    it('should use initial index type when provided', () => {
+      const { result } = renderHook(() =>
+        useIndexingConfig({
+          ...defaultOptions,
+          initialIndexType: IndexingType.ECONOMICAL,
+        }),
+      )
+      expect(result.current.indexType).toBe(IndexingType.ECONOMICAL)
+    })
+
+    it('should use initial embedding model when provided', () => {
+      const { result } = renderHook(() =>
+        useIndexingConfig({
+          ...defaultOptions,
+          initialEmbeddingModel: { provider: 'openai', model: 'text-embedding-3-small' },
+        }),
+      )
+      expect(result.current.embeddingModel).toEqual({
+        provider: 'openai',
+        model: 'text-embedding-3-small',
+      })
+    })
+
+    it('should use initial retrieval config when provided', () => {
+      const config = {
+        search_method: RETRIEVE_METHOD.fullText,
+        reranking_enable: false,
+        reranking_model: { reranking_provider_name: '', reranking_model_name: '' },
+        top_k: 5,
+        score_threshold_enabled: true,
+        score_threshold: 0.8,
+      }
+      const { result } = renderHook(() =>
+        useIndexingConfig({ ...defaultOptions, initialRetrievalConfig: config }),
+      )
+      expect(result.current.retrievalConfig.search_method).toBe(RETRIEVE_METHOD.fullText)
+      expect(result.current.retrievalConfig.top_k).toBe(5)
+    })
+  })
+
+  describe('setters', () => {
+    it('should update index type', () => {
+      const { result } = renderHook(() => useIndexingConfig(defaultOptions))
+
+      act(() => {
+        result.current.setIndexType(IndexingType.ECONOMICAL)
+      })
+      expect(result.current.indexType).toBe(IndexingType.ECONOMICAL)
+    })
+
+    it('should update embedding model', () => {
+      const { result } = renderHook(() => useIndexingConfig(defaultOptions))
+
+      act(() => {
+        result.current.setEmbeddingModel({ provider: 'cohere', model: 'embed-v3' })
+      })
+      expect(result.current.embeddingModel).toEqual({ provider: 'cohere', model: 'embed-v3' })
+    })
+
+    it('should update retrieval config', () => {
+      const { result } = renderHook(() => useIndexingConfig(defaultOptions))
+      const newConfig = {
+        ...result.current.retrievalConfig,
+        top_k: 10,
+      }
+
+      act(() => {
+        result.current.setRetrievalConfig(newConfig)
+      })
+      expect(result.current.retrievalConfig.top_k).toBe(10)
+    })
+  })
+
+  describe('getIndexingTechnique', () => {
+    it('should return initialIndexType when provided', () => {
+      const { result } = renderHook(() =>
+        useIndexingConfig({
+          ...defaultOptions,
+          initialIndexType: IndexingType.ECONOMICAL,
+        }),
+      )
+      expect(result.current.getIndexingTechnique()).toBe(IndexingType.ECONOMICAL)
+    })
+
+    it('should return current indexType when no initialIndexType', () => {
+      const { result } = renderHook(() => useIndexingConfig(defaultOptions))
+      expect(result.current.getIndexingTechnique()).toBe(IndexingType.QUALIFIED)
+    })
+  })
+
+  describe('computed properties', () => {
+    it('should expose hasSetIndexType from options', () => {
+      const { result } = renderHook(() =>
+        useIndexingConfig({ ...defaultOptions, hasSetIndexType: true }),
+      )
+      expect(result.current.hasSetIndexType).toBe(true)
+    })
+
+    it('should expose showMultiModalTip as boolean', () => {
+      const { result } = renderHook(() => useIndexingConfig(defaultOptions))
+      expect(typeof result.current.showMultiModalTip).toBe('boolean')
+    })
+  })
+})

web/app/components/datasets/create/step-two/hooks/use-indexing-estimate.spec.ts

@@ -0,0 +1,127 @@
+import type { IndexingType } from './use-indexing-config'
+import type { NotionPage } from '@/models/common'
+import type { ChunkingMode, CrawlResultItem, CustomFile, ProcessRule } from '@/models/datasets'
+import { renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DataSourceType } from '@/models/datasets'
+
+// Hoisted mocks
+const mocks = vi.hoisted(() => ({
+  fileMutate: vi.fn(),
+  fileReset: vi.fn(),
+  notionMutate: vi.fn(),
+  notionReset: vi.fn(),
+  webMutate: vi.fn(),
+  webReset: vi.fn(),
+}))
+
+vi.mock('@/service/knowledge/use-create-dataset', () => ({
+  useFetchFileIndexingEstimateForFile: () => ({
+    mutate: mocks.fileMutate,
+    reset: mocks.fileReset,
+    data: { tokens: 100, total_segments: 5 },
+    isIdle: true,
+    isPending: false,
+  }),
+  useFetchFileIndexingEstimateForNotion: () => ({
+    mutate: mocks.notionMutate,
+    reset: mocks.notionReset,
+    data: null,
+    isIdle: true,
+    isPending: false,
+  }),
+  useFetchFileIndexingEstimateForWeb: () => ({
+    mutate: mocks.webMutate,
+    reset: mocks.webReset,
+    data: null,
+    isIdle: true,
+    isPending: false,
+  }),
+}))
+
+const { useIndexingEstimate } = await import('./use-indexing-estimate')
+
+describe('useIndexingEstimate', () => {
+  const defaultOptions = {
+    dataSourceType: DataSourceType.FILE,
+    currentDocForm: 'text_model' as ChunkingMode,
+    docLanguage: 'English',
+    files: [{ id: 'f-1', name: 'test.txt' }] as unknown as CustomFile[],
+    previewNotionPage: {} as unknown as NotionPage,
+    notionCredentialId: '',
+    previewWebsitePage: {} as unknown as CrawlResultItem,
+    indexingTechnique: 'high_quality' as unknown as IndexingType,
+    processRule: { mode: 'custom', rules: {} } as unknown as ProcessRule,
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('currentMutation selection', () => {
+    it('should select file mutation for FILE type', () => {
+      const { result } = renderHook(() => useIndexingEstimate(defaultOptions))
+      expect(result.current.estimate).toEqual({ tokens: 100, total_segments: 5 })
+    })
+
+    it('should select notion mutation for NOTION type', () => {
+      const { result } = renderHook(() => useIndexingEstimate({
+        ...defaultOptions,
+        dataSourceType: DataSourceType.NOTION,
+      }))
+      expect(result.current.estimate).toBeNull()
+    })
+
+    it('should select web mutation for WEB type', () => {
+      const { result } = renderHook(() => useIndexingEstimate({
+        ...defaultOptions,
+        dataSourceType: DataSourceType.WEB,
+      }))
+      expect(result.current.estimate).toBeNull()
+    })
+  })
+
+  describe('fetchEstimate', () => {
+    it('should call file mutate for FILE type', () => {
+      const { result } = renderHook(() => useIndexingEstimate(defaultOptions))
+      result.current.fetchEstimate()
+      expect(mocks.fileMutate).toHaveBeenCalledOnce()
+    })
+
+    it('should call notion mutate for NOTION type', () => {
+      const { result } = renderHook(() => useIndexingEstimate({
+        ...defaultOptions,
+        dataSourceType: DataSourceType.NOTION,
+      }))
+      result.current.fetchEstimate()
+      expect(mocks.notionMutate).toHaveBeenCalledOnce()
+    })
+
+    it('should call web mutate for WEB type', () => {
+      const { result } = renderHook(() => useIndexingEstimate({
+        ...defaultOptions,
+        dataSourceType: DataSourceType.WEB,
+      }))
+      result.current.fetchEstimate()
+      expect(mocks.webMutate).toHaveBeenCalledOnce()
+    })
+  })
+
+  describe('state properties', () => {
+    it('should expose isIdle', () => {
+      const { result } = renderHook(() => useIndexingEstimate(defaultOptions))
+      expect(result.current.isIdle).toBe(true)
+    })
+
+    it('should expose isPending', () => {
+      const { result } = renderHook(() => useIndexingEstimate(defaultOptions))
+      expect(result.current.isPending).toBe(false)
+    })
+
+    it('should expose reset function', () => {
+      const { result } = renderHook(() => useIndexingEstimate(defaultOptions))
+      result.current.reset()
+      expect(mocks.fileReset).toHaveBeenCalledOnce()
+    })
+  })
+})

web/app/components/datasets/create/step-two/hooks/use-preview-state.spec.ts

@@ -0,0 +1,198 @@
+import type { NotionPage } from '@/models/common'
+import type { CrawlResultItem, CustomFile } from '@/models/datasets'
+import { act, renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DataSourceType } from '@/models/datasets'
+import { usePreviewState } from './use-preview-state'
+
+// Factory functions
+const createFile = (id: string, name: string): CustomFile => ({
+  id,
+  name,
+  size: 1024,
+  type: 'text/plain',
+  extension: 'txt',
+  created_by: 'user',
+  created_at: Date.now(),
+} as unknown as CustomFile)
+
+const createNotionPage = (pageId: string, pageName: string): NotionPage => ({
+  page_id: pageId,
+  page_name: pageName,
+  page_icon: null,
+  parent_id: '',
+  type: 'page',
+  is_bound: true,
+} as unknown as NotionPage)
+
+const createWebsitePage = (url: string, title: string): CrawlResultItem => ({
+  source_url: url,
+  title,
+  markdown: '',
+  description: '',
+} as unknown as CrawlResultItem)
+
+describe('usePreviewState', () => {
+  const files = [createFile('f-1', 'file1.txt'), createFile('f-2', 'file2.txt')]
+  const notionPages = [createNotionPage('np-1', 'Page 1'), createNotionPage('np-2', 'Page 2')]
+  const websitePages = [createWebsitePage('https://a.com', 'Site A'), createWebsitePage('https://b.com', 'Site B')]
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('initial state for FILE', () => {
+    it('should set first file as preview', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.FILE,
+        files,
+        notionPages: [],
+        websitePages: [],
+      }))
+      expect(result.current.previewFile).toBe(files[0])
+    })
+  })
+
+  describe('initial state for NOTION', () => {
+    it('should set first notion page as preview', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.NOTION,
+        files: [],
+        notionPages,
+        websitePages: [],
+      }))
+      expect(result.current.previewNotionPage).toBe(notionPages[0])
+    })
+  })
+
+  describe('initial state for WEB', () => {
+    it('should set first website page as preview', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.WEB,
+        files: [],
+        notionPages: [],
+        websitePages,
+      }))
+      expect(result.current.previewWebsitePage).toBe(websitePages[0])
+    })
+  })
+
+  describe('getPreviewPickerItems', () => {
+    it('should return files for FILE type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.FILE,
+        files,
+        notionPages: [],
+        websitePages: [],
+      }))
+      const items = result.current.getPreviewPickerItems()
+      expect(items).toHaveLength(2)
+    })
+
+    it('should return mapped notion pages for NOTION type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.NOTION,
+        files: [],
+        notionPages,
+        websitePages: [],
+      }))
+      const items = result.current.getPreviewPickerItems()
+      expect(items).toHaveLength(2)
+      expect(items[0]).toEqual({ id: 'np-1', name: 'Page 1', extension: 'md' })
+    })
+
+    it('should return mapped website pages for WEB type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.WEB,
+        files: [],
+        notionPages: [],
+        websitePages,
+      }))
+      const items = result.current.getPreviewPickerItems()
+      expect(items).toHaveLength(2)
+      expect(items[0]).toEqual({ id: 'https://a.com', name: 'Site A', extension: 'md' })
+    })
+  })
+
+  describe('getPreviewPickerValue', () => {
+    it('should return current preview file for FILE type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.FILE,
+        files,
+        notionPages: [],
+        websitePages: [],
+      }))
+      const value = result.current.getPreviewPickerValue()
+      expect(value).toBe(files[0])
+    })
+
+    it('should return mapped notion page value for NOTION type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.NOTION,
+        files: [],
+        notionPages,
+        websitePages: [],
+      }))
+      const value = result.current.getPreviewPickerValue()
+      expect(value).toEqual({ id: 'np-1', name: 'Page 1', extension: 'md' })
+    })
+  })
+
+  describe('handlePreviewChange', () => {
+    it('should change preview file for FILE type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.FILE,
+        files,
+        notionPages: [],
+        websitePages: [],
+      }))
+
+      act(() => {
+        result.current.handlePreviewChange({ id: 'f-2', name: 'file2.txt' })
+      })
+      expect(result.current.previewFile).toEqual({ id: 'f-2', name: 'file2.txt' })
+    })
+
+    it('should change preview notion page for NOTION type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.NOTION,
+        files: [],
+        notionPages,
+        websitePages: [],
+      }))
+
+      act(() => {
+        result.current.handlePreviewChange({ id: 'np-2', name: 'Page 2' })
+      })
+      expect(result.current.previewNotionPage).toBe(notionPages[1])
+    })
+
+    it('should change preview website page for WEB type', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.WEB,
+        files: [],
+        notionPages: [],
+        websitePages,
+      }))
+
+      act(() => {
+        result.current.handlePreviewChange({ id: 'https://b.com', name: 'Site B' })
+      })
+      expect(result.current.previewWebsitePage).toBe(websitePages[1])
+    })
+
+    it('should not change if selected page not found (NOTION)', () => {
+      const { result } = renderHook(() => usePreviewState({
+        dataSourceType: DataSourceType.NOTION,
+        files: [],
+        notionPages,
+        websitePages: [],
+      }))
+
+      act(() => {
+        result.current.handlePreviewChange({ id: 'non-existent', name: 'x' })
+      })
+      expect(result.current.previewNotionPage).toBe(notionPages[0])
+    })
+  })
+})

web/app/components/datasets/create/step-two/hooks/use-segmentation-state.spec.ts

@@ -0,0 +1,373 @@
+import type { PreProcessingRule, Rules } from '@/models/datasets'
+import { act, renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { ChunkingMode, ProcessMode } from '@/models/datasets'
+import {
+  DEFAULT_MAXIMUM_CHUNK_LENGTH,
+  DEFAULT_OVERLAP,
+  DEFAULT_SEGMENT_IDENTIFIER,
+  defaultParentChildConfig,
+  useSegmentationState,
+} from './use-segmentation-state'
+
+describe('useSegmentationState', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // --- Default state ---
+  describe('default state', () => {
+    it('should initialize with default values', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      expect(result.current.segmentationType).toBe(ProcessMode.general)
+      expect(result.current.segmentIdentifier).toBe(DEFAULT_SEGMENT_IDENTIFIER)
+      expect(result.current.maxChunkLength).toBe(DEFAULT_MAXIMUM_CHUNK_LENGTH)
+      expect(result.current.overlap).toBe(DEFAULT_OVERLAP)
+      expect(result.current.rules).toEqual([])
+      expect(result.current.parentChildConfig).toEqual(defaultParentChildConfig)
+    })
+
+    it('should accept initial segmentation type', () => {
+      const { result } = renderHook(() =>
+        useSegmentationState({ initialSegmentationType: ProcessMode.parentChild }),
+      )
+      expect(result.current.segmentationType).toBe(ProcessMode.parentChild)
+    })
+
+    it('should accept initial summary index setting', () => {
+      const setting = { enable: true }
+      const { result } = renderHook(() =>
+        useSegmentationState({ initialSummaryIndexSetting: setting }),
+      )
+      expect(result.current.summaryIndexSetting).toEqual(setting)
+    })
+  })
+
+  // --- Setters ---
+  describe('setters', () => {
+    it('should update segmentation type', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setSegmentationType(ProcessMode.parentChild)
+      })
+      expect(result.current.segmentationType).toBe(ProcessMode.parentChild)
+    })
+
+    it('should update max chunk length', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setMaxChunkLength(2048)
+      })
+      expect(result.current.maxChunkLength).toBe(2048)
+    })
+
+    it('should update overlap', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setOverlap(100)
+      })
+      expect(result.current.overlap).toBe(100)
+    })
+
+    it('should update rules', () => {
+      const newRules: PreProcessingRule[] = [
+        { id: 'remove_extra_spaces', enabled: true },
+        { id: 'remove_urls_emails', enabled: false },
+      ]
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setRules(newRules)
+      })
+      expect(result.current.rules).toEqual(newRules)
+    })
+  })
+
+  // --- Segment identifier with escaping ---
+  describe('setSegmentIdentifier', () => {
+    it('should escape the value when setting', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setSegmentIdentifier('\n\n')
+      })
+      expect(result.current.segmentIdentifier).toBe('\\n\\n')
+    })
+
+    it('should reset to default when empty and canEmpty is false', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setSegmentIdentifier('')
+      })
+      expect(result.current.segmentIdentifier).toBe(DEFAULT_SEGMENT_IDENTIFIER)
+    })
+
+    it('should allow empty value when canEmpty is true', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setSegmentIdentifier('', true)
+      })
+      expect(result.current.segmentIdentifier).toBe('')
+    })
+  })
+
+  // --- Toggle rule ---
+  describe('toggleRule', () => {
+    it('should toggle a rule enabled state', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      const rules: PreProcessingRule[] = [
+        { id: 'remove_extra_spaces', enabled: true },
+        { id: 'remove_urls_emails', enabled: false },
+      ]
+
+      act(() => {
+        result.current.setRules(rules)
+      })
+      act(() => {
+        result.current.toggleRule('remove_extra_spaces')
+      })
+
+      expect(result.current.rules[0].enabled).toBe(false)
+      expect(result.current.rules[1].enabled).toBe(false)
+    })
+
+    it('should toggle second rule without affecting first', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      const rules: PreProcessingRule[] = [
+        { id: 'remove_extra_spaces', enabled: true },
+        { id: 'remove_urls_emails', enabled: false },
+      ]
+
+      act(() => {
+        result.current.setRules(rules)
+      })
+      act(() => {
+        result.current.toggleRule('remove_urls_emails')
+      })
+
+      expect(result.current.rules[0].enabled).toBe(true)
+      expect(result.current.rules[1].enabled).toBe(true)
+    })
+  })
+
+  // --- Parent-child config ---
+  describe('parent-child config', () => {
+    it('should update parent delimiter with escaping', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.updateParentConfig('delimiter', '\n')
+      })
+      expect(result.current.parentChildConfig.parent.delimiter).toBe('\\n')
+    })
+
+    it('should update parent maxLength', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.updateParentConfig('maxLength', 2048)
+      })
+      expect(result.current.parentChildConfig.parent.maxLength).toBe(2048)
+    })
+
+    it('should update child delimiter with escaping', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.updateChildConfig('delimiter', '\t')
+      })
+      expect(result.current.parentChildConfig.child.delimiter).toBe('\\t')
+    })
+
+    it('should update child maxLength', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.updateChildConfig('maxLength', 256)
+      })
+      expect(result.current.parentChildConfig.child.maxLength).toBe(256)
+    })
+
+    it('should set empty delimiter when value is empty', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.updateParentConfig('delimiter', '')
+      })
+      expect(result.current.parentChildConfig.parent.delimiter).toBe('')
+    })
+
+    it('should set chunk for context mode', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.setChunkForContext('full-doc')
+      })
+      expect(result.current.parentChildConfig.chunkForContext).toBe('full-doc')
+    })
+  })
+
+  // --- Reset to defaults ---
+  describe('resetToDefaults', () => {
+    it('should reset to default config when defaults are set', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      const defaultRules: Rules = {
+        pre_processing_rules: [{ id: 'remove_extra_spaces', enabled: true }],
+        segmentation: {
+          separator: '---',
+          max_tokens: 500,
+          chunk_overlap: 25,
+        },
+        parent_mode: 'paragraph',
+        subchunk_segmentation: {
+          separator: '\n',
+          max_tokens: 200,
+        },
+      }
+
+      act(() => {
+        result.current.setDefaultConfig(defaultRules)
+      })
+      // Change values
+      act(() => {
+        result.current.setMaxChunkLength(2048)
+        result.current.setOverlap(200)
+      })
+      // Reset
+      act(() => {
+        result.current.resetToDefaults()
+      })
+
+      expect(result.current.maxChunkLength).toBe(500)
+      expect(result.current.overlap).toBe(25)
+      expect(result.current.parentChildConfig).toEqual(defaultParentChildConfig)
+    })
+
+    it('should reset parent-child config even without default config', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      act(() => {
+        result.current.updateParentConfig('maxLength', 9999)
+      })
+      act(() => {
+        result.current.resetToDefaults()
+      })
+
+      expect(result.current.parentChildConfig).toEqual(defaultParentChildConfig)
+    })
+  })
+
+  // --- applyConfigFromRules ---
+  describe('applyConfigFromRules', () => {
+    it('should apply general config from rules', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      const rulesConfig: Rules = {
+        pre_processing_rules: [{ id: 'remove_extra_spaces', enabled: true }],
+        segmentation: {
+          separator: '|||',
+          max_tokens: 800,
+          chunk_overlap: 30,
+        },
+        parent_mode: 'paragraph',
+        subchunk_segmentation: {
+          separator: '\n',
+          max_tokens: 200,
+        },
+      }
+
+      act(() => {
+        result.current.applyConfigFromRules(rulesConfig, false)
+      })
+
+      expect(result.current.maxChunkLength).toBe(800)
+      expect(result.current.overlap).toBe(30)
+      expect(result.current.rules).toEqual(rulesConfig.pre_processing_rules)
+    })
+
+    it('should apply hierarchical config from rules', () => {
+      const { result } = renderHook(() => useSegmentationState())
+      const rulesConfig: Rules = {
+        pre_processing_rules: [],
+        segmentation: {
+          separator: '\n\n',
+          max_tokens: 1024,
+          chunk_overlap: 50,
+        },
+        parent_mode: 'full-doc',
+        subchunk_segmentation: {
+          separator: '\n',
+          max_tokens: 256,
+        },
+      }
+
+      act(() => {
+        result.current.applyConfigFromRules(rulesConfig, true)
+      })
+
+      expect(result.current.parentChildConfig.chunkForContext).toBe('full-doc')
+      expect(result.current.parentChildConfig.child.maxLength).toBe(256)
+    })
+  })
+
+  // --- getProcessRule ---
+  describe('getProcessRule', () => {
+    it('should build general process rule', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      const rule = result.current.getProcessRule(ChunkingMode.text)
+      expect(rule.mode).toBe(ProcessMode.general)
+      expect(rule.rules!.segmentation.max_tokens).toBe(DEFAULT_MAXIMUM_CHUNK_LENGTH)
+      expect(rule.rules!.segmentation.chunk_overlap).toBe(DEFAULT_OVERLAP)
+    })
+
+    it('should build parent-child process rule', () => {
+      const { result } = renderHook(() => useSegmentationState())
+
+      const rule = result.current.getProcessRule(ChunkingMode.parentChild)
+      expect(rule.mode).toBe('hierarchical')
+      expect(rule.rules!.parent_mode).toBe('paragraph')
+      expect(rule.rules!.subchunk_segmentation).toBeDefined()
+    })
+
+    it('should include summary index setting in process rule', () => {
+      const setting = { enable: true }
+      const { result } = renderHook(() =>
+        useSegmentationState({ initialSummaryIndexSetting: setting }),
+      )
+
+      const rule = result.current.getProcessRule(ChunkingMode.text)
+      expect(rule.summary_index_setting).toEqual(setting)
+    })
+  })
+
+  // --- Summary index setting ---
+  describe('handleSummaryIndexSettingChange', () => {
+    it('should update summary index setting', () => {
+      const { result } = renderHook(() =>
+        useSegmentationState({ initialSummaryIndexSetting: { enable: false } }),
+      )
+
+      act(() => {
+        result.current.handleSummaryIndexSettingChange({ enable: true })
+      })
+      expect(result.current.summaryIndexSetting).toEqual({ enable: true })
+    })
+
+    it('should merge with existing setting', () => {
+      const { result } = renderHook(() =>
+        useSegmentationState({ initialSummaryIndexSetting: { enable: true } }),
+      )
+
+      act(() => {
+        result.current.handleSummaryIndexSettingChange({ enable: false })
+      })
+      expect(result.current.summaryIndexSetting?.enable).toBe(false)
+    })
+  })
+})

web/app/components/datasets/create/step-two/index.spec.tsx

@@ -10,7 +10,7 @@ import type {
   Rules,
 } from '@/models/datasets'
 import type { RetrievalConfig } from '@/types/app'
-import { act, fireEvent, render, renderHook, screen } from '@testing-library/react'
+import { act, cleanup, fireEvent, render, renderHook, screen } from '@testing-library/react'
 import { ConfigurationMethodEnum, ModelStatusEnum, ModelTypeEnum } from '@/app/components/header/account-setting/model-provider-page/declarations'
 import { ChunkingMode, DataSourceType, ProcessMode } from '@/models/datasets'
 import { RETRIEVE_METHOD } from '@/types/app'
@@ -30,12 +30,8 @@ import {
 } from './hooks'
 import escape from './hooks/escape'
 import unescape from './hooks/unescape'
+import StepTwo from './index'
 
-// ============================================
-// Mock external dependencies
-// ============================================
-
-// Mock dataset detail context
 const mockDataset = {
   id: 'test-dataset-id',
   doc_form: ChunkingMode.text,
@@ -60,10 +56,6 @@ vi.mock('@/context/dataset-detail', () => ({
     selector({ dataset: mockCurrentDataset, mutateDatasetRes: mockMutateDatasetRes }),
 }))
 
-// Note: @/context/i18n is globally mocked in vitest.setup.ts, no need to mock here
-// Note: @/hooks/use-breakpoints uses real import
-
-// Mock model hooks
 const mockEmbeddingModelList = [
   { provider: 'openai', model: 'text-embedding-ada-002' },
   { provider: 'cohere', model: 'embed-english-v3.0' },
@@ -170,18 +162,55 @@ vi.mock('@/app/components/base/amplitude', () => ({
   trackEvent: vi.fn(),
 }))
 
-// Note: @/app/components/base/toast - uses real import (base component)
-// Note: @/app/components/datasets/common/check-rerank-model - uses real import
-// Note: @/app/components/base/float-right-container - uses real import (base component)
+// Enable IS_CE_EDITION to show QA checkbox in tests
+vi.mock('@/config', async () => {
+  const actual = await vi.importActual('@/config')
+  return { ...actual, IS_CE_EDITION: true }
+})
+
+// Mock PreviewDocumentPicker to allow testing handlePickerChange
+vi.mock('@/app/components/datasets/common/document-picker/preview-document-picker', () => ({
+  // eslint-disable-next-line ts/no-explicit-any
+  default: ({ onChange, value, files }: { onChange: (item: any) => void, value: any, files: any[] }) => (
+    <div data-testid="preview-picker">
+      <span>{value?.name}</span>
+      {files?.map((f: { id: string, name: string }) => (
+        <button key={f.id} data-testid={`picker-${f.id}`} onClick={() => onChange(f)}>
+          {f.name}
+        </button>
+      ))}
+    </div>
+  ),
+}))
 
-// Mock checkShowMultiModalTip - requires complex model list structure
 vi.mock('@/app/components/datasets/settings/utils', () => ({
   checkShowMultiModalTip: () => false,
 }))
 
-// ============================================
-// Test data factories
-// ============================================
+// Mock complex child components to avoid deep dependency chains when rendering StepTwo
+vi.mock('@/app/components/header/account-setting/model-provider-page/model-selector', () => ({
+  default: ({ onSelect, readonly }: { onSelect?: (val: Record<string, string>) => void, readonly?: boolean }) => (
+    <div data-testid="model-selector" data-readonly={readonly}>
+      <button onClick={() => onSelect?.({ provider: 'openai', model: 'text-embedding-3-small' })}>Select Model</button>
+    </div>
+  ),
+}))
+
+vi.mock('@/app/components/datasets/common/retrieval-method-config', () => ({
+  default: ({ disabled }: { disabled?: boolean }) => (
+    <div data-testid="retrieval-method-config" data-disabled={disabled}>
+      Retrieval Config
+    </div>
+  ),
+}))
+
+vi.mock('@/app/components/datasets/common/economical-retrieval-method-config', () => ({
+  default: ({ disabled }: { disabled?: boolean }) => (
+    <div data-testid="economical-retrieval-config" data-disabled={disabled}>
+      Economical Config
+    </div>
+  ),
+}))
 
 const createMockFile = (overrides?: Partial<CustomFile>): CustomFile => ({
   id: 'file-1',
@@ -371,10 +400,6 @@ describe('unescape utility', () => {
   })
 })
 
-// ============================================
-// useSegmentationState Hook Tests
-// ============================================
-
 describe('useSegmentationState', () => {
   beforeEach(() => {
     vi.clearAllMocks()
@@ -2195,3 +2220,364 @@ describe('Integration Scenarios', () => {
     })
   })
 })
+
+// ============================================
+// StepTwo Component Tests
+// ============================================
+
+describe('StepTwo Component', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mockCurrentDataset = null
+  })
+
+  afterEach(() => {
+    cleanup()
+  })
+
+  const defaultStepTwoProps = {
+    dataSourceType: DataSourceType.FILE,
+    files: [createMockFile()],
+    isAPIKeySet: true,
+    onSetting: vi.fn(),
+    notionCredentialId: '',
+    onStepChange: vi.fn(),
+  }
+
+  describe('Rendering', () => {
+    it('should render without crashing', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+    })
+
+    it('should show general chunking options when not in upload', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // Should render the segmentation section
+      expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+    })
+
+    it('should show footer with Previous and Next buttons', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      expect(screen.getByText(/stepTwo\.previousStep/i)).toBeInTheDocument()
+      expect(screen.getByText(/stepTwo\.nextStep/i)).toBeInTheDocument()
+    })
+  })
+
+  describe('Initialization', () => {
+    it('should fetch default process rule when not in setting mode', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      expect(mockFetchDefaultProcessRuleMutate).toHaveBeenCalledWith('/datasets/process-rule')
+    })
+
+    it('should apply config from rules when in setting mode with document detail', () => {
+      const docDetail = createMockDocumentDetail()
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          isSetting={true}
+          documentDetail={docDetail}
+          datasetId="test-id"
+        />,
+      )
+      // Should not fetch default rule when isSetting
+      expect(mockFetchDefaultProcessRuleMutate).not.toHaveBeenCalled()
+    })
+  })
+
+  describe('User Interactions', () => {
+    it('should call onStepChange(-1) when Previous button is clicked', () => {
+      const onStepChange = vi.fn()
+      render(<StepTwo {...defaultStepTwoProps} onStepChange={onStepChange} />)
+      fireEvent.click(screen.getByText(/stepTwo\.previousStep/i))
+      expect(onStepChange).toHaveBeenCalledWith(-1)
+    })
+
+    it('should trigger handleCreate when Next Step button is clicked', async () => {
+      const onStepChange = vi.fn()
+      render(<StepTwo {...defaultStepTwoProps} onStepChange={onStepChange} />)
+      await act(async () => {
+        fireEvent.click(screen.getByText(/stepTwo\.nextStep/i))
+      })
+      // handleCreate validates, builds params, and calls executeCreation
+      // which calls onStepChange(1) on success
+      expect(onStepChange).toHaveBeenCalledWith(1)
+    })
+
+    it('should trigger updatePreview when preview button is clicked', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // GeneralChunkingOptions renders a "Preview Chunk" button
+      const previewButtons = screen.getAllByText(/stepTwo\.previewChunk/i)
+      fireEvent.click(previewButtons[0])
+      // updatePreview calls estimateHook.fetchEstimate()
+      // No error means the handler executed successfully
+    })
+
+    it('should trigger handleDocFormChange through parent-child option switch', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // ParentChildOptions renders an OptionCard; find the title element and click its parent card
+      const parentChildTitles = screen.getAllByText(/stepTwo\.parentChild/i)
+      // The first match is the title; click it to trigger onDocFormChange
+      fireEvent.click(parentChildTitles[0])
+      // handleDocFormChange sets docForm, segmentationType, and resets estimate
+    })
+  })
+
+  describe('Conditional Rendering', () => {
+    it('should show options based on currentDataset doc_form', () => {
+      mockCurrentDataset = { ...mockDataset, doc_form: ChunkingMode.parentChild }
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          datasetId="test-id"
+        />,
+      )
+      // When currentDataset has parentChild doc_form, should show parent-child option
+      expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+    })
+
+    it('should render setting mode with Save/Cancel buttons', () => {
+      const docDetail = createMockDocumentDetail()
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          isSetting={true}
+          documentDetail={docDetail}
+          datasetId="test-id"
+        />,
+      )
+      expect(screen.getByText(/stepTwo\.save/i)).toBeInTheDocument()
+      expect(screen.getByText(/stepTwo\.cancel/i)).toBeInTheDocument()
+    })
+
+    it('should call onCancel when Cancel button is clicked in setting mode', () => {
+      const onCancel = vi.fn()
+      const docDetail = createMockDocumentDetail()
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          isSetting={true}
+          documentDetail={docDetail}
+          datasetId="test-id"
+          onCancel={onCancel}
+        />,
+      )
+      fireEvent.click(screen.getByText(/stepTwo\.cancel/i))
+      expect(onCancel).toHaveBeenCalled()
+    })
+
+    it('should trigger handleCreate (Save) in setting mode', async () => {
+      const onSave = vi.fn()
+      const docDetail = createMockDocumentDetail()
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          isSetting={true}
+          documentDetail={docDetail}
+          datasetId="test-id"
+          onSave={onSave}
+        />,
+      )
+      await act(async () => {
+        fireEvent.click(screen.getByText(/stepTwo\.save/i))
+      })
+      // handleCreate → validateParams → buildCreationParams → executeCreation → onSave
+      expect(onSave).toHaveBeenCalled()
+    })
+
+    it('should show both general and parent-child options in create page', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // When isInInit (no datasetId, no isSetting), both options should show
+      expect(screen.getByText('datasetCreation.stepTwo.general')).toBeInTheDocument()
+      expect(screen.getByText('datasetCreation.stepTwo.parentChild')).toBeInTheDocument()
+    })
+
+    it('should only show parent-child option when dataset has parentChild doc_form', () => {
+      mockCurrentDataset = { ...mockDataset, doc_form: ChunkingMode.parentChild }
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          datasetId="test-id"
+        />,
+      )
+      // showGeneralOption should be false (parentChild not in [text, qa])
+      // showParentChildOption should be true
+      expect(screen.getByText('datasetCreation.stepTwo.parentChild')).toBeInTheDocument()
+    })
+
+    it('should show general option only when dataset has text doc_form', () => {
+      mockCurrentDataset = { ...mockDataset, doc_form: ChunkingMode.text }
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          datasetId="test-id"
+        />,
+      )
+      // showGeneralOption should be true (text is in [text, qa])
+      expect(screen.getByText('datasetCreation.stepTwo.general')).toBeInTheDocument()
+    })
+  })
+
+  describe('Upload in Dataset', () => {
+    it('should show general option when in upload with text doc_form', () => {
+      mockCurrentDataset = { ...mockDataset, doc_form: ChunkingMode.text }
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          datasetId="test-id"
+        />,
+      )
+      expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+    })
+
+    it('should show general option for empty dataset (no doc_form)', () => {
+      // eslint-disable-next-line ts/no-explicit-any
+      mockCurrentDataset = { ...mockDataset, doc_form: undefined as any }
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          datasetId="test-id"
+        />,
+      )
+      expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+    })
+
+    it('should show both options in empty dataset upload', () => {
+      // eslint-disable-next-line ts/no-explicit-any
+      mockCurrentDataset = { ...mockDataset, doc_form: undefined as any }
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          datasetId="test-id"
+        />,
+      )
+      // isUploadInEmptyDataset=true shows both options
+      expect(screen.getByText('datasetCreation.stepTwo.general')).toBeInTheDocument()
+      expect(screen.getByText('datasetCreation.stepTwo.parentChild')).toBeInTheDocument()
+    })
+  })
+
+  describe('Indexing Mode', () => {
+    it('should render indexing mode section', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // IndexingModeSection renders the index mode title
+      expect(screen.getByText(/stepTwo\.indexMode/i)).toBeInTheDocument()
+    })
+
+    it('should render embedding model selector when QUALIFIED', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // ModelSelector is mocked and rendered with data-testid
+      expect(screen.getByTestId('model-selector')).toBeInTheDocument()
+    })
+
+    it('should render retrieval method config', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // RetrievalMethodConfig is mocked with data-testid
+      expect(screen.getByTestId('retrieval-method-config')).toBeInTheDocument()
+    })
+
+    it('should disable model and retrieval config when datasetId has existing data source', () => {
+      mockCurrentDataset = { ...mockDataset, data_source_type: DataSourceType.FILE }
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          datasetId="test-id"
+        />,
+      )
+      // isModelAndRetrievalConfigDisabled should be true
+      const modelSelector = screen.getByTestId('model-selector')
+      expect(modelSelector).toHaveAttribute('data-readonly', 'true')
+    })
+  })
+
+  describe('Preview Panel', () => {
+    it('should render preview panel', () => {
+      render(<StepTwo {...defaultStepTwoProps} />)
+      expect(screen.getByText('datasetCreation.stepTwo.preview')).toBeInTheDocument()
+    })
+
+    it('should hide document picker in setting mode', () => {
+      const docDetail = createMockDocumentDetail()
+      render(
+        <StepTwo
+          {...defaultStepTwoProps}
+          isSetting={true}
+          documentDetail={docDetail}
+          datasetId="test-id"
+        />,
+      )
+      // Preview panel should still render
+      expect(screen.getByText('datasetCreation.stepTwo.preview')).toBeInTheDocument()
+    })
+  })
+
+  describe('Handler Functions - Uncovered Paths', () => {
+    beforeEach(() => {
+      vi.clearAllMocks()
+      mockCurrentDataset = null
+    })
+
+    afterEach(() => {
+      cleanup()
+    })
+
+    it('should switch to QUALIFIED when selecting parentChild in ECONOMICAL mode', async () => {
+      render(<StepTwo {...defaultStepTwoProps} isAPIKeySet={false} />)
+      await vi.waitFor(() => {
+        expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+      })
+      // Click parentChild option to trigger handleDocFormChange(ChunkingMode.parentChild) with ECONOMICAL
+      const parentChildTitles = screen.getAllByText(/stepTwo\.parentChild/i)
+      fireEvent.click(parentChildTitles[0])
+    })
+
+    it('should open QA confirm dialog and confirm switch when QA selected in ECONOMICAL mode', async () => {
+      render(<StepTwo {...defaultStepTwoProps} isAPIKeySet={false} />)
+      await vi.waitFor(() => {
+        expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+      })
+      // Click QA checkbox (visible because IS_CE_EDITION is mocked as true)
+      const qaCheckbox = screen.getByText(/stepTwo\.useQALanguage/i)
+      fireEvent.click(qaCheckbox)
+      // Dialog should open → click Switch to confirm (triggers handleQAConfirm)
+      const switchButton = await screen.findByText(/stepTwo\.switch/i)
+      expect(switchButton).toBeInTheDocument()
+      fireEvent.click(switchButton)
+    })
+
+    it('should close QA confirm dialog when cancel is clicked', async () => {
+      render(<StepTwo {...defaultStepTwoProps} isAPIKeySet={false} />)
+      await vi.waitFor(() => {
+        expect(screen.getByText(/stepTwo\.segmentation/i)).toBeInTheDocument()
+      })
+      // Open QA confirm dialog
+      const qaCheckbox = screen.getByText(/stepTwo\.useQALanguage/i)
+      fireEvent.click(qaCheckbox)
+      // Click the dialog cancel button (onQAConfirmDialogClose)
+      const dialogCancelButtons = await screen.findAllByText(/stepTwo\.cancel/i)
+      fireEvent.click(dialogCancelButtons[0])
+    })
+
+    it('should handle picker change when selecting a different file', () => {
+      const files = [
+        createMockFile({ id: 'file-1', name: 'first.pdf', extension: 'pdf' }),
+        createMockFile({ id: 'file-2', name: 'second.pdf', extension: 'pdf' }),
+      ]
+      render(<StepTwo {...defaultStepTwoProps} files={files} />)
+      // Click on the second file in the mocked picker (triggers handlePickerChange)
+      const pickerButton = screen.getByTestId('picker-file-2')
+      fireEvent.click(pickerButton)
+    })
+
+    it('should show error toast when preview is clicked with maxChunkLength exceeding limit', () => {
+      // Set a high maxChunkLength via the DOM attribute
+      document.body.setAttribute('data-public-indexing-max-segmentation-tokens-length', '100')
+      render(<StepTwo {...defaultStepTwoProps} />)
+      // The default maxChunkLength (1024) now exceeds the limit (100)
+      // Click preview button to trigger updatePreview error path
+      const previewButtons = screen.getAllByText(/stepTwo\.previewChunk/i)
+      fireEvent.click(previewButtons[0])
+      // Restore
+      document.body.removeAttribute('data-public-indexing-max-segmentation-tokens-length')
+    })
+  })
+})

web/app/components/datasets/create/stepper/step.spec.tsx

@@ -0,0 +1,32 @@
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it } from 'vitest'
+import { StepperStep } from './step'
+
+describe('StepperStep', () => {
+  it('should render step name', () => {
+    render(<StepperStep name="Configure" index={0} activeIndex={0} />)
+    expect(screen.getByText('Configure')).toBeInTheDocument()
+  })
+
+  it('should show "STEP N" label for active step', () => {
+    render(<StepperStep name="Configure" index={1} activeIndex={1} />)
+    expect(screen.getByText('STEP 2')).toBeInTheDocument()
+  })
+
+  it('should show just number for non-active step', () => {
+    render(<StepperStep name="Configure" index={1} activeIndex={0} />)
+    expect(screen.getByText('2')).toBeInTheDocument()
+  })
+
+  it('should apply accent style for active step', () => {
+    render(<StepperStep name="Step A" index={0} activeIndex={0} />)
+    const nameEl = screen.getByText('Step A')
+    expect(nameEl.className).toContain('text-text-accent')
+  })
+
+  it('should apply disabled style for future step', () => {
+    render(<StepperStep name="Step C" index={2} activeIndex={0} />)
+    const nameEl = screen.getByText('Step C')
+    expect(nameEl.className).toContain('text-text-quaternary')
+  })
+})

web/app/components/datasets/create/website/base/checkbox-with-label.spec.tsx

@@ -0,0 +1,43 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import CheckboxWithLabel from './checkbox-with-label'
+
+vi.mock('@/app/components/base/tooltip', () => ({
+  default: ({ popupContent }: { popupContent?: React.ReactNode }) => <div data-testid="tooltip">{popupContent}</div>,
+}))
+
+describe('CheckboxWithLabel', () => {
+  const onChange = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render label', () => {
+    render(<CheckboxWithLabel isChecked={false} onChange={onChange} label="Accept terms" />)
+    expect(screen.getByText('Accept terms')).toBeInTheDocument()
+  })
+
+  it('should render tooltip when provided', () => {
+    render(
+      <CheckboxWithLabel
+        isChecked={false}
+        onChange={onChange}
+        label="Option"
+        tooltip="Help text"
+      />,
+    )
+    expect(screen.getByTestId('tooltip')).toBeInTheDocument()
+  })
+
+  it('should not render tooltip when not provided', () => {
+    render(<CheckboxWithLabel isChecked={false} onChange={onChange} label="Option" />)
+    expect(screen.queryByTestId('tooltip')).not.toBeInTheDocument()
+  })
+
+  it('should toggle checked state on checkbox click', () => {
+    render(<CheckboxWithLabel isChecked={false} onChange={onChange} label="Toggle" testId="my-check" />)
+    fireEvent.click(screen.getByTestId('checkbox-my-check'))
+    expect(onChange).toHaveBeenCalledWith(true)
+  })
+})

web/app/components/datasets/create/website/base/crawled-result-item.spec.tsx

@@ -0,0 +1,47 @@
+import type { CrawlResultItem as CrawlResultItemType } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import CrawledResultItem from './crawled-result-item'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({ t: (key: string) => key }),
+}))
+
+describe('CrawledResultItem', () => {
+  const defaultProps = {
+    payload: { title: 'Example Page', source_url: 'https://example.com/page' } as CrawlResultItemType,
+    isChecked: false,
+    isPreview: false,
+    onCheckChange: vi.fn(),
+    onPreview: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render title and url', () => {
+    render(<CrawledResultItem {...defaultProps} />)
+    expect(screen.getByText('Example Page')).toBeInTheDocument()
+    expect(screen.getByText('https://example.com/page')).toBeInTheDocument()
+  })
+
+  it('should apply active styling when isPreview', () => {
+    const { container } = render(<CrawledResultItem {...defaultProps} isPreview={true} />)
+    expect((container.firstChild as HTMLElement).className).toContain('bg-state-base-active')
+  })
+
+  it('should call onCheckChange with true when unchecked checkbox is clicked', () => {
+    render(<CrawledResultItem {...defaultProps} isChecked={false} testId="crawl-item" />)
+    const checkbox = screen.getByTestId('checkbox-crawl-item')
+    fireEvent.click(checkbox)
+    expect(defaultProps.onCheckChange).toHaveBeenCalledWith(true)
+  })
+
+  it('should call onCheckChange with false when checked checkbox is clicked', () => {
+    render(<CrawledResultItem {...defaultProps} isChecked={true} testId="crawl-item" />)
+    const checkbox = screen.getByTestId('checkbox-crawl-item')
+    fireEvent.click(checkbox)
+    expect(defaultProps.onCheckChange).toHaveBeenCalledWith(false)
+  })
+})

web/app/components/datasets/create/website/base/crawled-result.spec.tsx

@@ -0,0 +1,313 @@
+import type { CrawlResultItem } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import CrawledResult from './crawled-result'
+
+vi.mock('./checkbox-with-label', () => ({
+  default: ({ isChecked, onChange, label, testId }: {
+    isChecked: boolean
+    onChange: (checked: boolean) => void
+    label: string
+    testId?: string
+  }) => (
+    <label data-testid={testId}>
+      <input
+        type="checkbox"
+        checked={isChecked}
+        onChange={() => onChange(!isChecked)}
+        data-testid={`checkbox-${testId}`}
+      />
+      <span>{label}</span>
+    </label>
+  ),
+}))
+
+vi.mock('./crawled-result-item', () => ({
+  default: ({ payload, isChecked, isPreview, onCheckChange, onPreview, testId }: {
+    payload: CrawlResultItem
+    isChecked: boolean
+    isPreview: boolean
+    onCheckChange: (checked: boolean) => void
+    onPreview: () => void
+    testId?: string
+  }) => (
+    <div data-testid={testId} data-preview={isPreview}>
+      <input
+        type="checkbox"
+        checked={isChecked}
+        onChange={() => onCheckChange(!isChecked)}
+        data-testid={`check-${testId}`}
+      />
+      <span>{payload.title}</span>
+      <span>{payload.source_url}</span>
+      <button onClick={onPreview} data-testid={`preview-${testId}`}>Preview</button>
+    </div>
+  ),
+}))
+
+const createMockItem = (overrides: Partial<CrawlResultItem> = {}): CrawlResultItem => ({
+  title: 'Test Page',
+  markdown: '# Test',
+  description: 'A test page',
+  source_url: 'https://example.com',
+  ...overrides,
+})
+
+const createMockList = (): CrawlResultItem[] => [
+  createMockItem({ title: 'Page 1', source_url: 'https://example.com/1' }),
+  createMockItem({ title: 'Page 2', source_url: 'https://example.com/2' }),
+  createMockItem({ title: 'Page 3', source_url: 'https://example.com/3' }),
+]
+
+describe('CrawledResult', () => {
+  const mockOnSelectedChange = vi.fn()
+  const mockOnPreview = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('Rendering', () => {
+    it('should render select all checkbox', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      expect(screen.getByTestId('select-all')).toBeInTheDocument()
+    })
+
+    it('should render all items from list', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      expect(screen.getByTestId('item-0')).toBeInTheDocument()
+      expect(screen.getByTestId('item-1')).toBeInTheDocument()
+      expect(screen.getByTestId('item-2')).toBeInTheDocument()
+    })
+
+    it('should render scrap time info', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      expect(screen.getByText(/scrapTimeInfo/i)).toBeInTheDocument()
+    })
+
+    it('should apply custom className', () => {
+      const list = createMockList()
+      const { container } = render(
+        <CrawledResult
+          className="custom-class"
+          list={list}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      const rootElement = container.firstChild as HTMLElement
+      expect(rootElement).toHaveClass('custom-class')
+    })
+  })
+
+  describe('Select All', () => {
+    it('should call onSelectedChange with full list when not all checked', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[list[0]]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      const selectAllCheckbox = screen.getByTestId('checkbox-select-all')
+      fireEvent.click(selectAllCheckbox)
+
+      expect(mockOnSelectedChange).toHaveBeenCalledWith(list)
+    })
+
+    it('should call onSelectedChange with empty array when all checked', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={list}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      const selectAllCheckbox = screen.getByTestId('checkbox-select-all')
+      fireEvent.click(selectAllCheckbox)
+
+      expect(mockOnSelectedChange).toHaveBeenCalledWith([])
+    })
+
+    it('should show selectAll label when not all checked', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[list[0]]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      expect(screen.getByText(/selectAll/i)).toBeInTheDocument()
+    })
+
+    it('should show resetAll label when all checked', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={list}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      expect(screen.getByText(/resetAll/i)).toBeInTheDocument()
+    })
+  })
+
+  describe('Individual Item Check', () => {
+    it('should call onSelectedChange with added item when checking', () => {
+      const list = createMockList()
+      const checkedList = [list[0]]
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={checkedList}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      const item1Checkbox = screen.getByTestId('check-item-1')
+      fireEvent.click(item1Checkbox)
+
+      expect(mockOnSelectedChange).toHaveBeenCalledWith([list[0], list[1]])
+    })
+
+    it('should call onSelectedChange with removed item when unchecking', () => {
+      const list = createMockList()
+      const checkedList = [list[0], list[1]]
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={checkedList}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      const item0Checkbox = screen.getByTestId('check-item-0')
+      fireEvent.click(item0Checkbox)
+
+      expect(mockOnSelectedChange).toHaveBeenCalledWith([list[1]])
+    })
+  })
+
+  describe('Preview', () => {
+    it('should call onPreview with correct item when preview clicked', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      const previewButton = screen.getByTestId('preview-item-1')
+      fireEvent.click(previewButton)
+
+      expect(mockOnPreview).toHaveBeenCalledWith(list[1])
+    })
+
+    it('should update preview state when preview button is clicked', () => {
+      const list = createMockList()
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={1.5}
+        />,
+      )
+
+      const previewButton = screen.getByTestId('preview-item-0')
+      fireEvent.click(previewButton)
+
+      const item0 = screen.getByTestId('item-0')
+      expect(item0).toHaveAttribute('data-preview', 'true')
+    })
+  })
+
+  describe('Edge Cases', () => {
+    it('should render empty list without crashing', () => {
+      render(
+        <CrawledResult
+          list={[]}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={0}
+        />,
+      )
+
+      expect(screen.getByTestId('select-all')).toBeInTheDocument()
+    })
+
+    it('should handle single item list', () => {
+      const list = [createMockItem()]
+      render(
+        <CrawledResult
+          list={list}
+          checkedList={[]}
+          onSelectedChange={mockOnSelectedChange}
+          onPreview={mockOnPreview}
+          usedTime={0.5}
+        />,
+      )
+
+      expect(screen.getByTestId('item-0')).toBeInTheDocument()
+    })
+  })
+})

web/app/components/datasets/create/website/base/crawling.spec.tsx

@@ -0,0 +1,23 @@
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import Crawling from './crawling'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({ t: (key: string) => key }),
+}))
+vi.mock('@/app/components/base/icons/src/public/other', () => ({
+  RowStruct: (props: React.HTMLAttributes<HTMLDivElement>) => <div data-testid="row-struct" {...props} />,
+}))
+
+describe('Crawling', () => {
+  it('should render crawled count and total', () => {
+    render(<Crawling crawledNum={3} totalNum={10} />)
+    expect(screen.getByText(/3/)).toBeInTheDocument()
+    expect(screen.getByText(/10/)).toBeInTheDocument()
+  })
+
+  it('should render skeleton rows', () => {
+    render(<Crawling crawledNum={0} totalNum={5} />)
+    expect(screen.getAllByTestId('row-struct')).toHaveLength(4)
+  })
+})

web/app/components/datasets/create/website/base/error-message.spec.tsx

@@ -0,0 +1,29 @@
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import ErrorMessage from './error-message'
+
+vi.mock('@/app/components/base/icons/src/vender/solid/alertsAndFeedback', () => ({
+  AlertTriangle: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="alert-icon" {...props} />,
+}))
+
+describe('ErrorMessage', () => {
+  it('should render title', () => {
+    render(<ErrorMessage title="Something went wrong" />)
+    expect(screen.getByText('Something went wrong')).toBeInTheDocument()
+  })
+
+  it('should render error message when provided', () => {
+    render(<ErrorMessage title="Error" errorMsg="Detailed error info" />)
+    expect(screen.getByText('Detailed error info')).toBeInTheDocument()
+  })
+
+  it('should not render error message when not provided', () => {
+    render(<ErrorMessage title="Error" />)
+    expect(screen.queryByText('Detailed error info')).not.toBeInTheDocument()
+  })
+
+  it('should render alert icon', () => {
+    render(<ErrorMessage title="Error" />)
+    expect(screen.getByTestId('alert-icon')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/create/website/base/field.spec.tsx

@@ -0,0 +1,46 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Field from './field'
+
+vi.mock('@/app/components/base/tooltip', () => ({
+  default: ({ popupContent }: { popupContent?: React.ReactNode }) => <div data-testid="tooltip">{popupContent}</div>,
+}))
+
+describe('WebsiteField', () => {
+  const onChange = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render label', () => {
+    render(<Field label="URL" value="" onChange={onChange} />)
+    expect(screen.getByText('URL')).toBeInTheDocument()
+  })
+
+  it('should render required asterisk when isRequired', () => {
+    render(<Field label="URL" value="" onChange={onChange} isRequired />)
+    expect(screen.getByText('*')).toBeInTheDocument()
+  })
+
+  it('should not render required asterisk by default', () => {
+    render(<Field label="URL" value="" onChange={onChange} />)
+    expect(screen.queryByText('*')).not.toBeInTheDocument()
+  })
+
+  it('should render tooltip when provided', () => {
+    render(<Field label="URL" value="" onChange={onChange} tooltip="Enter full URL" />)
+    expect(screen.getByTestId('tooltip')).toBeInTheDocument()
+  })
+
+  it('should pass value and onChange to Input', () => {
+    render(<Field label="URL" value="https://example.com" onChange={onChange} />)
+    expect(screen.getByDisplayValue('https://example.com')).toBeInTheDocument()
+  })
+
+  it('should call onChange when input changes', () => {
+    render(<Field label="URL" value="" onChange={onChange} />)
+    fireEvent.change(screen.getByRole('textbox'), { target: { value: 'new' } })
+    expect(onChange).toHaveBeenCalledWith('new')
+  })
+})

web/app/components/datasets/create/website/base/header.spec.tsx

@@ -0,0 +1,50 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Header from './header'
+
+vi.mock('@remixicon/react', () => ({
+  RiBookOpenLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="book-icon" {...props} />,
+  RiEqualizer2Line: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="config-icon" {...props} />,
+}))
+
+describe('WebsiteHeader', () => {
+  const defaultProps = {
+    title: 'Jina Reader',
+    docTitle: 'Documentation',
+    docLink: 'https://docs.example.com',
+    onClickConfiguration: vi.fn(),
+    buttonText: 'Config',
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render title', () => {
+    render(<Header {...defaultProps} />)
+    expect(screen.getByText('Jina Reader')).toBeInTheDocument()
+  })
+
+  it('should render doc link with correct href', () => {
+    render(<Header {...defaultProps} />)
+    const link = screen.getByText('Documentation').closest('a')
+    expect(link).toHaveAttribute('href', 'https://docs.example.com')
+    expect(link).toHaveAttribute('target', '_blank')
+  })
+
+  it('should render configuration button with text when not in pipeline', () => {
+    render(<Header {...defaultProps} />)
+    expect(screen.getByText('Config')).toBeInTheDocument()
+  })
+
+  it('should call onClickConfiguration on button click', () => {
+    render(<Header {...defaultProps} />)
+    fireEvent.click(screen.getByText('Config').closest('button')!)
+    expect(defaultProps.onClickConfiguration).toHaveBeenCalledOnce()
+  })
+
+  it('should hide button text when isInPipeline', () => {
+    render(<Header {...defaultProps} isInPipeline={true} />)
+    expect(screen.queryByText('Config')).not.toBeInTheDocument()
+  })
+})

web/app/components/datasets/create/website/base/input.spec.tsx

@@ -0,0 +1,52 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Input from './input'
+
+describe('WebsiteInput', () => {
+  const onChange = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render text input by default', () => {
+    render(<Input value="hello" onChange={onChange} />)
+    const input = screen.getByDisplayValue('hello')
+    expect(input).toHaveAttribute('type', 'text')
+  })
+
+  it('should render number input when isNumber is true', () => {
+    render(<Input value={42} onChange={onChange} isNumber />)
+    const input = screen.getByDisplayValue('42')
+    expect(input).toHaveAttribute('type', 'number')
+  })
+
+  it('should call onChange with string value for text input', () => {
+    render(<Input value="" onChange={onChange} />)
+    fireEvent.change(screen.getByRole('textbox'), { target: { value: 'new value' } })
+    expect(onChange).toHaveBeenCalledWith('new value')
+  })
+
+  it('should call onChange with parsed integer for number input', () => {
+    render(<Input value={0} onChange={onChange} isNumber />)
+    fireEvent.change(screen.getByRole('spinbutton'), { target: { value: '10' } })
+    expect(onChange).toHaveBeenCalledWith(10)
+  })
+
+  it('should call onChange with empty string for NaN number input', () => {
+    render(<Input value={0} onChange={onChange} isNumber />)
+    fireEvent.change(screen.getByRole('spinbutton'), { target: { value: 'abc' } })
+    expect(onChange).toHaveBeenCalledWith('')
+  })
+
+  it('should clamp negative numbers to 0', () => {
+    render(<Input value={0} onChange={onChange} isNumber />)
+    fireEvent.change(screen.getByRole('spinbutton'), { target: { value: '-5' } })
+    expect(onChange).toHaveBeenCalledWith(0)
+  })
+
+  it('should render placeholder', () => {
+    render(<Input value="" onChange={onChange} placeholder="Enter URL" />)
+    expect(screen.getByPlaceholderText('Enter URL')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/create/website/base/options-wrap.spec.tsx

@@ -0,0 +1,51 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import OptionsWrap from './options-wrap'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({ t: (key: string) => key }),
+}))
+vi.mock('@remixicon/react', () => ({
+  RiEqualizer2Line: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="config-icon" {...props} />,
+}))
+vi.mock('@/app/components/base/icons/src/vender/line/arrows', () => ({
+  ChevronRight: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="chevron-icon" {...props} />,
+}))
+
+describe('OptionsWrap', () => {
+  it('should render children when not folded', () => {
+    render(
+      <OptionsWrap>
+        <div data-testid="child-content">Options here</div>
+      </OptionsWrap>,
+    )
+    expect(screen.getByTestId('child-content')).toBeInTheDocument()
+  })
+
+  it('should toggle fold on click', () => {
+    render(
+      <OptionsWrap>
+        <div data-testid="child-content">Options here</div>
+      </OptionsWrap>,
+    )
+    // Initially visible
+    expect(screen.getByTestId('child-content')).toBeInTheDocument()
+
+    // Click to fold
+    fireEvent.click(screen.getByText('stepOne.website.options'))
+    expect(screen.queryByTestId('child-content')).not.toBeInTheDocument()
+
+    // Click to unfold
+    fireEvent.click(screen.getByText('stepOne.website.options'))
+    expect(screen.getByTestId('child-content')).toBeInTheDocument()
+  })
+
+  it('should render options label', () => {
+    render(
+      <OptionsWrap>
+        <div>Content</div>
+      </OptionsWrap>,
+    )
+    expect(screen.getByText('stepOne.website.options')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/create/website/index.spec.tsx

@@ -0,0 +1,286 @@
+import type { DataSourceAuth } from '@/app/components/header/account-setting/data-source-page-new/types'
+import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { CredentialTypeEnum } from '@/app/components/plugins/plugin-auth/types'
+import Website from './index'
+
+const mockSetShowAccountSettingModal = vi.fn()
+
+vi.mock('@/context/modal-context', () => ({
+  useModalContext: () => ({
+    setShowAccountSettingModal: mockSetShowAccountSettingModal,
+  }),
+}))
+
+vi.mock('./index.module.css', () => ({
+  default: {
+    jinaLogo: 'jina-logo',
+    watercrawlLogo: 'watercrawl-logo',
+  },
+}))
+
+vi.mock('./firecrawl', () => ({
+  default: (props: Record<string, unknown>) => <div data-testid="firecrawl-component" data-props={JSON.stringify(props)} />,
+}))
+
+vi.mock('./jina-reader', () => ({
+  default: (props: Record<string, unknown>) => <div data-testid="jina-reader-component" data-props={JSON.stringify(props)} />,
+}))
+
+vi.mock('./watercrawl', () => ({
+  default: (props: Record<string, unknown>) => <div data-testid="watercrawl-component" data-props={JSON.stringify(props)} />,
+}))
+
+vi.mock('./no-data', () => ({
+  default: ({ onConfig, provider }: { onConfig: () => void, provider: string }) => (
+    <div data-testid="no-data-component" data-provider={provider}>
+      <button onClick={onConfig} data-testid="no-data-config-button">Configure</button>
+    </div>
+  ),
+}))
+
+let mockEnableJinaReader = true
+let mockEnableFirecrawl = true
+let mockEnableWatercrawl = true
+
+vi.mock('@/config', () => ({
+  get ENABLE_WEBSITE_JINAREADER() { return mockEnableJinaReader },
+  get ENABLE_WEBSITE_FIRECRAWL() { return mockEnableFirecrawl },
+  get ENABLE_WEBSITE_WATERCRAWL() { return mockEnableWatercrawl },
+}))
+
+const createMockCrawlOptions = (overrides: Partial<CrawlOptions> = {}): CrawlOptions => ({
+  crawl_sub_pages: true,
+  limit: 10,
+  max_depth: 2,
+  excludes: '',
+  includes: '',
+  only_main_content: false,
+  use_sitemap: false,
+  ...overrides,
+})
+
+const createMockDataSourceAuth = (
+  provider: string,
+  credentialsCount = 1,
+): DataSourceAuth => ({
+  author: 'test',
+  provider,
+  plugin_id: `${provider}-plugin`,
+  plugin_unique_identifier: `${provider}-unique`,
+  icon: 'icon.png',
+  name: provider,
+  label: { en_US: provider, zh_Hans: provider },
+  description: { en_US: `${provider} description`, zh_Hans: `${provider} description` },
+  credentials_list: Array.from({ length: credentialsCount }, (_, i) => ({
+    credential: {},
+    type: CredentialTypeEnum.API_KEY,
+    name: `cred-${i}`,
+    id: `cred-${i}`,
+    is_default: i === 0,
+    avatar_url: '',
+  })),
+})
+
+type RenderProps = {
+  authedDataSourceList?: DataSourceAuth[]
+  enableJina?: boolean
+  enableFirecrawl?: boolean
+  enableWatercrawl?: boolean
+}
+
+const renderWebsite = ({
+  authedDataSourceList = [],
+  enableJina = true,
+  enableFirecrawl = true,
+  enableWatercrawl = true,
+}: RenderProps = {}) => {
+  mockEnableJinaReader = enableJina
+  mockEnableFirecrawl = enableFirecrawl
+  mockEnableWatercrawl = enableWatercrawl
+
+  const props = {
+    onPreview: vi.fn() as (payload: CrawlResultItem) => void,
+    checkedCrawlResult: [] as CrawlResultItem[],
+    onCheckedCrawlResultChange: vi.fn() as (payload: CrawlResultItem[]) => void,
+    onCrawlProviderChange: vi.fn(),
+    onJobIdChange: vi.fn(),
+    crawlOptions: createMockCrawlOptions(),
+    onCrawlOptionsChange: vi.fn() as (payload: CrawlOptions) => void,
+    authedDataSourceList,
+  }
+
+  const result = render(<Website {...props} />)
+  return { ...result, props }
+}
+
+describe('Website', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mockEnableJinaReader = true
+    mockEnableFirecrawl = true
+    mockEnableWatercrawl = true
+  })
+
+  describe('Rendering', () => {
+    it('should render provider selection section', () => {
+      renderWebsite()
+      expect(screen.getByText(/chooseProvider/i)).toBeInTheDocument()
+    })
+
+    it('should show Jina Reader button when ENABLE_WEBSITE_JINAREADER is true', () => {
+      renderWebsite({ enableJina: true })
+      expect(screen.getByText('Jina Reader')).toBeInTheDocument()
+    })
+
+    it('should not show Jina Reader button when ENABLE_WEBSITE_JINAREADER is false', () => {
+      renderWebsite({ enableJina: false })
+      expect(screen.queryByText('Jina Reader')).not.toBeInTheDocument()
+    })
+
+    it('should show Firecrawl button when ENABLE_WEBSITE_FIRECRAWL is true', () => {
+      renderWebsite({ enableFirecrawl: true })
+      expect(screen.getByText(/Firecrawl/)).toBeInTheDocument()
+    })
+
+    it('should not show Firecrawl button when ENABLE_WEBSITE_FIRECRAWL is false', () => {
+      renderWebsite({ enableFirecrawl: false })
+      expect(screen.queryByText(/Firecrawl/)).not.toBeInTheDocument()
+    })
+
+    it('should show WaterCrawl button when ENABLE_WEBSITE_WATERCRAWL is true', () => {
+      renderWebsite({ enableWatercrawl: true })
+      expect(screen.getByText('WaterCrawl')).toBeInTheDocument()
+    })
+
+    it('should not show WaterCrawl button when ENABLE_WEBSITE_WATERCRAWL is false', () => {
+      renderWebsite({ enableWatercrawl: false })
+      expect(screen.queryByText('WaterCrawl')).not.toBeInTheDocument()
+    })
+  })
+
+  describe('Provider Selection', () => {
+    it('should select Jina Reader by default', () => {
+      const authedDataSourceList = [createMockDataSourceAuth('jinareader')]
+      renderWebsite({ authedDataSourceList })
+
+      expect(screen.getByTestId('jina-reader-component')).toBeInTheDocument()
+    })
+
+    it('should switch to Firecrawl when Firecrawl button clicked', () => {
+      const authedDataSourceList = [
+        createMockDataSourceAuth('jinareader'),
+        createMockDataSourceAuth('firecrawl'),
+      ]
+      renderWebsite({ authedDataSourceList })
+
+      const firecrawlButton = screen.getByText(/Firecrawl/)
+      fireEvent.click(firecrawlButton)
+
+      expect(screen.getByTestId('firecrawl-component')).toBeInTheDocument()
+      expect(screen.queryByTestId('jina-reader-component')).not.toBeInTheDocument()
+    })
+
+    it('should switch to WaterCrawl when WaterCrawl button clicked', () => {
+      const authedDataSourceList = [
+        createMockDataSourceAuth('jinareader'),
+        createMockDataSourceAuth('watercrawl'),
+      ]
+      renderWebsite({ authedDataSourceList })
+
+      const watercrawlButton = screen.getByText('WaterCrawl')
+      fireEvent.click(watercrawlButton)
+
+      expect(screen.getByTestId('watercrawl-component')).toBeInTheDocument()
+      expect(screen.queryByTestId('jina-reader-component')).not.toBeInTheDocument()
+    })
+
+    it('should call onCrawlProviderChange when provider switched', () => {
+      const authedDataSourceList = [
+        createMockDataSourceAuth('jinareader'),
+        createMockDataSourceAuth('firecrawl'),
+      ]
+      const { props } = renderWebsite({ authedDataSourceList })
+
+      const firecrawlButton = screen.getByText(/Firecrawl/)
+      fireEvent.click(firecrawlButton)
+
+      expect(props.onCrawlProviderChange).toHaveBeenCalledWith('firecrawl')
+    })
+  })
+
+  describe('Provider Content', () => {
+    it('should show JinaReader component when selected and available', () => {
+      const authedDataSourceList = [createMockDataSourceAuth('jinareader')]
+      renderWebsite({ authedDataSourceList })
+
+      expect(screen.getByTestId('jina-reader-component')).toBeInTheDocument()
+    })
+
+    it('should show Firecrawl component when selected and available', () => {
+      const authedDataSourceList = [
+        createMockDataSourceAuth('jinareader'),
+        createMockDataSourceAuth('firecrawl'),
+      ]
+      renderWebsite({ authedDataSourceList })
+
+      const firecrawlButton = screen.getByText(/Firecrawl/)
+      fireEvent.click(firecrawlButton)
+
+      expect(screen.getByTestId('firecrawl-component')).toBeInTheDocument()
+    })
+
+    it('should show NoData when selected provider has no credentials', () => {
+      const authedDataSourceList = [createMockDataSourceAuth('jinareader', 0)]
+      renderWebsite({ authedDataSourceList })
+
+      expect(screen.getByTestId('no-data-component')).toBeInTheDocument()
+    })
+
+    it('should show NoData when no data source available for selected provider', () => {
+      renderWebsite({ authedDataSourceList: [] })
+
+      expect(screen.getByTestId('no-data-component')).toBeInTheDocument()
+    })
+  })
+
+  describe('NoData Config', () => {
+    it('should call setShowAccountSettingModal when NoData onConfig is triggered', () => {
+      renderWebsite({ authedDataSourceList: [] })
+
+      const configButton = screen.getByTestId('no-data-config-button')
+      fireEvent.click(configButton)
+
+      expect(mockSetShowAccountSettingModal).toHaveBeenCalledWith({
+        payload: 'data-source',
+      })
+    })
+  })
+
+  describe('Edge Cases', () => {
+    it('should handle no providers enabled', () => {
+      renderWebsite({
+        enableJina: false,
+        enableFirecrawl: false,
+        enableWatercrawl: false,
+      })
+
+      expect(screen.queryByText('Jina Reader')).not.toBeInTheDocument()
+      expect(screen.queryByText(/Firecrawl/)).not.toBeInTheDocument()
+      expect(screen.queryByText('WaterCrawl')).not.toBeInTheDocument()
+    })
+
+    it('should handle only one provider enabled', () => {
+      renderWebsite({
+        enableJina: true,
+        enableFirecrawl: false,
+        enableWatercrawl: false,
+      })
+
+      expect(screen.getByText('Jina Reader')).toBeInTheDocument()
+      expect(screen.queryByText(/Firecrawl/)).not.toBeInTheDocument()
+      expect(screen.queryByText('WaterCrawl')).not.toBeInTheDocument()
+    })
+  })
+})

web/app/components/datasets/create/website/jina-reader/base/url-input.spec.tsx

@@ -0,0 +1,212 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import userEvent from '@testing-library/user-event'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+
+// ============================================================================
+// Component Imports (after mocks)
+// ============================================================================
+
+import UrlInput from './url-input'
+
+// ============================================================================
+// Mock Setup
+// ============================================================================
+
+vi.mock('@/context/i18n', () => ({
+  useDocLink: vi.fn(() => () => 'https://docs.example.com'),
+}))
+
+// ============================================================================
+// Jina Reader UrlInput Component Tests
+// ============================================================================
+
+describe('UrlInput (jina-reader)', () => {
+  const mockOnRun = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // --------------------------------------------------------------------------
+  // Rendering Tests
+  // --------------------------------------------------------------------------
+  describe('Rendering', () => {
+    it('should render input and run button', () => {
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+      expect(screen.getByRole('textbox')).toBeInTheDocument()
+      expect(screen.getByRole('button')).toBeInTheDocument()
+    })
+
+    it('should render input with placeholder from docLink', () => {
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+      const input = screen.getByRole('textbox')
+      expect(input).toHaveAttribute('placeholder', 'https://docs.example.com')
+    })
+
+    it('should show run text when not running', () => {
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+      const button = screen.getByRole('button')
+      expect(button).toHaveTextContent(/run/i)
+    })
+
+    it('should hide run text when running', () => {
+      render(<UrlInput isRunning={true} onRun={mockOnRun} />)
+      const button = screen.getByRole('button')
+      expect(button).not.toHaveTextContent(/run/i)
+    })
+
+    it('should show loading state on button when running', () => {
+      render(<UrlInput isRunning={true} onRun={mockOnRun} />)
+      const button = screen.getByRole('button')
+      expect(button).toHaveTextContent(/loading/i)
+    })
+
+    it('should not show loading state on button when not running', () => {
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+      const button = screen.getByRole('button')
+      expect(button).not.toHaveTextContent(/loading/i)
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // User Interactions Tests
+  // --------------------------------------------------------------------------
+  describe('User Interactions', () => {
+    it('should update url when user types in input', async () => {
+      const user = userEvent.setup()
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      const input = screen.getByRole('textbox')
+      await user.type(input, 'https://example.com')
+
+      expect(input).toHaveValue('https://example.com')
+    })
+
+    it('should call onRun with url when run button clicked and not running', async () => {
+      const user = userEvent.setup()
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      const input = screen.getByRole('textbox')
+      await user.type(input, 'https://example.com')
+
+      const button = screen.getByRole('button')
+      await user.click(button)
+
+      expect(mockOnRun).toHaveBeenCalledWith('https://example.com')
+      expect(mockOnRun).toHaveBeenCalledTimes(1)
+    })
+
+    it('should NOT call onRun when isRunning is true', async () => {
+      const user = userEvent.setup()
+      render(<UrlInput isRunning={true} onRun={mockOnRun} />)
+
+      const input = screen.getByRole('textbox')
+      fireEvent.change(input, { target: { value: 'https://example.com' } })
+
+      const button = screen.getByRole('button')
+      await user.click(button)
+
+      expect(mockOnRun).not.toHaveBeenCalled()
+    })
+
+    it('should call onRun with empty string when button clicked with empty input', async () => {
+      const user = userEvent.setup()
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      const button = screen.getByRole('button')
+      await user.click(button)
+
+      expect(mockOnRun).toHaveBeenCalledWith('')
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Props Variations Tests
+  // --------------------------------------------------------------------------
+  describe('Props Variations', () => {
+    it('should update button state when isRunning changes from false to true', () => {
+      const { rerender } = render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      expect(screen.getByRole('button')).toHaveTextContent(/run/i)
+
+      rerender(<UrlInput isRunning={true} onRun={mockOnRun} />)
+
+      expect(screen.getByRole('button')).not.toHaveTextContent(/run/i)
+    })
+
+    it('should preserve input value when isRunning prop changes', async () => {
+      const user = userEvent.setup()
+      const { rerender } = render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      const input = screen.getByRole('textbox')
+      await user.type(input, 'https://preserved.com')
+      expect(input).toHaveValue('https://preserved.com')
+
+      rerender(<UrlInput isRunning={true} onRun={mockOnRun} />)
+      expect(input).toHaveValue('https://preserved.com')
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Edge Cases Tests
+  // --------------------------------------------------------------------------
+  describe('Edge Cases', () => {
+    it('should handle special characters in url', async () => {
+      const user = userEvent.setup()
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      const input = screen.getByRole('textbox')
+      const specialUrl = 'https://example.com/path?query=test&param=value#anchor'
+      await user.type(input, specialUrl)
+
+      const button = screen.getByRole('button')
+      await user.click(button)
+
+      expect(mockOnRun).toHaveBeenCalledWith(specialUrl)
+    })
+
+    it('should handle rapid input changes', () => {
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      const input = screen.getByRole('textbox')
+      fireEvent.change(input, { target: { value: 'a' } })
+      fireEvent.change(input, { target: { value: 'ab' } })
+      fireEvent.change(input, { target: { value: 'https://final.com' } })
+
+      expect(input).toHaveValue('https://final.com')
+
+      fireEvent.click(screen.getByRole('button'))
+      expect(mockOnRun).toHaveBeenCalledWith('https://final.com')
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Integration Tests
+  // --------------------------------------------------------------------------
+  describe('Integration', () => {
+    it('should complete full workflow: type url -> click run -> verify callback', async () => {
+      const user = userEvent.setup()
+      render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      const input = screen.getByRole('textbox')
+      await user.type(input, 'https://mywebsite.com')
+
+      const button = screen.getByRole('button')
+      await user.click(button)
+
+      expect(mockOnRun).toHaveBeenCalledWith('https://mywebsite.com')
+    })
+
+    it('should show correct states during running workflow', () => {
+      const { rerender } = render(<UrlInput isRunning={false} onRun={mockOnRun} />)
+
+      expect(screen.getByRole('button')).toHaveTextContent(/run/i)
+
+      rerender(<UrlInput isRunning={true} onRun={mockOnRun} />)
+      expect(screen.getByRole('button')).not.toHaveTextContent(/run/i)
+
+      rerender(<UrlInput isRunning={false} onRun={mockOnRun} />)
+      expect(screen.getByRole('button')).toHaveTextContent(/run/i)
+    })
+  })
+})

web/app/components/datasets/create/website/jina-reader/options.spec.tsx

@@ -0,0 +1,209 @@
+import type { CrawlOptions } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Options from './options'
+
+// ============================================================================
+// Test Data Factory
+// ============================================================================
+
+const createMockCrawlOptions = (overrides: Partial<CrawlOptions> = {}): CrawlOptions => ({
+  crawl_sub_pages: true,
+  limit: 10,
+  max_depth: 2,
+  excludes: '',
+  includes: '',
+  only_main_content: false,
+  use_sitemap: false,
+  ...overrides,
+})
+
+// ============================================================================
+// Jina Reader Options Component Tests
+// ============================================================================
+
+describe('Options (jina-reader)', () => {
+  const mockOnChange = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  const getCheckboxes = (container: HTMLElement) => {
+    return container.querySelectorAll('[data-testid^="checkbox-"]')
+  }
+
+  // --------------------------------------------------------------------------
+  // Rendering Tests
+  // --------------------------------------------------------------------------
+  describe('Rendering', () => {
+    it('should render crawlSubPage and useSitemap checkboxes and limit field', () => {
+      const payload = createMockCrawlOptions()
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByText(/crawlSubPage/i)).toBeInTheDocument()
+      expect(screen.getByText(/useSitemap/i)).toBeInTheDocument()
+      expect(screen.getByText(/limit/i)).toBeInTheDocument()
+    })
+
+    it('should render two checkboxes', () => {
+      const payload = createMockCrawlOptions()
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes.length).toBe(2)
+    })
+
+    it('should render limit field with required indicator', () => {
+      const payload = createMockCrawlOptions()
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const requiredIndicator = screen.getByText('*')
+      expect(requiredIndicator).toBeInTheDocument()
+    })
+
+    it('should render with custom className', () => {
+      const payload = createMockCrawlOptions()
+      const { container } = render(
+        <Options payload={payload} onChange={mockOnChange} className="custom-class" />,
+      )
+
+      const rootElement = container.firstChild as HTMLElement
+      expect(rootElement).toHaveClass('custom-class')
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Props Display Tests
+  // --------------------------------------------------------------------------
+  describe('Props Display', () => {
+    it('should display crawl_sub_pages checkbox with check icon when true', () => {
+      const payload = createMockCrawlOptions({ crawl_sub_pages: true })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[0].querySelector('svg')).toBeInTheDocument()
+    })
+
+    it('should display crawl_sub_pages checkbox without check icon when false', () => {
+      const payload = createMockCrawlOptions({ crawl_sub_pages: false })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[0].querySelector('svg')).not.toBeInTheDocument()
+    })
+
+    it('should display use_sitemap checkbox with check icon when true', () => {
+      const payload = createMockCrawlOptions({ use_sitemap: true })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[1].querySelector('svg')).toBeInTheDocument()
+    })
+
+    it('should display use_sitemap checkbox without check icon when false', () => {
+      const payload = createMockCrawlOptions({ use_sitemap: false })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[1].querySelector('svg')).not.toBeInTheDocument()
+    })
+
+    it('should display limit value in input', () => {
+      const payload = createMockCrawlOptions({ limit: 25 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByDisplayValue('25')).toBeInTheDocument()
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // User Interactions Tests
+  // --------------------------------------------------------------------------
+  describe('User Interactions', () => {
+    it('should call onChange with updated crawl_sub_pages when checkbox is clicked', () => {
+      const payload = createMockCrawlOptions({ crawl_sub_pages: true })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      fireEvent.click(checkboxes[0])
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        crawl_sub_pages: false,
+      })
+    })
+
+    it('should call onChange with updated use_sitemap when checkbox is clicked', () => {
+      const payload = createMockCrawlOptions({ use_sitemap: false })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      fireEvent.click(checkboxes[1])
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        use_sitemap: true,
+      })
+    })
+
+    it('should call onChange with updated limit when input changes', () => {
+      const payload = createMockCrawlOptions({ limit: 10 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const limitInput = screen.getByDisplayValue('10')
+      fireEvent.change(limitInput, { target: { value: '50' } })
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        limit: 50,
+      })
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Edge Cases Tests
+  // --------------------------------------------------------------------------
+  describe('Edge Cases', () => {
+    it('should handle zero limit value', () => {
+      const payload = createMockCrawlOptions({ limit: 0 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const zeroInputs = screen.getAllByDisplayValue('0')
+      expect(zeroInputs.length).toBeGreaterThanOrEqual(1)
+    })
+
+    it('should preserve other payload fields when updating one field', () => {
+      const payload = createMockCrawlOptions({
+        crawl_sub_pages: true,
+        limit: 10,
+        use_sitemap: true,
+      })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const limitInput = screen.getByDisplayValue('10')
+      fireEvent.change(limitInput, { target: { value: '20' } })
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        limit: 20,
+      })
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Memoization Tests
+  // --------------------------------------------------------------------------
+  describe('Memoization', () => {
+    it('should re-render when payload changes', () => {
+      const payload1 = createMockCrawlOptions({ limit: 10 })
+      const payload2 = createMockCrawlOptions({ limit: 20 })
+
+      const { rerender } = render(<Options payload={payload1} onChange={mockOnChange} />)
+      expect(screen.getByDisplayValue('10')).toBeInTheDocument()
+
+      rerender(<Options payload={payload2} onChange={mockOnChange} />)
+      expect(screen.getByDisplayValue('20')).toBeInTheDocument()
+    })
+  })
+})

web/app/components/datasets/create/website/no-data.spec.tsx

@@ -0,0 +1,230 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DataSourceProvider } from '@/models/common'
+import NoData from './no-data'
+
+// ============================================================================
+// Mock Setup
+// ============================================================================
+
+// Mock CSS module
+vi.mock('./index.module.css', () => ({
+  default: {
+    jinaLogo: 'jinaLogo',
+    watercrawlLogo: 'watercrawlLogo',
+  },
+}))
+
+// Feature flags - default all enabled
+let mockEnableFirecrawl = true
+let mockEnableJinaReader = true
+let mockEnableWaterCrawl = true
+
+vi.mock('@/config', () => ({
+  get ENABLE_WEBSITE_FIRECRAWL() { return mockEnableFirecrawl },
+  get ENABLE_WEBSITE_JINAREADER() { return mockEnableJinaReader },
+  get ENABLE_WEBSITE_WATERCRAWL() { return mockEnableWaterCrawl },
+}))
+
+// ============================================================================
+// NoData Component Tests
+// ============================================================================
+
+describe('NoData', () => {
+  const mockOnConfig = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mockEnableFirecrawl = true
+    mockEnableJinaReader = true
+    mockEnableWaterCrawl = true
+  })
+
+  // --------------------------------------------------------------------------
+  // Rendering Tests - Per Provider
+  // --------------------------------------------------------------------------
+  describe('Rendering per provider', () => {
+    it('should render fireCrawl provider with emoji and not-configured message', () => {
+      // Arrange & Act
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.fireCrawl} />)
+
+      // Assert
+      expect(screen.getByText('🔥')).toBeInTheDocument()
+      const titleAndDesc = screen.getAllByText(/fireCrawlNotConfigured/i)
+      expect(titleAndDesc).toHaveLength(2)
+    })
+
+    it('should render jinaReader provider with jina logo and not-configured message', () => {
+      // Arrange & Act
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.jinaReader} />)
+
+      // Assert
+      const titleAndDesc = screen.getAllByText(/jinaReaderNotConfigured/i)
+      expect(titleAndDesc).toHaveLength(2)
+    })
+
+    it('should render waterCrawl provider with emoji and not-configured message', () => {
+      // Arrange & Act
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.waterCrawl} />)
+
+      // Assert
+      expect(screen.getByText('💧')).toBeInTheDocument()
+      const titleAndDesc = screen.getAllByText(/waterCrawlNotConfigured/i)
+      expect(titleAndDesc).toHaveLength(2)
+    })
+
+    it('should render configure button for each provider', () => {
+      // Arrange & Act
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.fireCrawl} />)
+
+      // Assert
+      expect(screen.getByRole('button', { name: /configure/i })).toBeInTheDocument()
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // User Interactions
+  // --------------------------------------------------------------------------
+  describe('User Interactions', () => {
+    it('should call onConfig when configure button is clicked', () => {
+      // Arrange
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.fireCrawl} />)
+
+      // Act
+      fireEvent.click(screen.getByRole('button', { name: /configure/i }))
+
+      // Assert
+      expect(mockOnConfig).toHaveBeenCalledTimes(1)
+    })
+
+    it('should call onConfig for jinaReader provider', () => {
+      // Arrange
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.jinaReader} />)
+
+      // Act
+      fireEvent.click(screen.getByRole('button', { name: /configure/i }))
+
+      // Assert
+      expect(mockOnConfig).toHaveBeenCalledTimes(1)
+    })
+
+    it('should call onConfig for waterCrawl provider', () => {
+      // Arrange
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.waterCrawl} />)
+
+      // Act
+      fireEvent.click(screen.getByRole('button', { name: /configure/i }))
+
+      // Assert
+      expect(mockOnConfig).toHaveBeenCalledTimes(1)
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Feature Flag Disabled - Returns null
+  // --------------------------------------------------------------------------
+  describe('Disabled providers (feature flag off)', () => {
+    it('should fall back to jinaReader when fireCrawl is disabled but jinaReader enabled', () => {
+      // Arrange — fireCrawl config is null, falls back to providerConfig.jinareader
+      mockEnableFirecrawl = false
+
+      // Act
+      const { container } = render(
+        <NoData onConfig={mockOnConfig} provider={DataSourceProvider.fireCrawl} />,
+      )
+
+      // Assert — renders the jinaReader fallback (not null)
+      expect(container.innerHTML).not.toBe('')
+      expect(screen.getAllByText(/jinaReaderNotConfigured/).length).toBeGreaterThan(0)
+    })
+
+    it('should return null when jinaReader is disabled', () => {
+      // Arrange — jinaReader is the only provider without a fallback
+      mockEnableJinaReader = false
+
+      // Act
+      const { container } = render(
+        <NoData onConfig={mockOnConfig} provider={DataSourceProvider.jinaReader} />,
+      )
+
+      // Assert
+      expect(container.innerHTML).toBe('')
+    })
+
+    it('should fall back to jinaReader when waterCrawl is disabled but jinaReader enabled', () => {
+      // Arrange — waterCrawl config is null, falls back to providerConfig.jinareader
+      mockEnableWaterCrawl = false
+
+      // Act
+      const { container } = render(
+        <NoData onConfig={mockOnConfig} provider={DataSourceProvider.waterCrawl} />,
+      )
+
+      // Assert — renders the jinaReader fallback (not null)
+      expect(container.innerHTML).not.toBe('')
+      expect(screen.getAllByText(/jinaReaderNotConfigured/).length).toBeGreaterThan(0)
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Fallback behavior
+  // --------------------------------------------------------------------------
+  describe('Fallback behavior', () => {
+    it('should fall back to jinaReader config for unknown provider value', () => {
+      // Arrange - the || fallback goes to providerConfig.jinareader
+      // Since DataSourceProvider only has 3 values, we test the fallback
+      // by checking that jinaReader is the fallback when provider doesn't match
+      mockEnableJinaReader = true
+
+      // Act
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.jinaReader} />)
+
+      // Assert
+      expect(screen.getAllByText(/jinaReaderNotConfigured/i).length).toBeGreaterThan(0)
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Edge Cases
+  // --------------------------------------------------------------------------
+  describe('Edge Cases', () => {
+    it('should not call onConfig without user interaction', () => {
+      // Arrange & Act
+      render(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.fireCrawl} />)
+
+      // Assert
+      expect(mockOnConfig).not.toHaveBeenCalled()
+    })
+
+    it('should render correctly when all providers are enabled', () => {
+      // Arrange - all flags are true by default
+
+      // Act
+      const { rerender } = render(
+        <NoData onConfig={mockOnConfig} provider={DataSourceProvider.fireCrawl} />,
+      )
+      expect(screen.getByText('🔥')).toBeInTheDocument()
+
+      rerender(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.jinaReader} />)
+      expect(screen.getAllByText(/jinaReaderNotConfigured/i).length).toBeGreaterThan(0)
+
+      rerender(<NoData onConfig={mockOnConfig} provider={DataSourceProvider.waterCrawl} />)
+      expect(screen.getByText('💧')).toBeInTheDocument()
+    })
+
+    it('should return null when all providers are disabled and fireCrawl is selected', () => {
+      // Arrange
+      mockEnableFirecrawl = false
+      mockEnableJinaReader = false
+      mockEnableWaterCrawl = false
+
+      // Act
+      const { container } = render(
+        <NoData onConfig={mockOnConfig} provider={DataSourceProvider.fireCrawl} />,
+      )
+
+      // Assert
+      expect(container.innerHTML).toBe('')
+    })
+  })
+})

web/app/components/datasets/create/website/preview.spec.tsx

@@ -0,0 +1,256 @@
+import type { CrawlResultItem } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import WebsitePreview from './preview'
+
+// ============================================================================
+// Mock Setup
+// ============================================================================
+
+// Mock the CSS module import - returns class names as-is
+vi.mock('../file-preview/index.module.css', () => ({
+  default: {
+    filePreview: 'filePreview',
+    previewHeader: 'previewHeader',
+    title: 'title',
+    previewContent: 'previewContent',
+    fileContent: 'fileContent',
+  },
+}))
+
+// ============================================================================
+// Test Data Factory
+// ============================================================================
+
+const createPayload = (overrides: Partial<CrawlResultItem> = {}): CrawlResultItem => ({
+  title: 'Test Page Title',
+  markdown: 'This is **markdown** content',
+  description: 'A test description',
+  source_url: 'https://example.com/page',
+  ...overrides,
+})
+
+// ============================================================================
+// WebsitePreview Component Tests
+// ============================================================================
+
+describe('WebsitePreview', () => {
+  const mockHidePreview = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // --------------------------------------------------------------------------
+  // Rendering Tests
+  // --------------------------------------------------------------------------
+  describe('Rendering', () => {
+    it('should render without crashing', () => {
+      // Arrange
+      const payload = createPayload()
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      expect(screen.getByText('Test Page Title')).toBeInTheDocument()
+    })
+
+    it('should render the page preview header text', () => {
+      // Arrange
+      const payload = createPayload()
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert - i18n returns the key path
+      expect(screen.getByText(/pagePreview/i)).toBeInTheDocument()
+    })
+
+    it('should render the payload title', () => {
+      // Arrange
+      const payload = createPayload({ title: 'My Custom Page' })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      expect(screen.getByText('My Custom Page')).toBeInTheDocument()
+    })
+
+    it('should render the payload source_url', () => {
+      // Arrange
+      const payload = createPayload({ source_url: 'https://docs.dify.ai/intro' })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      const urlElement = screen.getByText('https://docs.dify.ai/intro')
+      expect(urlElement).toBeInTheDocument()
+      expect(urlElement).toHaveAttribute('title', 'https://docs.dify.ai/intro')
+    })
+
+    it('should render the payload markdown content', () => {
+      // Arrange
+      const payload = createPayload({ markdown: 'Hello world markdown' })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      expect(screen.getByText('Hello world markdown')).toBeInTheDocument()
+    })
+
+    it('should render the close button (XMarkIcon)', () => {
+      // Arrange
+      const payload = createPayload()
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert - the close button container is a div with cursor-pointer
+      const closeButton = screen.getByText(/pagePreview/i).parentElement?.querySelector('.cursor-pointer')
+      expect(closeButton).toBeInTheDocument()
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // User Interactions
+  // --------------------------------------------------------------------------
+  describe('User Interactions', () => {
+    it('should call hidePreview when close button is clicked', () => {
+      // Arrange
+      const payload = createPayload()
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Act - find the close button div with cursor-pointer class
+      const closeButton = screen.getByText(/pagePreview/i)
+        .closest('[class*="title"]')!
+        .querySelector('.cursor-pointer') as HTMLElement
+      fireEvent.click(closeButton)
+
+      // Assert
+      expect(mockHidePreview).toHaveBeenCalledTimes(1)
+    })
+
+    it('should call hidePreview exactly once per click', () => {
+      // Arrange
+      const payload = createPayload()
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Act
+      const closeButton = screen.getByText(/pagePreview/i)
+        .closest('[class*="title"]')!
+        .querySelector('.cursor-pointer') as HTMLElement
+      fireEvent.click(closeButton)
+      fireEvent.click(closeButton)
+
+      // Assert
+      expect(mockHidePreview).toHaveBeenCalledTimes(2)
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Props Display Tests
+  // --------------------------------------------------------------------------
+  describe('Props Display', () => {
+    it('should display all payload fields simultaneously', () => {
+      // Arrange
+      const payload = createPayload({
+        title: 'Full Title',
+        source_url: 'https://full.example.com',
+        markdown: 'Full markdown text',
+      })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      expect(screen.getByText('Full Title')).toBeInTheDocument()
+      expect(screen.getByText('https://full.example.com')).toBeInTheDocument()
+      expect(screen.getByText('Full markdown text')).toBeInTheDocument()
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Edge Cases
+  // --------------------------------------------------------------------------
+  describe('Edge Cases', () => {
+    it('should render with empty title', () => {
+      // Arrange
+      const payload = createPayload({ title: '' })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert - component still renders, url is visible
+      expect(screen.getByText('https://example.com/page')).toBeInTheDocument()
+    })
+
+    it('should render with empty markdown', () => {
+      // Arrange
+      const payload = createPayload({ markdown: '' })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      expect(screen.getByText('Test Page Title')).toBeInTheDocument()
+    })
+
+    it('should render with empty source_url', () => {
+      // Arrange
+      const payload = createPayload({ source_url: '' })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      expect(screen.getByText('Test Page Title')).toBeInTheDocument()
+    })
+
+    it('should render with very long content', () => {
+      // Arrange
+      const longMarkdown = 'A'.repeat(5000)
+      const payload = createPayload({ markdown: longMarkdown })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert
+      expect(screen.getByText(longMarkdown)).toBeInTheDocument()
+    })
+
+    it('should render with special characters in title', () => {
+      // Arrange
+      const payload = createPayload({ title: '<script>alert("xss")</script>' })
+
+      // Act
+      render(<WebsitePreview payload={payload} hidePreview={mockHidePreview} />)
+
+      // Assert - React escapes HTML by default
+      expect(screen.getByText('<script>alert("xss")</script>')).toBeInTheDocument()
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // CSS Module Classes
+  // --------------------------------------------------------------------------
+  describe('CSS Module Classes', () => {
+    it('should apply filePreview class to root container', () => {
+      // Arrange
+      const payload = createPayload()
+
+      // Act
+      const { container } = render(
+        <WebsitePreview payload={payload} hidePreview={mockHidePreview} />,
+      )
+
+      // Assert
+      const root = container.firstElementChild
+      expect(root?.className).toContain('filePreview')
+      expect(root?.className).toContain('h-full')
+    })
+  })
+})

web/app/components/datasets/create/website/watercrawl/options.spec.tsx

@@ -0,0 +1,294 @@
+import type { CrawlOptions } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Options from './options'
+
+// ============================================================================
+// Test Data Factory
+// ============================================================================
+
+const createMockCrawlOptions = (overrides: Partial<CrawlOptions> = {}): CrawlOptions => ({
+  crawl_sub_pages: true,
+  limit: 10,
+  max_depth: 2,
+  excludes: '',
+  includes: '',
+  only_main_content: false,
+  use_sitemap: false,
+  ...overrides,
+})
+
+// ============================================================================
+// WaterCrawl Options Component Tests
+// ============================================================================
+
+describe('Options (watercrawl)', () => {
+  const mockOnChange = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  const getCheckboxes = (container: HTMLElement) => {
+    return container.querySelectorAll('[data-testid^="checkbox-"]')
+  }
+
+  // --------------------------------------------------------------------------
+  // Rendering Tests
+  // --------------------------------------------------------------------------
+  describe('Rendering', () => {
+    it('should render all form fields', () => {
+      const payload = createMockCrawlOptions()
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByText(/crawlSubPage/i)).toBeInTheDocument()
+      expect(screen.getByText(/extractOnlyMainContent/i)).toBeInTheDocument()
+      expect(screen.getByText(/limit/i)).toBeInTheDocument()
+      expect(screen.getByText(/maxDepth/i)).toBeInTheDocument()
+      expect(screen.getByText(/excludePaths/i)).toBeInTheDocument()
+      expect(screen.getByText(/includeOnlyPaths/i)).toBeInTheDocument()
+    })
+
+    it('should render two checkboxes', () => {
+      const payload = createMockCrawlOptions()
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes.length).toBe(2)
+    })
+
+    it('should render limit field with required indicator', () => {
+      const payload = createMockCrawlOptions()
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const requiredIndicator = screen.getByText('*')
+      expect(requiredIndicator).toBeInTheDocument()
+    })
+
+    it('should render placeholder for excludes field', () => {
+      const payload = createMockCrawlOptions()
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByPlaceholderText('blog/*, /about/*')).toBeInTheDocument()
+    })
+
+    it('should render placeholder for includes field', () => {
+      const payload = createMockCrawlOptions()
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByPlaceholderText('articles/*')).toBeInTheDocument()
+    })
+
+    it('should render with custom className', () => {
+      const payload = createMockCrawlOptions()
+      const { container } = render(
+        <Options payload={payload} onChange={mockOnChange} className="custom-class" />,
+      )
+
+      const rootElement = container.firstChild as HTMLElement
+      expect(rootElement).toHaveClass('custom-class')
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Props Display Tests
+  // --------------------------------------------------------------------------
+  describe('Props Display', () => {
+    it('should display crawl_sub_pages checkbox with check icon when true', () => {
+      const payload = createMockCrawlOptions({ crawl_sub_pages: true })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[0].querySelector('svg')).toBeInTheDocument()
+    })
+
+    it('should display crawl_sub_pages checkbox without check icon when false', () => {
+      const payload = createMockCrawlOptions({ crawl_sub_pages: false })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[0].querySelector('svg')).not.toBeInTheDocument()
+    })
+
+    it('should display only_main_content checkbox with check icon when true', () => {
+      const payload = createMockCrawlOptions({ only_main_content: true })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[1].querySelector('svg')).toBeInTheDocument()
+    })
+
+    it('should display only_main_content checkbox without check icon when false', () => {
+      const payload = createMockCrawlOptions({ only_main_content: false })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      expect(checkboxes[1].querySelector('svg')).not.toBeInTheDocument()
+    })
+
+    it('should display limit value in input', () => {
+      const payload = createMockCrawlOptions({ limit: 25 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByDisplayValue('25')).toBeInTheDocument()
+    })
+
+    it('should display max_depth value in input', () => {
+      const payload = createMockCrawlOptions({ max_depth: 5 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByDisplayValue('5')).toBeInTheDocument()
+    })
+
+    it('should display excludes value in input', () => {
+      const payload = createMockCrawlOptions({ excludes: 'test/*' })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByDisplayValue('test/*')).toBeInTheDocument()
+    })
+
+    it('should display includes value in input', () => {
+      const payload = createMockCrawlOptions({ includes: 'docs/*' })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      expect(screen.getByDisplayValue('docs/*')).toBeInTheDocument()
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // User Interactions Tests
+  // --------------------------------------------------------------------------
+  describe('User Interactions', () => {
+    it('should call onChange with updated crawl_sub_pages when checkbox is clicked', () => {
+      const payload = createMockCrawlOptions({ crawl_sub_pages: true })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      fireEvent.click(checkboxes[0])
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        crawl_sub_pages: false,
+      })
+    })
+
+    it('should call onChange with updated only_main_content when checkbox is clicked', () => {
+      const payload = createMockCrawlOptions({ only_main_content: false })
+      const { container } = render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const checkboxes = getCheckboxes(container)
+      fireEvent.click(checkboxes[1])
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        only_main_content: true,
+      })
+    })
+
+    it('should call onChange with updated limit when input changes', () => {
+      const payload = createMockCrawlOptions({ limit: 10 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const limitInput = screen.getByDisplayValue('10')
+      fireEvent.change(limitInput, { target: { value: '50' } })
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        limit: 50,
+      })
+    })
+
+    it('should call onChange with updated max_depth when input changes', () => {
+      const payload = createMockCrawlOptions({ max_depth: 2 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const maxDepthInput = screen.getByDisplayValue('2')
+      fireEvent.change(maxDepthInput, { target: { value: '10' } })
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        max_depth: 10,
+      })
+    })
+
+    it('should call onChange with updated excludes when input changes', () => {
+      const payload = createMockCrawlOptions({ excludes: '' })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const excludesInput = screen.getByPlaceholderText('blog/*, /about/*')
+      fireEvent.change(excludesInput, { target: { value: 'admin/*' } })
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        excludes: 'admin/*',
+      })
+    })
+
+    it('should call onChange with updated includes when input changes', () => {
+      const payload = createMockCrawlOptions({ includes: '' })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const includesInput = screen.getByPlaceholderText('articles/*')
+      fireEvent.change(includesInput, { target: { value: 'public/*' } })
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        ...payload,
+        includes: 'public/*',
+      })
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Edge Cases Tests
+  // --------------------------------------------------------------------------
+  describe('Edge Cases', () => {
+    it('should preserve other payload fields when updating one field', () => {
+      const payload = createMockCrawlOptions({
+        crawl_sub_pages: true,
+        limit: 10,
+        max_depth: 2,
+        excludes: 'test/*',
+        includes: 'docs/*',
+        only_main_content: true,
+      })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const limitInput = screen.getByDisplayValue('10')
+      fireEvent.change(limitInput, { target: { value: '20' } })
+
+      expect(mockOnChange).toHaveBeenCalledWith({
+        crawl_sub_pages: true,
+        limit: 20,
+        max_depth: 2,
+        excludes: 'test/*',
+        includes: 'docs/*',
+        only_main_content: true,
+        use_sitemap: false,
+      })
+    })
+
+    it('should handle zero values', () => {
+      const payload = createMockCrawlOptions({ limit: 0, max_depth: 0 })
+      render(<Options payload={payload} onChange={mockOnChange} />)
+
+      const zeroInputs = screen.getAllByDisplayValue('0')
+      expect(zeroInputs.length).toBeGreaterThanOrEqual(1)
+    })
+  })
+
+  // --------------------------------------------------------------------------
+  // Memoization Tests
+  // --------------------------------------------------------------------------
+  describe('Memoization', () => {
+    it('should re-render when payload changes', () => {
+      const payload1 = createMockCrawlOptions({ limit: 10 })
+      const payload2 = createMockCrawlOptions({ limit: 20 })
+
+      const { rerender } = render(<Options payload={payload1} onChange={mockOnChange} />)
+      expect(screen.getByDisplayValue('10')).toBeInTheDocument()
+
+      rerender(<Options payload={payload2} onChange={mockOnChange} />)
+      expect(screen.getByDisplayValue('20')).toBeInTheDocument()
+    })
+  })
+})

web/app/components/datasets/documents/components/document-list/hooks/use-document-actions.spec.ts

@@ -0,0 +1,237 @@
+import { act, renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DocumentActionType } from '@/models/datasets'
+import { useDocumentActions } from './use-document-actions'
+
+const mockArchive = vi.fn()
+const mockSummary = vi.fn()
+const mockEnable = vi.fn()
+const mockDisable = vi.fn()
+const mockDelete = vi.fn()
+const mockRetryIndex = vi.fn()
+const mockDownloadZip = vi.fn()
+let mockIsDownloadingZip = false
+
+vi.mock('@/service/knowledge/use-document', () => ({
+  useDocumentArchive: () => ({ mutateAsync: mockArchive }),
+  useDocumentSummary: () => ({ mutateAsync: mockSummary }),
+  useDocumentEnable: () => ({ mutateAsync: mockEnable }),
+  useDocumentDisable: () => ({ mutateAsync: mockDisable }),
+  useDocumentDelete: () => ({ mutateAsync: mockDelete }),
+  useDocumentBatchRetryIndex: () => ({ mutateAsync: mockRetryIndex }),
+  useDocumentDownloadZip: () => ({ mutateAsync: mockDownloadZip, isPending: mockIsDownloadingZip }),
+}))
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
+}))
+
+const mockToastNotify = vi.fn()
+vi.mock('@/app/components/base/toast', () => ({
+  default: { notify: (...args: unknown[]) => mockToastNotify(...args) },
+}))
+
+const mockDownloadBlob = vi.fn()
+vi.mock('@/utils/download', () => ({
+  downloadBlob: (...args: unknown[]) => mockDownloadBlob(...args),
+}))
+
+describe('useDocumentActions', () => {
+  const defaultOptions = {
+    datasetId: 'ds-1',
+    selectedIds: ['doc-1', 'doc-2'],
+    downloadableSelectedIds: ['doc-1'],
+    onUpdate: vi.fn(),
+    onClearSelection: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mockIsDownloadingZip = false
+  })
+
+  it('should return expected functions and state', () => {
+    const { result } = renderHook(() => useDocumentActions(defaultOptions))
+    expect(result.current.handleAction).toBeInstanceOf(Function)
+    expect(result.current.handleBatchReIndex).toBeInstanceOf(Function)
+    expect(result.current.handleBatchDownload).toBeInstanceOf(Function)
+    expect(typeof result.current.isDownloadingZip).toBe('boolean')
+  })
+
+  describe('handleAction', () => {
+    it('should call archive API and show success toast', async () => {
+      mockArchive.mockResolvedValue({ result: 'success' })
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleAction(DocumentActionType.archive)()
+      })
+
+      expect(mockArchive).toHaveBeenCalledWith({
+        datasetId: 'ds-1',
+        documentIds: ['doc-1', 'doc-2'],
+      })
+      expect(mockToastNotify).toHaveBeenCalledWith(
+        expect.objectContaining({ type: 'success' }),
+      )
+      expect(defaultOptions.onUpdate).toHaveBeenCalled()
+    })
+
+    it('should call enable API on enable action', async () => {
+      mockEnable.mockResolvedValue({ result: 'success' })
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleAction(DocumentActionType.enable)()
+      })
+
+      expect(mockEnable).toHaveBeenCalledWith({
+        datasetId: 'ds-1',
+        documentIds: ['doc-1', 'doc-2'],
+      })
+      expect(defaultOptions.onUpdate).toHaveBeenCalled()
+    })
+
+    it('should call disable API on disable action', async () => {
+      mockDisable.mockResolvedValue({ result: 'success' })
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleAction(DocumentActionType.disable)()
+      })
+
+      expect(mockDisable).toHaveBeenCalled()
+    })
+
+    it('should call summary API on summary action', async () => {
+      mockSummary.mockResolvedValue({ result: 'success' })
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleAction(DocumentActionType.summary)()
+      })
+
+      expect(mockSummary).toHaveBeenCalled()
+    })
+
+    it('should call onClearSelection on delete action success', async () => {
+      mockDelete.mockResolvedValue({ result: 'success' })
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleAction(DocumentActionType.delete)()
+      })
+
+      expect(mockDelete).toHaveBeenCalled()
+      expect(defaultOptions.onClearSelection).toHaveBeenCalled()
+      expect(defaultOptions.onUpdate).toHaveBeenCalled()
+    })
+
+    it('should not call onClearSelection on non-delete action success', async () => {
+      mockArchive.mockResolvedValue({ result: 'success' })
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleAction(DocumentActionType.archive)()
+      })
+
+      expect(defaultOptions.onClearSelection).not.toHaveBeenCalled()
+    })
+
+    it('should show error toast on action failure', async () => {
+      mockArchive.mockRejectedValue(new Error('fail'))
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleAction(DocumentActionType.archive)()
+      })
+
+      expect(mockToastNotify).toHaveBeenCalledWith(
+        expect.objectContaining({ type: 'error' }),
+      )
+      expect(defaultOptions.onUpdate).not.toHaveBeenCalled()
+    })
+  })
+
+  describe('handleBatchReIndex', () => {
+    it('should call retry index API and show success toast', async () => {
+      mockRetryIndex.mockResolvedValue({ result: 'success' })
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleBatchReIndex()
+      })
+
+      expect(mockRetryIndex).toHaveBeenCalledWith({
+        datasetId: 'ds-1',
+        documentIds: ['doc-1', 'doc-2'],
+      })
+      expect(defaultOptions.onClearSelection).toHaveBeenCalled()
+      expect(defaultOptions.onUpdate).toHaveBeenCalled()
+    })
+
+    it('should show error toast on reindex failure', async () => {
+      mockRetryIndex.mockRejectedValue(new Error('fail'))
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleBatchReIndex()
+      })
+
+      expect(mockToastNotify).toHaveBeenCalledWith(
+        expect.objectContaining({ type: 'error' }),
+      )
+    })
+  })
+
+  describe('handleBatchDownload', () => {
+    it('should download blob on success', async () => {
+      const blob = new Blob(['test'])
+      mockDownloadZip.mockResolvedValue(blob)
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleBatchDownload()
+      })
+
+      expect(mockDownloadZip).toHaveBeenCalledWith({
+        datasetId: 'ds-1',
+        documentIds: ['doc-1'],
+      })
+      expect(mockDownloadBlob).toHaveBeenCalledWith(
+        expect.objectContaining({
+          data: blob,
+          fileName: expect.stringContaining('-docs.zip'),
+        }),
+      )
+    })
+
+    it('should show error toast on download failure', async () => {
+      mockDownloadZip.mockRejectedValue(new Error('fail'))
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleBatchDownload()
+      })
+
+      expect(mockToastNotify).toHaveBeenCalledWith(
+        expect.objectContaining({ type: 'error' }),
+      )
+    })
+
+    it('should show error toast when blob is null', async () => {
+      mockDownloadZip.mockResolvedValue(null)
+      const { result } = renderHook(() => useDocumentActions(defaultOptions))
+
+      await act(async () => {
+        await result.current.handleBatchDownload()
+      })
+
+      expect(mockToastNotify).toHaveBeenCalledWith(
+        expect.objectContaining({ type: 'error' }),
+      )
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source-options/datasource-icon.spec.tsx

@@ -0,0 +1,33 @@
+import { render } from '@testing-library/react'
+import { describe, expect, it } from 'vitest'
+import DatasourceIcon from './datasource-icon'
+
+describe('DatasourceIcon', () => {
+  it('should render icon with background image', () => {
+    const { container } = render(<DatasourceIcon iconUrl="https://example.com/icon.png" />)
+    const iconDiv = container.querySelector('[style*="background-image"]')
+    expect(iconDiv).not.toBeNull()
+    expect(iconDiv?.getAttribute('style')).toContain('https://example.com/icon.png')
+  })
+
+  it('should apply size class for sm', () => {
+    const { container } = render(<DatasourceIcon iconUrl="/icon.png" size="sm" />)
+    const wrapper = container.firstChild as HTMLElement
+    expect(wrapper.className).toContain('w-5')
+    expect(wrapper.className).toContain('h-5')
+  })
+
+  it('should apply size class for md', () => {
+    const { container } = render(<DatasourceIcon iconUrl="/icon.png" size="md" />)
+    const wrapper = container.firstChild as HTMLElement
+    expect(wrapper.className).toContain('w-6')
+    expect(wrapper.className).toContain('h-6')
+  })
+
+  it('should apply size class for xs', () => {
+    const { container } = render(<DatasourceIcon iconUrl="/icon.png" size="xs" />)
+    const wrapper = container.firstChild as HTMLElement
+    expect(wrapper.className).toContain('w-4')
+    expect(wrapper.className).toContain('h-4')
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source-options/hooks.spec.tsx

@@ -0,0 +1,141 @@
+import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
+import { renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { useDatasourceIcon } from './hooks'
+
+const mockTransformDataSourceToTool = vi.fn()
+
+vi.mock('@/app/components/workflow/block-selector/utils', () => ({
+  transformDataSourceToTool: (...args: unknown[]) => mockTransformDataSourceToTool(...args),
+}))
+
+let mockDataSourceListReturn: {
+  data: Array<{
+    plugin_id: string
+    provider: string
+    declaration: { identity: { icon: string, author: string } }
+  }> | undefined
+  isSuccess: boolean
+}
+
+vi.mock('@/service/use-pipeline', () => ({
+  useDataSourceList: () => mockDataSourceListReturn,
+}))
+
+vi.mock('@/utils/var', () => ({
+  basePath: '',
+}))
+
+const createMockDataSourceNode = (overrides?: Partial<DataSourceNodeType>): DataSourceNodeType => ({
+  plugin_id: 'plugin-abc',
+  provider_type: 'builtin',
+  provider_name: 'web-scraper',
+  datasource_name: 'scraper',
+  datasource_label: 'Web Scraper',
+  datasource_parameters: {},
+  datasource_configurations: {},
+  title: 'DataSource',
+  desc: '',
+  type: '' as DataSourceNodeType['type'],
+  ...overrides,
+} as DataSourceNodeType)
+
+describe('useDatasourceIcon', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+    mockDataSourceListReturn = { data: undefined, isSuccess: false }
+    mockTransformDataSourceToTool.mockReset()
+  })
+
+  // Returns undefined when data has not loaded
+  describe('Loading State', () => {
+    it('should return undefined when data is not loaded (isSuccess false)', () => {
+      mockDataSourceListReturn = { data: undefined, isSuccess: false }
+
+      const { result } = renderHook(() =>
+        useDatasourceIcon(createMockDataSourceNode()),
+      )
+
+      expect(result.current).toBeUndefined()
+    })
+  })
+
+  // Returns correct icon when plugin_id matches
+  describe('Icon Resolution', () => {
+    it('should return correct icon when plugin_id matches', () => {
+      const mockIcon = 'https://example.com/icon.svg'
+      mockDataSourceListReturn = {
+        data: [
+          {
+            plugin_id: 'plugin-abc',
+            provider: 'web-scraper',
+            declaration: { identity: { icon: mockIcon, author: 'dify' } },
+          },
+        ],
+        isSuccess: true,
+      }
+      mockTransformDataSourceToTool.mockImplementation((item: { plugin_id: string, declaration: { identity: { icon: string } } }) => ({
+        plugin_id: item.plugin_id,
+        icon: item.declaration.identity.icon,
+      }))
+
+      const { result } = renderHook(() =>
+        useDatasourceIcon(createMockDataSourceNode({ plugin_id: 'plugin-abc' })),
+      )
+
+      expect(result.current).toBe(mockIcon)
+    })
+
+    it('should return undefined when plugin_id does not match', () => {
+      mockDataSourceListReturn = {
+        data: [
+          {
+            plugin_id: 'plugin-xyz',
+            provider: 'other',
+            declaration: { identity: { icon: '/icon.svg', author: 'dify' } },
+          },
+        ],
+        isSuccess: true,
+      }
+      mockTransformDataSourceToTool.mockImplementation((item: { plugin_id: string, declaration: { identity: { icon: string } } }) => ({
+        plugin_id: item.plugin_id,
+        icon: item.declaration.identity.icon,
+      }))
+
+      const { result } = renderHook(() =>
+        useDatasourceIcon(createMockDataSourceNode({ plugin_id: 'plugin-abc' })),
+      )
+
+      expect(result.current).toBeUndefined()
+    })
+  })
+
+  // basePath prepending
+  describe('basePath Prepending', () => {
+    it('should prepend basePath to icon URL when not already included', () => {
+      // basePath is mocked as '' so prepending '' to '/icon.png' results in '/icon.png'
+      // The important thing is that the forEach logic runs without error
+      mockDataSourceListReturn = {
+        data: [
+          {
+            plugin_id: 'plugin-abc',
+            provider: 'web-scraper',
+            declaration: { identity: { icon: '/icon.png', author: 'dify' } },
+          },
+        ],
+        isSuccess: true,
+      }
+      mockTransformDataSourceToTool.mockImplementation((item: { plugin_id: string, declaration: { identity: { icon: string } } }) => ({
+        plugin_id: item.plugin_id,
+        icon: item.declaration.identity.icon,
+      }))
+
+      const { result } = renderHook(() =>
+        useDatasourceIcon(createMockDataSourceNode({ plugin_id: 'plugin-abc' })),
+      )
+
+      // With empty basePath, icon stays as '/icon.png'
+      expect(result.current).toBe('/icon.png')
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source-options/option-card.spec.tsx

@@ -0,0 +1,110 @@
+import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import OptionCard from './option-card'
+
+const TEST_ICON_URL = 'https://example.com/test-icon.png'
+
+vi.mock('./hooks', () => ({
+  useDatasourceIcon: () => TEST_ICON_URL,
+}))
+
+vi.mock('./datasource-icon', () => ({
+  default: ({ iconUrl }: { iconUrl: string }) => (
+    <img data-testid="datasource-icon" src={iconUrl} alt="datasource" />
+  ),
+}))
+
+const createMockNodeData = (overrides: Partial<DataSourceNodeType> = {}): DataSourceNodeType => ({
+  title: 'Test Node',
+  desc: '',
+  type: {} as DataSourceNodeType['type'],
+  plugin_id: 'test-plugin',
+  provider_type: 'builtin',
+  provider_name: 'test-provider',
+  datasource_name: 'test-ds',
+  datasource_label: 'Test DS',
+  datasource_parameters: {},
+  datasource_configurations: {},
+  ...overrides,
+} as DataSourceNodeType)
+
+describe('OptionCard', () => {
+  const defaultProps = {
+    label: 'Google Drive',
+    selected: false,
+    nodeData: createMockNodeData(),
+    onClick: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering: label text and icon
+  describe('Rendering', () => {
+    it('should render label text', () => {
+      render(<OptionCard {...defaultProps} />)
+
+      expect(screen.getByText('Google Drive')).toBeInTheDocument()
+    })
+
+    it('should render datasource icon with correct URL', () => {
+      render(<OptionCard {...defaultProps} />)
+
+      const icon = screen.getByTestId('datasource-icon')
+      expect(icon).toHaveAttribute('src', TEST_ICON_URL)
+    })
+
+    it('should set title attribute on label element', () => {
+      render(<OptionCard {...defaultProps} />)
+
+      expect(screen.getByTitle('Google Drive')).toBeInTheDocument()
+    })
+  })
+
+  // User interactions: clicking the card
+  describe('User Interactions', () => {
+    it('should call onClick when clicked', () => {
+      render(<OptionCard {...defaultProps} />)
+
+      fireEvent.click(screen.getByText('Google Drive'))
+
+      expect(defaultProps.onClick).toHaveBeenCalledOnce()
+    })
+
+    it('should not throw when onClick is undefined', () => {
+      expect(() => {
+        const { container } = render(
+          <OptionCard {...defaultProps} onClick={undefined} />,
+        )
+        fireEvent.click(container.firstElementChild!)
+      }).not.toThrow()
+    })
+  })
+
+  // Props: selected state applies different styles
+  describe('Props', () => {
+    it('should apply selected styles when selected is true', () => {
+      const { container } = render(<OptionCard {...defaultProps} selected />)
+
+      const card = container.firstElementChild
+      expect(card?.className).toContain('border-components-option-card-option-selected-border')
+      expect(card?.className).toContain('bg-components-option-card-option-selected-bg')
+    })
+
+    it('should apply default styles when selected is false', () => {
+      const { container } = render(<OptionCard {...defaultProps} selected={false} />)
+
+      const card = container.firstElementChild
+      expect(card?.className).not.toContain('border-components-option-card-option-selected-border')
+    })
+
+    it('should apply text-text-primary class to label when selected', () => {
+      render(<OptionCard {...defaultProps} selected />)
+
+      const labelEl = screen.getByTitle('Google Drive')
+      expect(labelEl.className).toContain('text-text-primary')
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/base/credential-selector/item.spec.tsx

@@ -0,0 +1,45 @@
+import type { DataSourceCredential } from '@/types/pipeline'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Item from './item'
+
+vi.mock('@remixicon/react', () => ({
+  RiCheckLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="check-icon" {...props} />,
+}))
+vi.mock('@/app/components/datasets/common/credential-icon', () => ({
+  CredentialIcon: () => <span data-testid="credential-icon" />,
+}))
+
+describe('CredentialSelectorItem', () => {
+  const defaultProps = {
+    credential: { id: 'cred-1', name: 'My Account', avatar_url: 'https://example.com/avatar.png' } as DataSourceCredential,
+    isSelected: false,
+    onCredentialChange: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render credential name and icon', () => {
+    render(<Item {...defaultProps} />)
+    expect(screen.getByText('My Account')).toBeInTheDocument()
+    expect(screen.getByTestId('credential-icon')).toBeInTheDocument()
+  })
+
+  it('should show check icon when selected', () => {
+    render(<Item {...defaultProps} isSelected={true} />)
+    expect(screen.getByTestId('check-icon')).toBeInTheDocument()
+  })
+
+  it('should not show check icon when not selected', () => {
+    render(<Item {...defaultProps} isSelected={false} />)
+    expect(screen.queryByTestId('check-icon')).not.toBeInTheDocument()
+  })
+
+  it('should call onCredentialChange with credential id on click', () => {
+    render(<Item {...defaultProps} />)
+    fireEvent.click(screen.getByText('My Account'))
+    expect(defaultProps.onCredentialChange).toHaveBeenCalledWith('cred-1')
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/base/credential-selector/list.spec.tsx

@@ -0,0 +1,46 @@
+import type { DataSourceCredential } from '@/types/pipeline'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import List from './list'
+
+vi.mock('@remixicon/react', () => ({
+  RiCheckLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="check-icon" {...props} />,
+}))
+vi.mock('@/app/components/datasets/common/credential-icon', () => ({
+  CredentialIcon: () => <span data-testid="credential-icon" />,
+}))
+
+describe('CredentialSelectorList', () => {
+  const mockCredentials: DataSourceCredential[] = [
+    { id: 'cred-1', name: 'Account A', avatar_url: '' } as DataSourceCredential,
+    { id: 'cred-2', name: 'Account B', avatar_url: '' } as DataSourceCredential,
+  ]
+
+  const defaultProps = {
+    currentCredentialId: 'cred-1',
+    credentials: mockCredentials,
+    onCredentialChange: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render all credentials', () => {
+    render(<List {...defaultProps} />)
+    expect(screen.getByText('Account A')).toBeInTheDocument()
+    expect(screen.getByText('Account B')).toBeInTheDocument()
+  })
+
+  it('should mark selected credential with check icon', () => {
+    render(<List {...defaultProps} />)
+    const checkIcons = screen.getAllByTestId('check-icon')
+    expect(checkIcons).toHaveLength(1)
+  })
+
+  it('should call onCredentialChange on item click', () => {
+    render(<List {...defaultProps} />)
+    fireEvent.click(screen.getByText('Account B'))
+    expect(defaultProps.onCredentialChange).toHaveBeenCalledWith('cred-2')
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/base/credential-selector/trigger.spec.tsx

@@ -0,0 +1,49 @@
+import type { DataSourceCredential } from '@/types/pipeline'
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import Trigger from './trigger'
+
+vi.mock('@remixicon/react', () => ({
+  RiArrowDownSLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="arrow-icon" {...props} />,
+}))
+vi.mock('@/app/components/datasets/common/credential-icon', () => ({
+  CredentialIcon: () => <span data-testid="credential-icon" />,
+}))
+
+describe('CredentialSelectorTrigger', () => {
+  it('should render credential name when provided', () => {
+    render(
+      <Trigger
+        currentCredential={{ id: 'cred-1', name: 'Account A', avatar_url: '' } as DataSourceCredential}
+        isOpen={false}
+      />,
+    )
+    expect(screen.getByText('Account A')).toBeInTheDocument()
+  })
+
+  it('should render empty name when no credential', () => {
+    render(<Trigger currentCredential={undefined} isOpen={false} />)
+    expect(screen.getByTestId('credential-icon')).toBeInTheDocument()
+  })
+
+  it('should render arrow icon', () => {
+    render(
+      <Trigger
+        currentCredential={{ id: 'cred-1', name: 'A', avatar_url: '' } as DataSourceCredential}
+        isOpen={false}
+      />,
+    )
+    expect(screen.getByTestId('arrow-icon')).toBeInTheDocument()
+  })
+
+  it('should apply hover style when open', () => {
+    const { container } = render(
+      <Trigger
+        currentCredential={{ id: 'cred-1', name: 'A', avatar_url: '' } as DataSourceCredential}
+        isOpen={true}
+      />,
+    )
+    const wrapper = container.firstChild as HTMLElement
+    expect(wrapper.className).toContain('bg-state-base-hover')
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/base/header.spec.tsx

@@ -1,658 +1,64 @@
-import type { DataSourceCredential } from '@/types/pipeline'
-import { fireEvent, render, screen } from '@testing-library/react'
-import * as React from 'react'
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
 import Header from './header'
 
-// Mock CredentialTypeEnum to avoid deep import chain issues
-enum MockCredentialTypeEnum {
-  OAUTH2 = 'oauth2',
-  API_KEY = 'api_key',
-}
-
-// Mock plugin-auth module to avoid deep import chain issues
-vi.mock('@/app/components/plugins/plugin-auth', () => ({
-  CredentialTypeEnum: {
-    OAUTH2: 'oauth2',
-    API_KEY: 'api_key',
-  },
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string, opts?: { pluginName?: string }) => opts?.pluginName ? `${key}-${opts.pluginName}` : key,
+  }),
 }))
 
-// Mock portal-to-follow-elem - required for CredentialSelector
-vi.mock('@/app/components/base/portal-to-follow-elem', () => {
-  const MockPortalToFollowElem = ({ children, open }: any) => {
-    return (
-      <div data-testid="portal-root" data-open={open}>
-        {React.Children.map(children, (child: any) => {
-          if (!child)
-            return null
-          return React.cloneElement(child, { __portalOpen: open })
-        })}
-      </div>
-    )
-  }
+vi.mock('@remixicon/react', () => ({
+  RiBookOpenLine: () => <span data-testid="book-icon" />,
+  RiEqualizer2Line: ({ onClick }: { onClick?: () => void }) => <span data-testid="config-icon" onClick={onClick} />,
+}))
 
-  const MockPortalToFollowElemTrigger = ({ children, onClick, className, __portalOpen }: any) => (
-    <div data-testid="portal-trigger" onClick={onClick} className={className} data-open={__portalOpen}>
-      {children}
-    </div>
-  )
+vi.mock('@/app/components/base/button', () => ({
+  default: ({ children }: { children: React.ReactNode }) => <button>{children}</button>,
+}))
 
-  const MockPortalToFollowElemContent = ({ children, className, __portalOpen }: any) => {
-    if (!__portalOpen)
-      return null
-    return (
-      <div data-testid="portal-content" className={className}>
-        {children}
-      </div>
-    )
-  }
+vi.mock('@/app/components/base/divider', () => ({
+  default: () => <span data-testid="divider" />,
+}))
 
-  return {
-    PortalToFollowElem: MockPortalToFollowElem,
-    PortalToFollowElemTrigger: MockPortalToFollowElemTrigger,
-    PortalToFollowElemContent: MockPortalToFollowElemContent,
-  }
-})
+vi.mock('@/app/components/base/tooltip', () => ({
+  default: ({ children }: { children: React.ReactNode }) => <div data-testid="tooltip">{children}</div>,
+}))
 
-// ==========================================
-// Test Data Builders
-// ==========================================
-const createMockCredential = (overrides?: Partial<DataSourceCredential>): DataSourceCredential => ({
-  id: 'cred-1',
-  name: 'Test Credential',
-  avatar_url: 'https://example.com/avatar.png',
-  credential: { key: 'value' },
-  is_default: false,
-  type: MockCredentialTypeEnum.OAUTH2 as unknown as DataSourceCredential['type'],
-  ...overrides,
-})
-
-const createMockCredentials = (count: number = 3): DataSourceCredential[] =>
-  Array.from({ length: count }, (_, i) =>
-    createMockCredential({
-      id: `cred-${i + 1}`,
-      name: `Credential ${i + 1}`,
-      avatar_url: `https://example.com/avatar-${i + 1}.png`,
-      is_default: i === 0,
-    }))
-
-type HeaderProps = React.ComponentProps<typeof Header>
-
-const createDefaultProps = (overrides?: Partial<HeaderProps>): HeaderProps => ({
-  docTitle: 'Documentation',
-  docLink: 'https://docs.example.com',
-  pluginName: 'Test Plugin',
-  currentCredentialId: 'cred-1',
-  onCredentialChange: vi.fn(),
-  credentials: createMockCredentials(),
-  ...overrides,
-})
+vi.mock('./credential-selector', () => ({
+  default: () => <div data-testid="credential-selector" />,
+}))
 
 describe('Header', () => {
-  beforeEach(() => {
-    vi.clearAllMocks()
+  const defaultProps = {
+    docTitle: 'Documentation',
+    docLink: 'https://docs.example.com',
+    onClickConfiguration: vi.fn(),
+    pluginName: 'TestPlugin',
+    credentials: [],
+    currentCredentialId: '',
+    onCredentialChange: vi.fn(),
+  }
+
+  it('should render doc link with title', () => {
+    render(<Header {...defaultProps} />)
+    expect(screen.getByText('Documentation')).toBeInTheDocument()
   })
 
-  // ==========================================
-  // Rendering Tests
-  // ==========================================
-  describe('Rendering', () => {
-    it('should render without crashing', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      expect(screen.getByText('Documentation')).toBeInTheDocument()
-    })
-
-    it('should render documentation link with correct attributes', () => {
-      // Arrange
-      const props = createDefaultProps({
-        docTitle: 'API Docs',
-        docLink: 'https://api.example.com/docs',
-      })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      const link = screen.getByRole('link', { name: /API Docs/i })
-      expect(link).toHaveAttribute('href', 'https://api.example.com/docs')
-      expect(link).toHaveAttribute('target', '_blank')
-      expect(link).toHaveAttribute('rel', 'noopener noreferrer')
-    })
-
-    it('should render document title with title attribute', () => {
-      // Arrange
-      const props = createDefaultProps({ docTitle: 'My Documentation' })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      const titleSpan = screen.getByText('My Documentation')
-      expect(titleSpan).toHaveAttribute('title', 'My Documentation')
-    })
-
-    it('should render CredentialSelector with correct props', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert - CredentialSelector should render current credential name
-      expect(screen.getByText('Credential 1')).toBeInTheDocument()
-    })
-
-    it('should render configuration button', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should render book icon in documentation link', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert - RiBookOpenLine renders as SVG
-      const link = screen.getByRole('link')
-      const svg = link.querySelector('svg')
-      expect(svg).toBeInTheDocument()
-    })
-
-    it('should render divider between credential selector and configuration button', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      const { container } = render(<Header {...props} />)
-
-      // Assert - Divider component should be rendered
-      // Divider typically renders as a div with specific styling
-      const divider = container.querySelector('[class*="divider"]') || container.querySelector('.mx-1.h-3\\.5')
-      expect(divider).toBeInTheDocument()
-    })
+  it('should render credential selector', () => {
+    render(<Header {...defaultProps} />)
+    expect(screen.getByTestId('credential-selector')).toBeInTheDocument()
   })
 
-  // ==========================================
-  // Props Testing
-  // ==========================================
-  describe('Props', () => {
-    describe('docTitle prop', () => {
-      it('should display the document title', () => {
-        // Arrange
-        const props = createDefaultProps({ docTitle: 'Getting Started Guide' })
-
-        // Act
-        render(<Header {...props} />)
-
-        // Assert
-        expect(screen.getByText('Getting Started Guide')).toBeInTheDocument()
-      })
-
-      it.each([
-        'Quick Start',
-        'API Reference',
-        'Configuration Guide',
-        'Plugin Documentation',
-      ])('should display "%s" as document title', (title) => {
-        // Arrange
-        const props = createDefaultProps({ docTitle: title })
-
-        // Act
-        render(<Header {...props} />)
-
-        // Assert
-        expect(screen.getByText(title)).toBeInTheDocument()
-      })
-    })
-
-    describe('docLink prop', () => {
-      it('should set correct href on documentation link', () => {
-        // Arrange
-        const props = createDefaultProps({ docLink: 'https://custom.docs.com/guide' })
-
-        // Act
-        render(<Header {...props} />)
-
-        // Assert
-        const link = screen.getByRole('link')
-        expect(link).toHaveAttribute('href', 'https://custom.docs.com/guide')
-      })
-
-      it.each([
-        'https://docs.dify.ai',
-        'https://example.com/api',
-        '/local/docs',
-      ])('should accept "%s" as docLink', (link) => {
-        // Arrange
-        const props = createDefaultProps({ docLink: link })
-
-        // Act
-        render(<Header {...props} />)
-
-        // Assert
-        expect(screen.getByRole('link')).toHaveAttribute('href', link)
-      })
-    })
-
-    describe('pluginName prop', () => {
-      it('should pass pluginName to translation function', () => {
-        // Arrange
-        const props = createDefaultProps({ pluginName: 'MyPlugin' })
-
-        // Act
-        render(<Header {...props} />)
-
-        // Assert - The translation mock returns the key with options
-        // Tooltip uses the translated content
-        expect(screen.getByRole('button')).toBeInTheDocument()
-      })
-    })
-
-    describe('onClickConfiguration prop', () => {
-      it('should call onClickConfiguration when configuration icon is clicked', () => {
-        // Arrange
-        const mockOnClick = vi.fn()
-        const props = createDefaultProps({ onClickConfiguration: mockOnClick })
-        render(<Header {...props} />)
-
-        // Act - Find the configuration button and click the icon inside
-        // The button contains the RiEqualizer2Line icon with onClick handler
-        const configButton = screen.getByRole('button')
-        const configIcon = configButton.querySelector('svg')
-        expect(configIcon).toBeInTheDocument()
-        fireEvent.click(configIcon!)
-
-        // Assert
-        expect(mockOnClick).toHaveBeenCalledTimes(1)
-      })
-
-      it('should not crash when onClickConfiguration is undefined', () => {
-        // Arrange
-        const props = createDefaultProps({ onClickConfiguration: undefined })
-        render(<Header {...props} />)
-
-        // Act - Find the configuration button and click the icon inside
-        const configButton = screen.getByRole('button')
-        const configIcon = configButton.querySelector('svg')
-        expect(configIcon).toBeInTheDocument()
-        fireEvent.click(configIcon!)
-
-        // Assert - Component should still be rendered (no crash)
-        expect(screen.getByRole('button')).toBeInTheDocument()
-      })
-    })
-
-    describe('CredentialSelector props passthrough', () => {
-      it('should pass currentCredentialId to CredentialSelector', () => {
-        // Arrange
-        const props = createDefaultProps({ currentCredentialId: 'cred-2' })
-
-        // Act
-        render(<Header {...props} />)
-
-        // Assert - Should display the second credential
-        expect(screen.getByText('Credential 2')).toBeInTheDocument()
-      })
-
-      it('should pass credentials to CredentialSelector', () => {
-        // Arrange
-        const customCredentials = [
-          createMockCredential({ id: 'custom-1', name: 'Custom Credential' }),
-        ]
-        const props = createDefaultProps({
-          credentials: customCredentials,
-          currentCredentialId: 'custom-1',
-        })
-
-        // Act
-        render(<Header {...props} />)
-
-        // Assert
-        expect(screen.getByText('Custom Credential')).toBeInTheDocument()
-      })
-
-      it('should pass onCredentialChange to CredentialSelector', () => {
-        // Arrange
-        const mockOnChange = vi.fn()
-        const props = createDefaultProps({ onCredentialChange: mockOnChange })
-        render(<Header {...props} />)
-
-        // Act - Open dropdown and select a credential
-        // Use getAllByTestId and select the first one (CredentialSelector's trigger)
-        const triggers = screen.getAllByTestId('portal-trigger')
-        fireEvent.click(triggers[0])
-        const credential2 = screen.getByText('Credential 2')
-        fireEvent.click(credential2)
-
-        // Assert
-        expect(mockOnChange).toHaveBeenCalledWith('cred-2')
-      })
-    })
+  it('should render configuration button', () => {
+    render(<Header {...defaultProps} />)
+    expect(screen.getByTestId('config-icon')).toBeInTheDocument()
   })
 
-  // ==========================================
-  // User Interactions
-  // ==========================================
-  describe('User Interactions', () => {
-    it('should open external link in new tab when clicking documentation link', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert - Link has target="_blank" for new tab
-      const link = screen.getByRole('link')
-      expect(link).toHaveAttribute('target', '_blank')
-    })
-
-    it('should allow credential selection through CredentialSelector', () => {
-      // Arrange
-      const mockOnChange = vi.fn()
-      const props = createDefaultProps({ onCredentialChange: mockOnChange })
-      render(<Header {...props} />)
-
-      // Act - Open dropdown (use first trigger which is CredentialSelector's)
-      const triggers = screen.getAllByTestId('portal-trigger')
-      fireEvent.click(triggers[0])
-
-      // Assert - Dropdown should be open
-      expect(screen.getByTestId('portal-content')).toBeInTheDocument()
-    })
-
-    it('should trigger configuration callback when clicking config icon', () => {
-      // Arrange
-      const mockOnConfig = vi.fn()
-      const props = createDefaultProps({ onClickConfiguration: mockOnConfig })
-      const { container } = render(<Header {...props} />)
-
-      // Act
-      const configIcon = container.querySelector('.h-4.w-4')
-      fireEvent.click(configIcon!)
-
-      // Assert
-      expect(mockOnConfig).toHaveBeenCalled()
-    })
-  })
-
-  // ==========================================
-  // Component Memoization
-  // ==========================================
-  describe('Component Memoization', () => {
-    it('should be wrapped with React.memo', () => {
-      // Assert
-      expect(Header.$$typeof).toBe(Symbol.for('react.memo'))
-    })
-
-    it('should not re-render when props remain the same', () => {
-      // Arrange
-      const props = createDefaultProps()
-      const renderSpy = vi.fn()
-
-      const TrackedHeader: React.FC<HeaderProps> = (trackedProps) => {
-        renderSpy()
-        return <Header {...trackedProps} />
-      }
-      const MemoizedTracked = React.memo(TrackedHeader)
-
-      // Act
-      const { rerender } = render(<MemoizedTracked {...props} />)
-      rerender(<MemoizedTracked {...props} />)
-
-      // Assert - Should only render once due to same props
-      expect(renderSpy).toHaveBeenCalledTimes(1)
-    })
-
-    it('should re-render when docTitle changes', () => {
-      // Arrange
-      const props = createDefaultProps({ docTitle: 'Original Title' })
-      const { rerender } = render(<Header {...props} />)
-
-      // Assert initial
-      expect(screen.getByText('Original Title')).toBeInTheDocument()
-
-      // Act
-      rerender(<Header {...props} docTitle="Updated Title" />)
-
-      // Assert
-      expect(screen.getByText('Updated Title')).toBeInTheDocument()
-    })
-
-    it('should re-render when currentCredentialId changes', () => {
-      // Arrange
-      const props = createDefaultProps({ currentCredentialId: 'cred-1' })
-      const { rerender } = render(<Header {...props} />)
-
-      // Assert initial
-      expect(screen.getByText('Credential 1')).toBeInTheDocument()
-
-      // Act
-      rerender(<Header {...props} currentCredentialId="cred-2" />)
-
-      // Assert
-      expect(screen.getByText('Credential 2')).toBeInTheDocument()
-    })
-  })
-
-  // ==========================================
-  // Edge Cases
-  // ==========================================
-  describe('Edge Cases', () => {
-    it('should handle empty docTitle', () => {
-      // Arrange
-      const props = createDefaultProps({ docTitle: '' })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert - Should render without crashing
-      const link = screen.getByRole('link')
-      expect(link).toBeInTheDocument()
-    })
-
-    it('should handle very long docTitle', () => {
-      // Arrange
-      const longTitle = 'A'.repeat(200)
-      const props = createDefaultProps({ docTitle: longTitle })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      expect(screen.getByText(longTitle)).toBeInTheDocument()
-    })
-
-    it('should handle special characters in docTitle', () => {
-      // Arrange
-      const specialTitle = 'Docs & Guide <v2> "Special"'
-      const props = createDefaultProps({ docTitle: specialTitle })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      expect(screen.getByText(specialTitle)).toBeInTheDocument()
-    })
-
-    it('should handle empty credentials array', () => {
-      // Arrange
-      const props = createDefaultProps({
-        credentials: [],
-        currentCredentialId: '',
-      })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert - Should render without crashing
-      expect(screen.getByRole('link')).toBeInTheDocument()
-    })
-
-    it('should handle special characters in pluginName', () => {
-      // Arrange
-      const props = createDefaultProps({ pluginName: 'Plugin & Tool <v1>' })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert - Should render without crashing
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should handle unicode characters in docTitle', () => {
-      // Arrange
-      const props = createDefaultProps({ docTitle: '文档说明 📚' })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      expect(screen.getByText('文档说明 📚')).toBeInTheDocument()
-    })
-  })
-
-  // ==========================================
-  // Styling
-  // ==========================================
-  describe('Styling', () => {
-    it('should apply correct classes to container', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      const { container } = render(<Header {...props} />)
-
-      // Assert
-      const rootDiv = container.firstChild as HTMLElement
-      expect(rootDiv).toHaveClass('flex', 'items-center', 'justify-between', 'gap-x-2')
-    })
-
-    it('should apply correct classes to documentation link', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      const link = screen.getByRole('link')
-      expect(link).toHaveClass('system-xs-medium', 'text-text-accent')
-    })
-
-    it('should apply shrink-0 to documentation link', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      const link = screen.getByRole('link')
-      expect(link).toHaveClass('shrink-0')
-    })
-  })
-
-  // ==========================================
-  // Integration Tests
-  // ==========================================
-  describe('Integration', () => {
-    it('should work with full credential workflow', () => {
-      // Arrange
-      const mockOnCredentialChange = vi.fn()
-      const props = createDefaultProps({
-        onCredentialChange: mockOnCredentialChange,
-        currentCredentialId: 'cred-1',
-      })
-      render(<Header {...props} />)
-
-      // Assert initial state
-      expect(screen.getByText('Credential 1')).toBeInTheDocument()
-
-      // Act - Open dropdown and select different credential
-      // Use first trigger which is CredentialSelector's
-      const triggers = screen.getAllByTestId('portal-trigger')
-      fireEvent.click(triggers[0])
-
-      const credential3 = screen.getByText('Credential 3')
-      fireEvent.click(credential3)
-
-      // Assert
-      expect(mockOnCredentialChange).toHaveBeenCalledWith('cred-3')
-    })
-
-    it('should display all components together correctly', () => {
-      // Arrange
-      const mockOnConfig = vi.fn()
-      const props = createDefaultProps({
-        docTitle: 'Integration Test Docs',
-        docLink: 'https://test.com/docs',
-        pluginName: 'TestPlugin',
-        onClickConfiguration: mockOnConfig,
-      })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert - All main elements present
-      expect(screen.getByText('Credential 1')).toBeInTheDocument() // CredentialSelector
-      expect(screen.getByRole('button')).toBeInTheDocument() // Config button
-      expect(screen.getByText('Integration Test Docs')).toBeInTheDocument() // Doc link
-      expect(screen.getByRole('link')).toHaveAttribute('href', 'https://test.com/docs')
-    })
-  })
-
-  // ==========================================
-  // Accessibility
-  // ==========================================
-  describe('Accessibility', () => {
-    it('should have accessible link', () => {
-      // Arrange
-      const props = createDefaultProps({ docTitle: 'Accessible Docs' })
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      const link = screen.getByRole('link', { name: /Accessible Docs/i })
-      expect(link).toBeInTheDocument()
-    })
-
-    it('should have accessible button for configuration', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      const button = screen.getByRole('button')
-      expect(button).toBeInTheDocument()
-    })
-
-    it('should have noopener noreferrer for security on external links', () => {
-      // Arrange
-      const props = createDefaultProps()
-
-      // Act
-      render(<Header {...props} />)
-
-      // Assert
-      const link = screen.getByRole('link')
-      expect(link).toHaveAttribute('rel', 'noopener noreferrer')
-    })
+  it('should link to external doc', () => {
+    render(<Header {...defaultProps} />)
+    const link = screen.getByText('Documentation').closest('a')
+    expect(link).toHaveAttribute('href', 'https://docs.example.com')
+    expect(link).toHaveAttribute('target', '_blank')
   })
 })

web/app/components/datasets/documents/create-from-pipeline/data-source/online-documents/page-selector/utils.spec.ts

@@ -0,0 +1,100 @@
+import type { NotionPageTreeItem, NotionPageTreeMap } from './index'
+import type { DataSourceNotionPageMap } from '@/models/common'
+import { describe, expect, it } from 'vitest'
+import { recursivePushInParentDescendants } from './utils'
+
+const makePageEntry = (overrides: Partial<NotionPageTreeItem>): NotionPageTreeItem => ({
+  page_icon: null,
+  page_id: '',
+  page_name: '',
+  parent_id: '',
+  type: 'page',
+  is_bound: false,
+  children: new Set(),
+  descendants: new Set(),
+  depth: 0,
+  ancestors: [],
+  ...overrides,
+})
+
+describe('recursivePushInParentDescendants', () => {
+  it('should add child to parent descendants', () => {
+    const pagesMap = {
+      parent1: { page_id: 'parent1', parent_id: 'root', page_name: 'Parent' },
+      child1: { page_id: 'child1', parent_id: 'parent1', page_name: 'Child' },
+    } as unknown as DataSourceNotionPageMap
+
+    const listTreeMap: NotionPageTreeMap = {
+      child1: makePageEntry({ page_id: 'child1', parent_id: 'parent1', page_name: 'Child' }),
+    }
+
+    recursivePushInParentDescendants(pagesMap, listTreeMap, listTreeMap.child1, listTreeMap.child1)
+
+    expect(listTreeMap.parent1).toBeDefined()
+    expect(listTreeMap.parent1.children.has('child1')).toBe(true)
+    expect(listTreeMap.parent1.descendants.has('child1')).toBe(true)
+  })
+
+  it('should recursively populate ancestors for deeply nested items', () => {
+    const pagesMap = {
+      grandparent: { page_id: 'grandparent', parent_id: 'root', page_name: 'Grandparent' },
+      parent: { page_id: 'parent', parent_id: 'grandparent', page_name: 'Parent' },
+      child: { page_id: 'child', parent_id: 'parent', page_name: 'Child' },
+    } as unknown as DataSourceNotionPageMap
+
+    const listTreeMap: NotionPageTreeMap = {
+      parent: makePageEntry({ page_id: 'parent', parent_id: 'grandparent', page_name: 'Parent' }),
+      child: makePageEntry({ page_id: 'child', parent_id: 'parent', page_name: 'Child' }),
+    }
+
+    recursivePushInParentDescendants(pagesMap, listTreeMap, listTreeMap.child, listTreeMap.child)
+
+    expect(listTreeMap.child.depth).toBe(2)
+    expect(listTreeMap.child.ancestors).toContain('Grandparent')
+    expect(listTreeMap.child.ancestors).toContain('Parent')
+  })
+
+  it('should do nothing for root parent', () => {
+    const pagesMap = {
+      root_child: { page_id: 'root_child', parent_id: 'root', page_name: 'Root Child' },
+    } as unknown as DataSourceNotionPageMap
+
+    const listTreeMap: NotionPageTreeMap = {
+      root_child: makePageEntry({ page_id: 'root_child', parent_id: 'root', page_name: 'Root Child' }),
+    }
+
+    recursivePushInParentDescendants(pagesMap, listTreeMap, listTreeMap.root_child, listTreeMap.root_child)
+
+    // No new entries should be added since parent is root
+    expect(Object.keys(listTreeMap)).toEqual(['root_child'])
+  })
+
+  it('should handle missing parent_id gracefully', () => {
+    const pagesMap = {} as DataSourceNotionPageMap
+    const current = makePageEntry({ page_id: 'orphan', parent_id: undefined as unknown as string })
+    const listTreeMap: NotionPageTreeMap = { orphan: current }
+
+    // Should not throw
+    recursivePushInParentDescendants(pagesMap, listTreeMap, current, current)
+    expect(listTreeMap.orphan.depth).toBe(0)
+  })
+
+  it('should add to existing parent entry when parent already in tree', () => {
+    const pagesMap = {
+      parent: { page_id: 'parent', parent_id: 'root', page_name: 'Parent' },
+      child1: { page_id: 'child1', parent_id: 'parent', page_name: 'Child1' },
+      child2: { page_id: 'child2', parent_id: 'parent', page_name: 'Child2' },
+    } as unknown as DataSourceNotionPageMap
+
+    const listTreeMap: NotionPageTreeMap = {
+      parent: makePageEntry({ page_id: 'parent', parent_id: 'root', children: new Set(['child1']), descendants: new Set(['child1']), page_name: 'Parent' }),
+      child2: makePageEntry({ page_id: 'child2', parent_id: 'parent', page_name: 'Child2' }),
+    }
+
+    recursivePushInParentDescendants(pagesMap, listTreeMap, listTreeMap.child2, listTreeMap.child2)
+
+    expect(listTreeMap.parent.children.has('child2')).toBe(true)
+    expect(listTreeMap.parent.descendants.has('child2')).toBe(true)
+    expect(listTreeMap.parent.children.has('child1')).toBe(true)
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-documents/title.spec.tsx

@@ -0,0 +1,16 @@
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import Title from './title'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string, opts?: Record<string, unknown>) => `${key}:${(opts?.name as string) || ''}`,
+  }),
+}))
+
+describe('OnlineDocumentTitle', () => {
+  it('should render title with name prop', () => {
+    render(<Title name="Notion Workspace" />)
+    expect(screen.getByText('onlineDocument.pageSelectorTitle:Notion Workspace')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/header/breadcrumbs/bucket.spec.tsx

@@ -0,0 +1,60 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Bucket from './bucket'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({ t: (key: string) => key }),
+}))
+vi.mock('@/app/components/base/icons/src/public/knowledge/online-drive', () => ({
+  BucketsGray: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="buckets-gray" {...props} />,
+}))
+vi.mock('@/app/components/base/tooltip', () => ({
+  default: ({ children }: { children?: React.ReactNode }) => <div>{children}</div>,
+}))
+
+describe('Bucket', () => {
+  const defaultProps = {
+    bucketName: 'my-bucket',
+    handleBackToBucketList: vi.fn(),
+    handleClickBucketName: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render bucket name', () => {
+    render(<Bucket {...defaultProps} />)
+    expect(screen.getByText('my-bucket')).toBeInTheDocument()
+  })
+
+  it('should render bucket icon', () => {
+    render(<Bucket {...defaultProps} />)
+    expect(screen.getByTestId('buckets-gray')).toBeInTheDocument()
+  })
+
+  it('should call handleBackToBucketList on icon button click', () => {
+    render(<Bucket {...defaultProps} />)
+    const buttons = screen.getAllByRole('button')
+    fireEvent.click(buttons[0])
+    expect(defaultProps.handleBackToBucketList).toHaveBeenCalledOnce()
+  })
+
+  it('should call handleClickBucketName on name click', () => {
+    render(<Bucket {...defaultProps} />)
+    fireEvent.click(screen.getByText('my-bucket'))
+    expect(defaultProps.handleClickBucketName).toHaveBeenCalledOnce()
+  })
+
+  it('should not call handleClickBucketName when disabled', () => {
+    render(<Bucket {...defaultProps} disabled={true} />)
+    fireEvent.click(screen.getByText('my-bucket'))
+    expect(defaultProps.handleClickBucketName).not.toHaveBeenCalled()
+  })
+
+  it('should show separator by default', () => {
+    render(<Bucket {...defaultProps} />)
+    const separators = screen.getAllByText('/')
+    expect(separators.length).toBeGreaterThanOrEqual(2) // One after icon, one after name
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/header/breadcrumbs/drive.spec.tsx

@@ -0,0 +1,61 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Drive from './drive'
+
+describe('Drive', () => {
+  const defaultProps = {
+    breadcrumbs: [] as string[],
+    handleBackToRoot: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering: button text and separator visibility
+  describe('Rendering', () => {
+    it('should render "All Files" button text', () => {
+      render(<Drive {...defaultProps} />)
+
+      expect(screen.getByRole('button')).toHaveTextContent('datasetPipeline.onlineDrive.breadcrumbs.allFiles')
+    })
+
+    it('should show separator "/" when breadcrumbs has items', () => {
+      render(<Drive {...defaultProps} breadcrumbs={['Folder A']} />)
+
+      expect(screen.getByText('/')).toBeInTheDocument()
+    })
+
+    it('should hide separator when breadcrumbs is empty', () => {
+      render(<Drive {...defaultProps} breadcrumbs={[]} />)
+
+      expect(screen.queryByText('/')).not.toBeInTheDocument()
+    })
+  })
+
+  // Props: disabled state depends on breadcrumbs length
+  describe('Props', () => {
+    it('should disable button when breadcrumbs is empty', () => {
+      render(<Drive {...defaultProps} breadcrumbs={[]} />)
+
+      expect(screen.getByRole('button')).toBeDisabled()
+    })
+
+    it('should enable button when breadcrumbs has items', () => {
+      render(<Drive {...defaultProps} breadcrumbs={['Folder A', 'Folder B']} />)
+
+      expect(screen.getByRole('button')).not.toBeDisabled()
+    })
+  })
+
+  // User interactions: clicking the root button
+  describe('User Interactions', () => {
+    it('should call handleBackToRoot on click when enabled', () => {
+      render(<Drive {...defaultProps} breadcrumbs={['Folder A']} />)
+
+      fireEvent.click(screen.getByRole('button'))
+
+      expect(defaultProps.handleBackToRoot).toHaveBeenCalledOnce()
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/header/breadcrumbs/dropdown/item.spec.tsx

@@ -0,0 +1,44 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Item from './item'
+
+describe('Item', () => {
+  const defaultProps = {
+    name: 'Documents',
+    index: 2,
+    onBreadcrumbClick: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering: verify the breadcrumb name is displayed
+  describe('Rendering', () => {
+    it('should render breadcrumb name', () => {
+      render(<Item {...defaultProps} />)
+
+      expect(screen.getByText('Documents')).toBeInTheDocument()
+    })
+  })
+
+  // User interactions: clicking triggers callback with correct index
+  describe('User Interactions', () => {
+    it('should call onBreadcrumbClick with correct index on click', () => {
+      render(<Item {...defaultProps} />)
+
+      fireEvent.click(screen.getByText('Documents'))
+
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledOnce()
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(2)
+    })
+
+    it('should pass different index values correctly', () => {
+      render(<Item {...defaultProps} index={5} />)
+
+      fireEvent.click(screen.getByText('Documents'))
+
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(5)
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/header/breadcrumbs/dropdown/menu.spec.tsx

@@ -0,0 +1,79 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Menu from './menu'
+
+describe('Menu', () => {
+  const defaultProps = {
+    breadcrumbs: ['Folder A', 'Folder B', 'Folder C'],
+    startIndex: 1,
+    onBreadcrumbClick: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering: verify all breadcrumb items are displayed
+  describe('Rendering', () => {
+    it('should render all breadcrumb items', () => {
+      render(<Menu {...defaultProps} />)
+
+      expect(screen.getByText('Folder A')).toBeInTheDocument()
+      expect(screen.getByText('Folder B')).toBeInTheDocument()
+      expect(screen.getByText('Folder C')).toBeInTheDocument()
+    })
+
+    it('should render empty list when no breadcrumbs provided', () => {
+      const { container } = render(
+        <Menu breadcrumbs={[]} startIndex={0} onBreadcrumbClick={vi.fn()} />,
+      )
+
+      const menuContainer = container.firstElementChild
+      expect(menuContainer?.children).toHaveLength(0)
+    })
+  })
+
+  // Index mapping: startIndex offsets are applied correctly
+  describe('Index Mapping', () => {
+    it('should pass correct index (startIndex + offset) to each item', () => {
+      render(<Menu {...defaultProps} />)
+
+      fireEvent.click(screen.getByText('Folder A'))
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(1)
+
+      fireEvent.click(screen.getByText('Folder B'))
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(2)
+
+      fireEvent.click(screen.getByText('Folder C'))
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(3)
+    })
+
+    it('should offset from startIndex of zero', () => {
+      render(
+        <Menu
+          breadcrumbs={['First', 'Second']}
+          startIndex={0}
+          onBreadcrumbClick={defaultProps.onBreadcrumbClick}
+        />,
+      )
+
+      fireEvent.click(screen.getByText('First'))
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(0)
+
+      fireEvent.click(screen.getByText('Second'))
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(1)
+    })
+  })
+
+  // User interactions: clicking items triggers the callback
+  describe('User Interactions', () => {
+    it('should call onBreadcrumbClick with correct index when item clicked', () => {
+      render(<Menu {...defaultProps} />)
+
+      fireEvent.click(screen.getByText('Folder B'))
+
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledOnce()
+      expect(defaultProps.onBreadcrumbClick).toHaveBeenCalledWith(2)
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/header/breadcrumbs/item.spec.tsx

@@ -0,0 +1,48 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import BreadcrumbItem from './item'
+
+describe('BreadcrumbItem', () => {
+  const defaultProps = {
+    name: 'Documents',
+    index: 2,
+    handleClick: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render name', () => {
+    render(<BreadcrumbItem {...defaultProps} />)
+    expect(screen.getByText('Documents')).toBeInTheDocument()
+  })
+
+  it('should show separator by default', () => {
+    render(<BreadcrumbItem {...defaultProps} />)
+    expect(screen.getByText('/')).toBeInTheDocument()
+  })
+
+  it('should hide separator when showSeparator is false', () => {
+    render(<BreadcrumbItem {...defaultProps} showSeparator={false} />)
+    expect(screen.queryByText('/')).not.toBeInTheDocument()
+  })
+
+  it('should call handleClick with index on click', () => {
+    render(<BreadcrumbItem {...defaultProps} />)
+    fireEvent.click(screen.getByText('Documents'))
+    expect(defaultProps.handleClick).toHaveBeenCalledWith(2)
+  })
+
+  it('should not call handleClick when disabled', () => {
+    render(<BreadcrumbItem {...defaultProps} disabled={true} />)
+    fireEvent.click(screen.getByText('Documents'))
+    expect(defaultProps.handleClick).not.toHaveBeenCalled()
+  })
+
+  it('should apply active styling', () => {
+    render(<BreadcrumbItem {...defaultProps} isActive={true} />)
+    const btn = screen.getByRole('button')
+    expect(btn.className).toContain('system-sm-medium')
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/list/empty-folder.spec.tsx

@@ -1,38 +1,16 @@
-import { cleanup, render, screen } from '@testing-library/react'
-import { afterEach, describe, expect, it, vi } from 'vitest'
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
 import EmptyFolder from './empty-folder'
 
-// Mock react-i18next
 vi.mock('react-i18next', () => ({
   useTranslation: () => ({
     t: (key: string) => key,
   }),
 }))
 
-afterEach(() => {
-  cleanup()
-})
-
 describe('EmptyFolder', () => {
-  it('should render without crashing', () => {
+  it('should render empty folder message', () => {
     render(<EmptyFolder />)
     expect(screen.getByText('onlineDrive.emptyFolder')).toBeInTheDocument()
   })
-
-  it('should render the empty folder text', () => {
-    render(<EmptyFolder />)
-    expect(screen.getByText('onlineDrive.emptyFolder')).toBeInTheDocument()
-  })
-
-  it('should have proper styling classes', () => {
-    const { container } = render(<EmptyFolder />)
-    const wrapper = container.firstChild as HTMLElement
-    expect(wrapper).toHaveClass('flex')
-    expect(wrapper).toHaveClass('items-center')
-    expect(wrapper).toHaveClass('justify-center')
-  })
-
-  it('should be wrapped with React.memo', () => {
-    expect((EmptyFolder as unknown as { $$typeof: symbol }).$$typeof).toBe(Symbol.for('react.memo'))
-  })
 })

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/list/empty-search-result.spec.tsx

@@ -0,0 +1,34 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import EmptySearchResult from './empty-search-result'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({ t: (key: string) => key }),
+}))
+vi.mock('@/app/components/base/icons/src/vender/knowledge', () => ({
+  SearchMenu: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="search-icon" {...props} />,
+}))
+
+describe('EmptySearchResult', () => {
+  const onResetKeywords = vi.fn()
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render empty state message', () => {
+    render(<EmptySearchResult onResetKeywords={onResetKeywords} />)
+    expect(screen.getByText('onlineDrive.emptySearchResult')).toBeInTheDocument()
+  })
+
+  it('should render reset button', () => {
+    render(<EmptySearchResult onResetKeywords={onResetKeywords} />)
+    expect(screen.getByText('onlineDrive.resetKeywords')).toBeInTheDocument()
+  })
+
+  it('should call onResetKeywords when reset button clicked', () => {
+    render(<EmptySearchResult onResetKeywords={onResetKeywords} />)
+    fireEvent.click(screen.getByText('onlineDrive.resetKeywords'))
+    expect(onResetKeywords).toHaveBeenCalledOnce()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/list/file-icon.spec.tsx

@@ -0,0 +1,29 @@
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import { OnlineDriveFileType } from '@/models/pipeline'
+import FileIcon from './file-icon'
+
+vi.mock('@/app/components/base/file-uploader/file-type-icon', () => ({
+  default: ({ type }: { type: string }) => <span data-testid="file-type-icon">{type}</span>,
+}))
+vi.mock('@/app/components/base/icons/src/public/knowledge/online-drive', () => ({
+  BucketsBlue: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="bucket-icon" {...props} />,
+  Folder: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="folder-icon" {...props} />,
+}))
+
+describe('FileIcon', () => {
+  it('should render bucket icon for bucket type', () => {
+    render(<FileIcon type={OnlineDriveFileType.bucket} fileName="" />)
+    expect(screen.getByTestId('bucket-icon')).toBeInTheDocument()
+  })
+
+  it('should render folder icon for folder type', () => {
+    render(<FileIcon type={OnlineDriveFileType.folder} fileName="" />)
+    expect(screen.getByTestId('folder-icon')).toBeInTheDocument()
+  })
+
+  it('should render file type icon for file type', () => {
+    render(<FileIcon type={OnlineDriveFileType.file} fileName="doc.pdf" />)
+    expect(screen.getByTestId('file-type-icon')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/list/item.spec.tsx

@@ -0,0 +1,96 @@
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Item from './item'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
+}))
+
+vi.mock('@/app/components/base/checkbox', () => ({
+  default: ({ checked, onCheck, disabled }: { checked: boolean, onCheck: () => void, disabled?: boolean }) => (
+    <input type="checkbox" data-testid="checkbox" checked={checked} onChange={onCheck} disabled={disabled} />
+  ),
+}))
+
+vi.mock('@/app/components/base/radio/ui', () => ({
+  default: ({ isChecked, onCheck }: { isChecked: boolean, onCheck: () => void }) => (
+    <input type="radio" data-testid="radio" checked={isChecked} onChange={onCheck} />
+  ),
+}))
+
+vi.mock('@/app/components/base/tooltip', () => ({
+  default: ({ children, popupContent }: { children: React.ReactNode, popupContent: string }) => (
+    <div data-testid="tooltip" title={popupContent}>{children}</div>
+  ),
+}))
+
+vi.mock('./file-icon', () => ({
+  default: () => <span data-testid="file-icon" />,
+}))
+
+describe('Item', () => {
+  const makeFile = (type: string, name = 'test.pdf', size = 1024): OnlineDriveFile => ({
+    id: 'f-1',
+    name,
+    type: type as OnlineDriveFile['type'],
+    size,
+  })
+
+  const defaultProps = {
+    file: makeFile('file'),
+    isSelected: false,
+    onSelect: vi.fn(),
+    onOpen: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render file name', () => {
+    render(<Item {...defaultProps} />)
+    expect(screen.getByText('test.pdf')).toBeInTheDocument()
+  })
+
+  it('should render checkbox for file type in multiple choice mode', () => {
+    render(<Item {...defaultProps} />)
+    expect(screen.getByTestId('checkbox')).toBeInTheDocument()
+  })
+
+  it('should render radio for file type in single choice mode', () => {
+    render(<Item {...defaultProps} isMultipleChoice={false} />)
+    expect(screen.getByTestId('radio')).toBeInTheDocument()
+  })
+
+  it('should not render checkbox for bucket type', () => {
+    render(<Item {...defaultProps} file={makeFile('bucket', 'my-bucket')} />)
+    expect(screen.queryByTestId('checkbox')).not.toBeInTheDocument()
+  })
+
+  it('should call onOpen for folder click', () => {
+    const file = makeFile('folder', 'my-folder')
+    render(<Item {...defaultProps} file={file} />)
+    fireEvent.click(screen.getByText('my-folder'))
+    expect(defaultProps.onOpen).toHaveBeenCalledWith(file)
+  })
+
+  it('should call onSelect for file click', () => {
+    render(<Item {...defaultProps} />)
+    fireEvent.click(screen.getByText('test.pdf'))
+    expect(defaultProps.onSelect).toHaveBeenCalledWith(defaultProps.file)
+  })
+
+  it('should not call handlers when disabled', () => {
+    render(<Item {...defaultProps} disabled={true} />)
+    fireEvent.click(screen.getByText('test.pdf'))
+    expect(defaultProps.onSelect).not.toHaveBeenCalled()
+  })
+
+  it('should render file icon', () => {
+    render(<Item {...defaultProps} />)
+    expect(screen.getByTestId('file-icon')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/file-list/list/utils.spec.ts

@@ -0,0 +1,79 @@
+import { describe, expect, it } from 'vitest'
+import { FileAppearanceTypeEnum } from '@/app/components/base/file-uploader/types'
+import { getFileExtension, getFileType } from './utils'
+
+describe('getFileExtension', () => {
+  it('should return extension for normal file', () => {
+    expect(getFileExtension('test.pdf')).toBe('pdf')
+  })
+
+  it('should return lowercase extension', () => {
+    expect(getFileExtension('test.PDF')).toBe('pdf')
+  })
+
+  it('should return last extension for multiple dots', () => {
+    expect(getFileExtension('my.file.name.txt')).toBe('txt')
+  })
+
+  it('should return empty string for no extension', () => {
+    expect(getFileExtension('noext')).toBe('')
+  })
+
+  it('should return empty string for empty string', () => {
+    expect(getFileExtension('')).toBe('')
+  })
+
+  it('should return empty string for dotfile with no extension', () => {
+    expect(getFileExtension('.gitignore')).toBe('')
+  })
+})
+
+describe('getFileType', () => {
+  it('should return pdf for .pdf files', () => {
+    expect(getFileType('doc.pdf')).toBe(FileAppearanceTypeEnum.pdf)
+  })
+
+  it('should return markdown for .md files', () => {
+    expect(getFileType('readme.md')).toBe(FileAppearanceTypeEnum.markdown)
+  })
+
+  it('should return markdown for .mdx files', () => {
+    expect(getFileType('page.mdx')).toBe(FileAppearanceTypeEnum.markdown)
+  })
+
+  it('should return excel for .xlsx files', () => {
+    expect(getFileType('data.xlsx')).toBe(FileAppearanceTypeEnum.excel)
+  })
+
+  it('should return excel for .csv files', () => {
+    expect(getFileType('data.csv')).toBe(FileAppearanceTypeEnum.excel)
+  })
+
+  it('should return word for .docx files', () => {
+    expect(getFileType('doc.docx')).toBe(FileAppearanceTypeEnum.word)
+  })
+
+  it('should return ppt for .pptx files', () => {
+    expect(getFileType('slides.pptx')).toBe(FileAppearanceTypeEnum.ppt)
+  })
+
+  it('should return code for .html files', () => {
+    expect(getFileType('page.html')).toBe(FileAppearanceTypeEnum.code)
+  })
+
+  it('should return code for .json files', () => {
+    expect(getFileType('config.json')).toBe(FileAppearanceTypeEnum.code)
+  })
+
+  it('should return gif for .gif files', () => {
+    expect(getFileType('animation.gif')).toBe(FileAppearanceTypeEnum.gif)
+  })
+
+  it('should return custom for unknown extension', () => {
+    expect(getFileType('file.xyz')).toBe(FileAppearanceTypeEnum.custom)
+  })
+
+  it('should return custom for no extension', () => {
+    expect(getFileType('noext')).toBe(FileAppearanceTypeEnum.custom)
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/header.spec.tsx

@@ -0,0 +1,33 @@
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Header from './header'
+
+vi.mock('@remixicon/react', () => ({
+  RiBookOpenLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="book-icon" {...props} />,
+  RiEqualizer2Line: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="config-icon" {...props} />,
+}))
+
+describe('OnlineDriveHeader', () => {
+  const defaultProps = {
+    docTitle: 'S3 Guide',
+    docLink: 'https://docs.aws.com/s3',
+    onClickConfiguration: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render doc link with title', () => {
+    render(<Header {...defaultProps} />)
+    const link = screen.getByText('S3 Guide').closest('a')
+    expect(link).toHaveAttribute('href', 'https://docs.aws.com/s3')
+    expect(link).toHaveAttribute('target', '_blank')
+  })
+
+  it('should render book and config icons', () => {
+    render(<Header {...defaultProps} />)
+    expect(screen.getByTestId('book-icon')).toBeInTheDocument()
+    expect(screen.getByTestId('config-icon')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/online-drive/utils.spec.ts

@@ -0,0 +1,105 @@
+import type { OnlineDriveData } from '@/types/pipeline'
+import { describe, expect, it } from 'vitest'
+import { OnlineDriveFileType } from '@/models/pipeline'
+import { convertOnlineDriveData, isBucketListInitiation, isFile } from './utils'
+
+describe('online-drive utils', () => {
+  describe('isFile', () => {
+    it('should return true for file type', () => {
+      expect(isFile('file')).toBe(true)
+    })
+
+    it('should return false for folder type', () => {
+      expect(isFile('folder')).toBe(false)
+    })
+  })
+
+  describe('isBucketListInitiation', () => {
+    it('should return true when data has buckets and no prefix/bucket set', () => {
+      const data = [
+        { bucket: 'bucket-1', files: [], is_truncated: false, next_page_parameters: {} },
+        { bucket: 'bucket-2', files: [], is_truncated: false, next_page_parameters: {} },
+      ] as OnlineDriveData[]
+
+      expect(isBucketListInitiation(data, [], '')).toBe(true)
+    })
+
+    it('should return false when bucket is already set', () => {
+      const data = [
+        { bucket: 'bucket-1', files: [], is_truncated: false, next_page_parameters: {} },
+      ] as OnlineDriveData[]
+
+      expect(isBucketListInitiation(data, [], 'bucket-1')).toBe(false)
+    })
+
+    it('should return false when prefix is set', () => {
+      const data = [
+        { bucket: 'bucket-1', files: [], is_truncated: false, next_page_parameters: {} },
+      ] as OnlineDriveData[]
+
+      expect(isBucketListInitiation(data, ['folder/'], '')).toBe(false)
+    })
+
+    it('should return false when single bucket has files', () => {
+      const data = [
+        {
+          bucket: 'bucket-1',
+          files: [{ id: 'f1', name: 'test.txt', size: 100, type: 'file' as const }],
+          is_truncated: false,
+          next_page_parameters: {},
+        },
+      ] as OnlineDriveData[]
+
+      expect(isBucketListInitiation(data, [], '')).toBe(false)
+    })
+  })
+
+  describe('convertOnlineDriveData', () => {
+    it('should return empty result for empty data', () => {
+      const result = convertOnlineDriveData([], [], '')
+      expect(result.fileList).toEqual([])
+      expect(result.isTruncated).toBe(false)
+      expect(result.hasBucket).toBe(false)
+    })
+
+    it('should convert bucket list initiation to bucket items', () => {
+      const data = [
+        { bucket: 'bucket-1', files: [], is_truncated: false, next_page_parameters: {} },
+        { bucket: 'bucket-2', files: [], is_truncated: false, next_page_parameters: {} },
+      ] as OnlineDriveData[]
+
+      const result = convertOnlineDriveData(data, [], '')
+      expect(result.fileList).toHaveLength(2)
+      expect(result.fileList[0]).toEqual({
+        id: 'bucket-1',
+        name: 'bucket-1',
+        type: OnlineDriveFileType.bucket,
+      })
+      expect(result.hasBucket).toBe(true)
+    })
+
+    it('should convert files when not bucket list', () => {
+      const data = [
+        {
+          bucket: 'bucket-1',
+          files: [
+            { id: 'f1', name: 'test.txt', size: 100, type: 'file' as const },
+            { id: 'f2', name: 'folder', size: 0, type: 'folder' as const },
+          ],
+          is_truncated: true,
+          next_page_parameters: { token: 'next' },
+        },
+      ] as OnlineDriveData[]
+
+      const result = convertOnlineDriveData(data, [], 'bucket-1')
+      expect(result.fileList).toHaveLength(2)
+      expect(result.fileList[0].type).toBe(OnlineDriveFileType.file)
+      expect(result.fileList[0].size).toBe(100)
+      expect(result.fileList[1].type).toBe(OnlineDriveFileType.folder)
+      expect(result.fileList[1].size).toBeUndefined()
+      expect(result.isTruncated).toBe(true)
+      expect(result.nextPageParameters).toEqual({ token: 'next' })
+      expect(result.hasBucket).toBe(true)
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/store/index.spec.ts

@@ -0,0 +1,96 @@
+import type { FileItem } from '@/models/datasets'
+import { render, renderHook } from '@testing-library/react'
+import * as React from 'react'
+import { describe, expect, it } from 'vitest'
+import { createDataSourceStore, useDataSourceStore, useDataSourceStoreWithSelector } from './'
+import DataSourceProvider from './provider'
+
+describe('createDataSourceStore', () => {
+  it('should create a store with all slices combined', () => {
+    const store = createDataSourceStore()
+    const state = store.getState()
+
+    // Common slice
+    expect(state.currentCredentialId).toBe('')
+    expect(typeof state.setCurrentCredentialId).toBe('function')
+
+    // LocalFile slice
+    expect(state.localFileList).toEqual([])
+    expect(typeof state.setLocalFileList).toBe('function')
+
+    // OnlineDocument slice
+    expect(state.documentsData).toEqual([])
+    expect(typeof state.setDocumentsData).toBe('function')
+
+    // WebsiteCrawl slice
+    expect(state.websitePages).toEqual([])
+    expect(typeof state.setWebsitePages).toBe('function')
+
+    // OnlineDrive slice
+    expect(state.breadcrumbs).toEqual([])
+    expect(typeof state.setBreadcrumbs).toBe('function')
+  })
+
+  it('should allow cross-slice state updates', () => {
+    const store = createDataSourceStore()
+
+    store.getState().setCurrentCredentialId('cred-1')
+    store.getState().setLocalFileList([{ file: { id: 'f1' } }] as unknown as FileItem[])
+
+    expect(store.getState().currentCredentialId).toBe('cred-1')
+    expect(store.getState().localFileList).toHaveLength(1)
+  })
+
+  it('should create independent store instances', () => {
+    const store1 = createDataSourceStore()
+    const store2 = createDataSourceStore()
+
+    store1.getState().setCurrentCredentialId('cred-1')
+    expect(store2.getState().currentCredentialId).toBe('')
+  })
+})
+
+describe('useDataSourceStoreWithSelector', () => {
+  it('should throw when used outside provider', () => {
+    expect(() => {
+      renderHook(() => useDataSourceStoreWithSelector(s => s.currentCredentialId))
+    }).toThrow('Missing DataSourceContext.Provider in the tree')
+  })
+
+  it('should return selected state when used inside provider', () => {
+    const wrapper = ({ children }: { children: React.ReactNode }) =>
+      React.createElement(DataSourceProvider, null, children)
+    const { result } = renderHook(
+      () => useDataSourceStoreWithSelector(s => s.currentCredentialId),
+      { wrapper },
+    )
+    expect(result.current).toBe('')
+  })
+})
+
+describe('useDataSourceStore', () => {
+  it('should throw when used outside provider', () => {
+    expect(() => {
+      renderHook(() => useDataSourceStore())
+    }).toThrow('Missing DataSourceContext.Provider in the tree')
+  })
+
+  it('should return store when used inside provider', () => {
+    const wrapper = ({ children }: { children: React.ReactNode }) =>
+      React.createElement(DataSourceProvider, null, children)
+    const { result } = renderHook(
+      () => useDataSourceStore(),
+      { wrapper },
+    )
+    expect(result.current).toBeDefined()
+    expect(typeof result.current.getState).toBe('function')
+  })
+})
+
+describe('DataSourceProvider', () => {
+  it('should render children', () => {
+    const child = React.createElement('div', null, 'Child Content')
+    const { getByText } = render(React.createElement(DataSourceProvider, null, child))
+    expect(getByText('Child Content')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/store/provider.spec.tsx

@@ -0,0 +1,89 @@
+import { render, screen } from '@testing-library/react'
+import { useContext } from 'react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import DataSourceProvider, { DataSourceContext } from './provider'
+
+const mockStore = { getState: vi.fn(), setState: vi.fn(), subscribe: vi.fn() }
+
+vi.mock('./', () => ({
+  createDataSourceStore: () => mockStore,
+}))
+
+// Test consumer component that reads from context
+function ContextConsumer() {
+  const store = useContext(DataSourceContext)
+  return (
+    <div data-testid="context-value" data-has-store={store !== null}>
+      {store ? 'has-store' : 'no-store'}
+    </div>
+  )
+}
+
+describe('DataSourceProvider', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering: verifies children are passed through
+  describe('Rendering', () => {
+    it('should render children', () => {
+      render(
+        <DataSourceProvider>
+          <span data-testid="child">Hello</span>
+        </DataSourceProvider>,
+      )
+
+      expect(screen.getByTestId('child')).toBeInTheDocument()
+      expect(screen.getByText('Hello')).toBeInTheDocument()
+    })
+  })
+
+  // Context: verifies the store is provided to consumers
+  describe('Context', () => {
+    it('should provide store value to context consumers', () => {
+      render(
+        <DataSourceProvider>
+          <ContextConsumer />
+        </DataSourceProvider>,
+      )
+
+      expect(screen.getByTestId('context-value')).toHaveTextContent('has-store')
+      expect(screen.getByTestId('context-value')).toHaveAttribute('data-has-store', 'true')
+    })
+
+    it('should provide null when no provider wraps the consumer', () => {
+      render(<ContextConsumer />)
+
+      expect(screen.getByTestId('context-value')).toHaveTextContent('no-store')
+      expect(screen.getByTestId('context-value')).toHaveAttribute('data-has-store', 'false')
+    })
+  })
+
+  // Stability: verifies the store reference is stable across re-renders
+  describe('Store Stability', () => {
+    it('should reuse same store on re-render (stable reference)', () => {
+      const storeValues: Array<typeof mockStore | null> = []
+
+      function StoreCapture() {
+        const store = useContext(DataSourceContext)
+        storeValues.push(store as typeof mockStore | null)
+        return null
+      }
+
+      const { rerender } = render(
+        <DataSourceProvider>
+          <StoreCapture />
+        </DataSourceProvider>,
+      )
+
+      rerender(
+        <DataSourceProvider>
+          <StoreCapture />
+        </DataSourceProvider>,
+      )
+
+      expect(storeValues).toHaveLength(2)
+      expect(storeValues[0]).toBe(storeValues[1])
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/store/slices/common.spec.ts

@@ -0,0 +1,29 @@
+import type { CommonShape } from './common'
+import { describe, expect, it } from 'vitest'
+import { createStore } from 'zustand'
+import { createCommonSlice } from './common'
+
+const createTestStore = () => createStore<CommonShape>((...args) => createCommonSlice(...args))
+
+describe('createCommonSlice', () => {
+  it('should initialize with default values', () => {
+    const state = createTestStore().getState()
+
+    expect(state.currentCredentialId).toBe('')
+    expect(state.currentNodeIdRef.current).toBe('')
+    expect(state.currentCredentialIdRef.current).toBe('')
+  })
+
+  it('should update currentCredentialId', () => {
+    const store = createTestStore()
+    store.getState().setCurrentCredentialId('cred-123')
+    expect(store.getState().currentCredentialId).toBe('cred-123')
+  })
+
+  it('should update currentCredentialId multiple times', () => {
+    const store = createTestStore()
+    store.getState().setCurrentCredentialId('cred-1')
+    store.getState().setCurrentCredentialId('cred-2')
+    expect(store.getState().currentCredentialId).toBe('cred-2')
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/store/slices/local-file.spec.ts

@@ -0,0 +1,49 @@
+import type { LocalFileSliceShape } from './local-file'
+import type { CustomFile as File, FileItem } from '@/models/datasets'
+import { describe, expect, it } from 'vitest'
+import { createStore } from 'zustand'
+import { createLocalFileSlice } from './local-file'
+
+const createTestStore = () => createStore<LocalFileSliceShape>((...args) => createLocalFileSlice(...args))
+
+describe('createLocalFileSlice', () => {
+  it('should initialize with default values', () => {
+    const state = createTestStore().getState()
+
+    expect(state.localFileList).toEqual([])
+    expect(state.currentLocalFile).toBeUndefined()
+    expect(state.previewLocalFileRef.current).toBeUndefined()
+  })
+
+  it('should set local file list and update preview ref to first file', () => {
+    const store = createTestStore()
+    const files = [
+      { file: { id: 'f1', name: 'a.pdf' } },
+      { file: { id: 'f2', name: 'b.pdf' } },
+    ] as unknown as FileItem[]
+
+    store.getState().setLocalFileList(files)
+    expect(store.getState().localFileList).toEqual(files)
+    expect(store.getState().previewLocalFileRef.current).toEqual({ id: 'f1', name: 'a.pdf' })
+  })
+
+  it('should set preview ref to undefined for empty file list', () => {
+    const store = createTestStore()
+    store.getState().setLocalFileList([])
+    expect(store.getState().previewLocalFileRef.current).toBeUndefined()
+  })
+
+  it('should set current local file', () => {
+    const store = createTestStore()
+    const file = { id: 'f1', name: 'test.pdf' } as unknown as File
+    store.getState().setCurrentLocalFile(file)
+    expect(store.getState().currentLocalFile).toEqual(file)
+  })
+
+  it('should clear current local file with undefined', () => {
+    const store = createTestStore()
+    store.getState().setCurrentLocalFile({ id: 'f1' } as unknown as File)
+    store.getState().setCurrentLocalFile(undefined)
+    expect(store.getState().currentLocalFile).toBeUndefined()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/store/slices/online-document.spec.ts

@@ -0,0 +1,55 @@
+import type { OnlineDocumentSliceShape } from './online-document'
+import type { DataSourceNotionWorkspace, NotionPage } from '@/models/common'
+import { describe, expect, it } from 'vitest'
+import { createStore } from 'zustand'
+import { createOnlineDocumentSlice } from './online-document'
+
+const createTestStore = () => createStore<OnlineDocumentSliceShape>((...args) => createOnlineDocumentSlice(...args))
+
+describe('createOnlineDocumentSlice', () => {
+  it('should initialize with default values', () => {
+    const state = createTestStore().getState()
+
+    expect(state.documentsData).toEqual([])
+    expect(state.searchValue).toBe('')
+    expect(state.onlineDocuments).toEqual([])
+    expect(state.currentDocument).toBeUndefined()
+    expect(state.selectedPagesId).toEqual(new Set())
+    expect(state.previewOnlineDocumentRef.current).toBeUndefined()
+  })
+
+  it('should set documents data', () => {
+    const store = createTestStore()
+    const data = [{ workspace_id: 'w1', pages: [] }] as unknown as DataSourceNotionWorkspace[]
+    store.getState().setDocumentsData(data)
+    expect(store.getState().documentsData).toEqual(data)
+  })
+
+  it('should set search value', () => {
+    const store = createTestStore()
+    store.getState().setSearchValue('hello')
+    expect(store.getState().searchValue).toBe('hello')
+  })
+
+  it('should set online documents and update preview ref', () => {
+    const store = createTestStore()
+    const pages = [{ page_id: 'p1' }, { page_id: 'p2' }] as unknown as NotionPage[]
+    store.getState().setOnlineDocuments(pages)
+    expect(store.getState().onlineDocuments).toEqual(pages)
+    expect(store.getState().previewOnlineDocumentRef.current).toEqual({ page_id: 'p1' })
+  })
+
+  it('should set current document', () => {
+    const store = createTestStore()
+    const doc = { page_id: 'p1' } as unknown as NotionPage
+    store.getState().setCurrentDocument(doc)
+    expect(store.getState().currentDocument).toEqual(doc)
+  })
+
+  it('should set selected pages id', () => {
+    const store = createTestStore()
+    const ids = new Set(['p1', 'p2'])
+    store.getState().setSelectedPagesId(ids)
+    expect(store.getState().selectedPagesId).toEqual(ids)
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/store/slices/online-drive.spec.ts

@@ -0,0 +1,79 @@
+import type { OnlineDriveSliceShape } from './online-drive'
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { describe, expect, it } from 'vitest'
+import { createStore } from 'zustand'
+import { createOnlineDriveSlice } from './online-drive'
+
+const createTestStore = () => createStore<OnlineDriveSliceShape>((...args) => createOnlineDriveSlice(...args))
+
+describe('createOnlineDriveSlice', () => {
+  it('should initialize with default values', () => {
+    const state = createTestStore().getState()
+
+    expect(state.breadcrumbs).toEqual([])
+    expect(state.prefix).toEqual([])
+    expect(state.keywords).toBe('')
+    expect(state.selectedFileIds).toEqual([])
+    expect(state.onlineDriveFileList).toEqual([])
+    expect(state.bucket).toBe('')
+    expect(state.nextPageParameters).toEqual({})
+    expect(state.isTruncated.current).toBe(false)
+    expect(state.previewOnlineDriveFileRef.current).toBeUndefined()
+    expect(state.hasBucket).toBe(false)
+  })
+
+  it('should set breadcrumbs', () => {
+    const store = createTestStore()
+    store.getState().setBreadcrumbs(['root', 'folder'])
+    expect(store.getState().breadcrumbs).toEqual(['root', 'folder'])
+  })
+
+  it('should set prefix', () => {
+    const store = createTestStore()
+    store.getState().setPrefix(['a', 'b'])
+    expect(store.getState().prefix).toEqual(['a', 'b'])
+  })
+
+  it('should set keywords', () => {
+    const store = createTestStore()
+    store.getState().setKeywords('search term')
+    expect(store.getState().keywords).toBe('search term')
+  })
+
+  it('should set selected file ids and update preview ref', () => {
+    const store = createTestStore()
+    const files = [
+      { id: 'file-1', name: 'a.pdf', type: 'file' },
+      { id: 'file-2', name: 'b.pdf', type: 'file' },
+    ] as unknown as OnlineDriveFile[]
+    store.getState().setOnlineDriveFileList(files)
+    store.getState().setSelectedFileIds(['file-1'])
+
+    expect(store.getState().selectedFileIds).toEqual(['file-1'])
+    expect(store.getState().previewOnlineDriveFileRef.current).toEqual(files[0])
+  })
+
+  it('should set preview ref to undefined when selected id not found', () => {
+    const store = createTestStore()
+    store.getState().setSelectedFileIds(['non-existent'])
+    expect(store.getState().previewOnlineDriveFileRef.current).toBeUndefined()
+  })
+
+  it('should set bucket', () => {
+    const store = createTestStore()
+    store.getState().setBucket('my-bucket')
+    expect(store.getState().bucket).toBe('my-bucket')
+  })
+
+  it('should set next page parameters', () => {
+    const store = createTestStore()
+    store.getState().setNextPageParameters({ cursor: 'abc' })
+    expect(store.getState().nextPageParameters).toEqual({ cursor: 'abc' })
+  })
+
+  it('should set hasBucket', () => {
+    const store = createTestStore()
+    store.getState().setHasBucket(true)
+    expect(store.getState().hasBucket).toBe(true)
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/store/slices/website-crawl.spec.ts

@@ -0,0 +1,65 @@
+import type { WebsiteCrawlSliceShape } from './website-crawl'
+import type { CrawlResult, CrawlResultItem } from '@/models/datasets'
+import { describe, expect, it } from 'vitest'
+import { createStore } from 'zustand'
+import { CrawlStep } from '@/models/datasets'
+import { createWebsiteCrawlSlice } from './website-crawl'
+
+const createTestStore = () => createStore<WebsiteCrawlSliceShape>((...args) => createWebsiteCrawlSlice(...args))
+
+describe('createWebsiteCrawlSlice', () => {
+  it('should initialize with default values', () => {
+    const state = createTestStore().getState()
+
+    expect(state.websitePages).toEqual([])
+    expect(state.currentWebsite).toBeUndefined()
+    expect(state.crawlResult).toBeUndefined()
+    expect(state.step).toBe(CrawlStep.init)
+    expect(state.previewIndex).toBe(-1)
+    expect(state.previewWebsitePageRef.current).toBeUndefined()
+  })
+
+  it('should set website pages and update preview ref', () => {
+    const store = createTestStore()
+    const pages = [
+      { title: 'Page 1', source_url: 'https://a.com' },
+      { title: 'Page 2', source_url: 'https://b.com' },
+    ] as unknown as CrawlResultItem[]
+    store.getState().setWebsitePages(pages)
+    expect(store.getState().websitePages).toEqual(pages)
+    expect(store.getState().previewWebsitePageRef.current).toEqual(pages[0])
+  })
+
+  it('should set current website', () => {
+    const store = createTestStore()
+    const website = { title: 'Page 1' } as unknown as CrawlResultItem
+    store.getState().setCurrentWebsite(website)
+    expect(store.getState().currentWebsite).toEqual(website)
+  })
+
+  it('should set crawl result', () => {
+    const store = createTestStore()
+    const result = { data: { count: 5 } } as unknown as CrawlResult
+    store.getState().setCrawlResult(result)
+    expect(store.getState().crawlResult).toEqual(result)
+  })
+
+  it('should set step', () => {
+    const store = createTestStore()
+    store.getState().setStep(CrawlStep.running)
+    expect(store.getState().step).toBe(CrawlStep.running)
+  })
+
+  it('should set preview index', () => {
+    const store = createTestStore()
+    store.getState().setPreviewIndex(3)
+    expect(store.getState().previewIndex).toBe(3)
+  })
+
+  it('should clear current website with undefined', () => {
+    const store = createTestStore()
+    store.getState().setCurrentWebsite({ title: 'X' } as unknown as CrawlResultItem)
+    store.getState().setCurrentWebsite(undefined)
+    expect(store.getState().currentWebsite).toBeUndefined()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/website-crawl/base/checkbox-with-label.spec.tsx

@@ -0,0 +1,50 @@
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import CheckboxWithLabel from './checkbox-with-label'
+
+vi.mock('@/app/components/base/checkbox', () => ({
+  default: ({ checked, onCheck }: { checked: boolean, onCheck: () => void }) => (
+    <input type="checkbox" data-testid="checkbox" checked={checked} onChange={onCheck} />
+  ),
+}))
+
+vi.mock('@/app/components/base/tooltip', () => ({
+  default: ({ popupContent }: { popupContent: string }) => <div data-testid="tooltip">{popupContent}</div>,
+}))
+
+describe('CheckboxWithLabel', () => {
+  const defaultProps = {
+    isChecked: false,
+    onChange: vi.fn(),
+    label: 'Test Label',
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render label text', () => {
+    render(<CheckboxWithLabel {...defaultProps} />)
+    expect(screen.getByText('Test Label')).toBeInTheDocument()
+  })
+
+  it('should render checkbox', () => {
+    render(<CheckboxWithLabel {...defaultProps} />)
+    expect(screen.getByTestId('checkbox')).toBeInTheDocument()
+  })
+
+  it('should render tooltip when provided', () => {
+    render(<CheckboxWithLabel {...defaultProps} tooltip="Help text" />)
+    expect(screen.getByTestId('tooltip')).toBeInTheDocument()
+  })
+
+  it('should not render tooltip when not provided', () => {
+    render(<CheckboxWithLabel {...defaultProps} />)
+    expect(screen.queryByTestId('tooltip')).not.toBeInTheDocument()
+  })
+
+  it('should apply custom className', () => {
+    const { container } = render(<CheckboxWithLabel {...defaultProps} className="custom-cls" />)
+    expect(container.querySelector('.custom-cls')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/website-crawl/base/crawled-result-item.spec.tsx

@@ -0,0 +1,75 @@
+import type { CrawlResultItem as CrawlResultItemType } from '@/models/datasets'
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import CrawledResultItem from './crawled-result-item'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
+}))
+
+vi.mock('@/app/components/base/button', () => ({
+  default: ({ children, onClick }: { children: React.ReactNode, onClick: () => void }) => (
+    <button data-testid="preview-button" onClick={onClick}>{children}</button>
+  ),
+}))
+
+vi.mock('@/app/components/base/checkbox', () => ({
+  default: ({ checked, onCheck }: { checked: boolean, onCheck: () => void }) => (
+    <input type="checkbox" data-testid="checkbox" checked={checked} onChange={onCheck} />
+  ),
+}))
+
+vi.mock('@/app/components/base/radio/ui', () => ({
+  default: ({ isChecked, onCheck }: { isChecked: boolean, onCheck: () => void }) => (
+    <input type="radio" data-testid="radio" checked={isChecked} onChange={onCheck} />
+  ),
+}))
+
+describe('CrawledResultItem', () => {
+  const defaultProps = {
+    payload: {
+      title: 'Test Page',
+      source_url: 'https://example.com/page',
+      markdown: '',
+      description: '',
+    } satisfies CrawlResultItemType,
+    isChecked: false,
+    onCheckChange: vi.fn(),
+    isPreview: false,
+    showPreview: true,
+    onPreview: vi.fn(),
+    isMultipleChoice: true,
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should render title and URL', () => {
+    render(<CrawledResultItem {...defaultProps} />)
+    expect(screen.getByText('Test Page')).toBeInTheDocument()
+    expect(screen.getByText('https://example.com/page')).toBeInTheDocument()
+  })
+
+  it('should render checkbox in multiple choice mode', () => {
+    render(<CrawledResultItem {...defaultProps} />)
+    expect(screen.getByTestId('checkbox')).toBeInTheDocument()
+  })
+
+  it('should render radio in single choice mode', () => {
+    render(<CrawledResultItem {...defaultProps} isMultipleChoice={false} />)
+    expect(screen.getByTestId('radio')).toBeInTheDocument()
+  })
+
+  it('should show preview button when showPreview is true', () => {
+    render(<CrawledResultItem {...defaultProps} />)
+    expect(screen.getByTestId('preview-button')).toBeInTheDocument()
+  })
+
+  it('should not show preview button when showPreview is false', () => {
+    render(<CrawledResultItem {...defaultProps} showPreview={false} />)
+    expect(screen.queryByTestId('preview-button')).not.toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/website-crawl/base/crawled-result.spec.tsx

@@ -0,0 +1,218 @@
+import type { CrawlResultItem } from '@/models/datasets'
+import { fireEvent, render, screen } from '@testing-library/react'
+
+import CrawledResult from './crawled-result'
+
+vi.mock('./checkbox-with-label', () => ({
+  default: ({ isChecked, onChange, label }: { isChecked: boolean, onChange: () => void, label: string }) => (
+    <label>
+      <input
+        type="checkbox"
+        checked={isChecked}
+        onChange={onChange}
+        data-testid="check-all-checkbox"
+      />
+      {label}
+    </label>
+  ),
+}))
+
+vi.mock('./crawled-result-item', () => ({
+  default: ({
+    payload,
+    isChecked,
+    onCheckChange,
+    onPreview,
+  }: {
+    payload: CrawlResultItem
+    isChecked: boolean
+    onCheckChange: (checked: boolean) => void
+    onPreview: () => void
+  }) => (
+    <div data-testid={`crawled-item-${payload.source_url}`}>
+      <span data-testid="item-url">{payload.source_url}</span>
+      <button data-testid={`check-${payload.source_url}`} onClick={() => onCheckChange(!isChecked)}>
+        {isChecked ? 'uncheck' : 'check'}
+      </button>
+      <button data-testid={`preview-${payload.source_url}`} onClick={onPreview}>
+        preview
+      </button>
+    </div>
+  ),
+}))
+
+const createItem = (url: string): CrawlResultItem => ({
+  source_url: url,
+  title: `Title for ${url}`,
+  markdown: `# ${url}`,
+  description: `Desc for ${url}`,
+})
+
+const defaultList: CrawlResultItem[] = [
+  createItem('https://example.com/a'),
+  createItem('https://example.com/b'),
+  createItem('https://example.com/c'),
+]
+
+describe('CrawledResult', () => {
+  const defaultProps = {
+    list: defaultList,
+    checkedList: [] as CrawlResultItem[],
+    onSelectedChange: vi.fn(),
+    usedTime: 12.345,
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering tests
+  describe('Rendering', () => {
+    it('should render scrap time info with correct total and time', () => {
+      render(<CrawledResult {...defaultProps} />)
+
+      expect(
+        screen.getByText(/scrapTimeInfo/),
+      ).toBeInTheDocument()
+      // The global i18n mock serialises params, so verify total and time appear
+      expect(screen.getByText(/"total":3/)).toBeInTheDocument()
+      expect(screen.getByText(/"time":"12.3"/)).toBeInTheDocument()
+    })
+
+    it('should render all items from list', () => {
+      render(<CrawledResult {...defaultProps} />)
+
+      for (const item of defaultList) {
+        expect(screen.getByTestId(`crawled-item-${item.source_url}`)).toBeInTheDocument()
+      }
+    })
+
+    it('should apply custom className', () => {
+      const { container } = render(
+        <CrawledResult {...defaultProps} className="my-custom-class" />,
+      )
+
+      expect(container.firstChild).toHaveClass('my-custom-class')
+    })
+  })
+
+  // Check-all checkbox visibility
+  describe('Check All Checkbox', () => {
+    it('should show check-all checkbox in multiple choice mode', () => {
+      render(<CrawledResult {...defaultProps} isMultipleChoice={true} />)
+
+      expect(screen.getByTestId('check-all-checkbox')).toBeInTheDocument()
+    })
+
+    it('should hide check-all checkbox in single choice mode', () => {
+      render(<CrawledResult {...defaultProps} isMultipleChoice={false} />)
+
+      expect(screen.queryByTestId('check-all-checkbox')).not.toBeInTheDocument()
+    })
+  })
+
+  // Toggle all items
+  describe('Toggle All', () => {
+    it('should select all when not all checked', () => {
+      const onSelectedChange = vi.fn()
+      render(
+        <CrawledResult
+          {...defaultProps}
+          checkedList={[defaultList[0]]}
+          onSelectedChange={onSelectedChange}
+        />,
+      )
+
+      fireEvent.click(screen.getByTestId('check-all-checkbox'))
+
+      expect(onSelectedChange).toHaveBeenCalledWith(defaultList)
+    })
+
+    it('should deselect all when all checked', () => {
+      const onSelectedChange = vi.fn()
+      render(
+        <CrawledResult
+          {...defaultProps}
+          checkedList={[...defaultList]}
+          onSelectedChange={onSelectedChange}
+        />,
+      )
+
+      fireEvent.click(screen.getByTestId('check-all-checkbox'))
+
+      expect(onSelectedChange).toHaveBeenCalledWith([])
+    })
+  })
+
+  // Individual item check
+  describe('Individual Item Check', () => {
+    it('should add item to selection in multiple choice mode', () => {
+      const onSelectedChange = vi.fn()
+      render(
+        <CrawledResult
+          {...defaultProps}
+          checkedList={[defaultList[0]]}
+          onSelectedChange={onSelectedChange}
+          isMultipleChoice={true}
+        />,
+      )
+
+      // Click check on unchecked second item
+      fireEvent.click(screen.getByTestId(`check-${defaultList[1].source_url}`))
+
+      expect(onSelectedChange).toHaveBeenCalledWith([defaultList[0], defaultList[1]])
+    })
+
+    it('should replace selection in single choice mode', () => {
+      const onSelectedChange = vi.fn()
+      render(
+        <CrawledResult
+          {...defaultProps}
+          checkedList={[defaultList[0]]}
+          onSelectedChange={onSelectedChange}
+          isMultipleChoice={false}
+        />,
+      )
+
+      // Click check on unchecked second item
+      fireEvent.click(screen.getByTestId(`check-${defaultList[1].source_url}`))
+
+      expect(onSelectedChange).toHaveBeenCalledWith([defaultList[1]])
+    })
+
+    it('should remove item from selection when unchecked', () => {
+      const onSelectedChange = vi.fn()
+      render(
+        <CrawledResult
+          {...defaultProps}
+          checkedList={[defaultList[0], defaultList[1]]}
+          onSelectedChange={onSelectedChange}
+          isMultipleChoice={true}
+        />,
+      )
+
+      // Click uncheck on checked first item
+      fireEvent.click(screen.getByTestId(`check-${defaultList[0].source_url}`))
+
+      expect(onSelectedChange).toHaveBeenCalledWith([defaultList[1]])
+    })
+  })
+
+  // Preview
+  describe('Preview', () => {
+    it('should call onPreview with correct item and index', () => {
+      const onPreview = vi.fn()
+      render(
+        <CrawledResult
+          {...defaultProps}
+          onPreview={onPreview}
+          showPreview={true}
+        />,
+      )
+
+      fireEvent.click(screen.getByTestId(`preview-${defaultList[1].source_url}`))
+
+      expect(onPreview).toHaveBeenCalledWith(defaultList[1], 1)
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/website-crawl/base/crawling.spec.tsx

@@ -0,0 +1,27 @@
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import Crawling from './crawling'
+
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
+}))
+
+describe('Crawling', () => {
+  it('should render crawl progress', () => {
+    render(<Crawling crawledNum={5} totalNum={10} />)
+    expect(screen.getByText(/5/)).toBeInTheDocument()
+    expect(screen.getByText(/10/)).toBeInTheDocument()
+  })
+
+  it('should render total page scraped label', () => {
+    render(<Crawling crawledNum={0} totalNum={0} />)
+    expect(screen.getByText(/stepOne\.website\.totalPageScraped/)).toBeInTheDocument()
+  })
+
+  it('should apply custom className', () => {
+    const { container } = render(<Crawling crawledNum={1} totalNum={5} className="custom" />)
+    expect(container.querySelector('.custom')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/data-source/website-crawl/base/error-message.spec.tsx

@@ -0,0 +1,35 @@
+import { render, screen } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+import ErrorMessage from './error-message'
+
+vi.mock('@remixicon/react', () => ({
+  RiErrorWarningFill: () => <span data-testid="error-icon" />,
+}))
+
+describe('ErrorMessage', () => {
+  it('should render title', () => {
+    render(<ErrorMessage title="Something went wrong" />)
+    expect(screen.getByText('Something went wrong')).toBeInTheDocument()
+  })
+
+  it('should render error icon', () => {
+    render(<ErrorMessage title="Error" />)
+    expect(screen.getByTestId('error-icon')).toBeInTheDocument()
+  })
+
+  it('should render error message when provided', () => {
+    render(<ErrorMessage title="Error" errorMsg="Detailed error info" />)
+    expect(screen.getByText('Detailed error info')).toBeInTheDocument()
+  })
+
+  it('should not render error message when not provided', () => {
+    const { container } = render(<ErrorMessage title="Error" />)
+    const textElements = container.querySelectorAll('.system-xs-regular')
+    expect(textElements).toHaveLength(0)
+  })
+
+  it('should apply custom className', () => {
+    const { container } = render(<ErrorMessage title="Error" className="custom-cls" />)
+    expect(container.querySelector('.custom-cls')).toBeInTheDocument()
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/hooks/use-add-documents-steps.spec.ts

@@ -0,0 +1,50 @@
+import { act, renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { useAddDocumentsSteps } from './use-add-documents-steps'
+
+describe('useAddDocumentsSteps', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('should initialize with step 1', () => {
+    const { result } = renderHook(() => useAddDocumentsSteps())
+    expect(result.current.currentStep).toBe(1)
+  })
+
+  it('should return 3 steps', () => {
+    const { result } = renderHook(() => useAddDocumentsSteps())
+    expect(result.current.steps).toHaveLength(3)
+  })
+
+  it('should have correct step labels', () => {
+    const { result } = renderHook(() => useAddDocumentsSteps())
+    const labels = result.current.steps.map(s => s.label)
+    expect(labels[0]).toContain('chooseDatasource')
+    expect(labels[1]).toContain('processDocuments')
+    expect(labels[2]).toContain('processingDocuments')
+  })
+
+  it('should increment step on handleNextStep', () => {
+    const { result } = renderHook(() => useAddDocumentsSteps())
+    act(() => {
+      result.current.handleNextStep()
+    })
+    expect(result.current.currentStep).toBe(2)
+  })
+
+  it('should decrement step on handleBackStep', () => {
+    const { result } = renderHook(() => useAddDocumentsSteps())
+    act(() => {
+      result.current.handleNextStep()
+    })
+    act(() => {
+      result.current.handleNextStep()
+    })
+    expect(result.current.currentStep).toBe(3)
+    act(() => {
+      result.current.handleBackStep()
+    })
+    expect(result.current.currentStep).toBe(2)
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-actions.spec.ts

@@ -0,0 +1,204 @@
+import type { Datasource } from '@/app/components/rag-pipeline/components/panel/test-run/types'
+import type { DataSourceNotionPageMap, NotionPage } from '@/models/common'
+import type { CrawlResultItem, DocumentItem, FileItem } from '@/models/datasets'
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { act, renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DatasourceType } from '@/models/pipeline'
+import { createDataSourceStore } from '../data-source/store'
+import { useDatasourceActions } from './use-datasource-actions'
+
+const mockRunPublishedPipeline = vi.fn()
+vi.mock('@/service/use-pipeline', () => ({
+  useRunPublishedPipeline: () => ({
+    mutateAsync: mockRunPublishedPipeline,
+    isIdle: true,
+    isPending: false,
+  }),
+}))
+vi.mock('@/app/components/base/amplitude', () => ({
+  trackEvent: vi.fn(),
+}))
+
+describe('useDatasourceActions', () => {
+  let store: ReturnType<typeof createDataSourceStore>
+  const defaultParams = () => ({
+    datasource: { nodeId: 'node-1', nodeData: { provider_type: DatasourceType.localFile } } as unknown as Datasource,
+    datasourceType: DatasourceType.localFile,
+    pipelineId: 'pipeline-1',
+    dataSourceStore: store,
+    setEstimateData: vi.fn(),
+    setBatchId: vi.fn(),
+    setDocuments: vi.fn(),
+    handleNextStep: vi.fn(),
+    PagesMapAndSelectedPagesId: {},
+    currentWorkspacePages: undefined as { page_id: string }[] | undefined,
+    clearOnlineDocumentData: vi.fn(),
+    clearWebsiteCrawlData: vi.fn(),
+    clearOnlineDriveData: vi.fn(),
+    setDatasource: vi.fn(),
+  })
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    store = createDataSourceStore()
+  })
+
+  it('should return all action functions', () => {
+    const params = defaultParams()
+    const { result } = renderHook(() => useDatasourceActions(params))
+
+    expect(typeof result.current.onClickProcess).toBe('function')
+    expect(typeof result.current.onClickPreview).toBe('function')
+    expect(typeof result.current.handleSubmit).toBe('function')
+    expect(typeof result.current.handlePreviewFileChange).toBe('function')
+    expect(typeof result.current.handlePreviewOnlineDocumentChange).toBe('function')
+    expect(typeof result.current.handlePreviewWebsiteChange).toBe('function')
+    expect(typeof result.current.handlePreviewOnlineDriveFileChange).toBe('function')
+    expect(typeof result.current.handleSelectAll).toBe('function')
+    expect(typeof result.current.handleSwitchDataSource).toBe('function')
+    expect(typeof result.current.handleCredentialChange).toBe('function')
+    expect(result.current.isIdle).toBe(true)
+    expect(result.current.isPending).toBe(false)
+  })
+
+  it('should handle credential change by clearing data and setting new credential', () => {
+    const params = defaultParams()
+    const { result } = renderHook(() => useDatasourceActions(params))
+
+    act(() => {
+      result.current.handleCredentialChange('cred-new')
+    })
+
+    expect(store.getState().currentCredentialId).toBe('cred-new')
+  })
+
+  it('should handle switch data source', () => {
+    const params = defaultParams()
+    const newDatasource = {
+      nodeId: 'node-2',
+      nodeData: { provider_type: DatasourceType.onlineDocument },
+    } as unknown as Datasource
+
+    const { result } = renderHook(() => useDatasourceActions(params))
+    act(() => {
+      result.current.handleSwitchDataSource(newDatasource)
+    })
+
+    expect(store.getState().currentCredentialId).toBe('')
+    expect(store.getState().currentNodeIdRef.current).toBe('node-2')
+    expect(params.setDatasource).toHaveBeenCalledWith(newDatasource)
+  })
+
+  it('should handle preview file change by updating ref', () => {
+    const params = defaultParams()
+    params.dataSourceStore = store
+
+    const { result } = renderHook(() => useDatasourceActions(params))
+
+    // Set up formRef to prevent null error
+    result.current.formRef.current = { submit: vi.fn() }
+
+    const file = { id: 'f1', name: 'test.pdf' } as unknown as DocumentItem
+    act(() => {
+      result.current.handlePreviewFileChange(file)
+    })
+
+    expect(store.getState().previewLocalFileRef.current).toEqual(file)
+  })
+
+  it('should handle preview online document change', () => {
+    const params = defaultParams()
+    const { result } = renderHook(() => useDatasourceActions(params))
+    result.current.formRef.current = { submit: vi.fn() }
+
+    const page = { page_id: 'p1', page_name: 'My Page' } as unknown as NotionPage
+    act(() => {
+      result.current.handlePreviewOnlineDocumentChange(page)
+    })
+
+    expect(store.getState().previewOnlineDocumentRef.current).toEqual(page)
+  })
+
+  it('should handle preview website change', () => {
+    const params = defaultParams()
+    const { result } = renderHook(() => useDatasourceActions(params))
+    result.current.formRef.current = { submit: vi.fn() }
+
+    const website = { title: 'Page', source_url: 'https://example.com' } as unknown as CrawlResultItem
+    act(() => {
+      result.current.handlePreviewWebsiteChange(website)
+    })
+
+    expect(store.getState().previewWebsitePageRef.current).toEqual(website)
+  })
+
+  it('should handle select all for online documents', () => {
+    const params = defaultParams()
+    params.datasourceType = DatasourceType.onlineDocument
+    params.currentWorkspacePages = [{ page_id: 'p1' }, { page_id: 'p2' }] as unknown as NotionPage[]
+    params.PagesMapAndSelectedPagesId = {
+      p1: { page_id: 'p1', page_name: 'A', workspace_id: 'w1' },
+      p2: { page_id: 'p2', page_name: 'B', workspace_id: 'w1' },
+    } as unknown as DataSourceNotionPageMap
+
+    const { result } = renderHook(() => useDatasourceActions(params))
+
+    // First call: select all
+    act(() => {
+      result.current.handleSelectAll()
+    })
+    expect(store.getState().onlineDocuments).toHaveLength(2)
+
+    // Second call: deselect all
+    act(() => {
+      result.current.handleSelectAll()
+    })
+    expect(store.getState().onlineDocuments).toEqual([])
+  })
+
+  it('should handle select all for online drive', () => {
+    const params = defaultParams()
+    params.datasourceType = DatasourceType.onlineDrive
+
+    store.getState().setOnlineDriveFileList([
+      { id: 'f1', type: 'file' },
+      { id: 'f2', type: 'file' },
+      { id: 'b1', type: 'bucket' },
+    ] as unknown as OnlineDriveFile[])
+
+    const { result } = renderHook(() => useDatasourceActions(params))
+
+    act(() => {
+      result.current.handleSelectAll()
+    })
+    // Should select f1, f2 but not b1 (bucket)
+    expect(store.getState().selectedFileIds).toEqual(['f1', 'f2'])
+  })
+
+  it('should handle submit with preview mode', async () => {
+    const params = defaultParams()
+    store.getState().setLocalFileList([{ file: { id: 'f1', name: 'test.pdf' } }] as unknown as FileItem[])
+    store.getState().previewLocalFileRef.current = { id: 'f1', name: 'test.pdf' } as unknown as DocumentItem
+
+    mockRunPublishedPipeline.mockResolvedValue({ data: { outputs: { tokens: 100 } } })
+
+    const { result } = renderHook(() => useDatasourceActions(params))
+
+    // Set preview mode
+    result.current.isPreview.current = true
+
+    await act(async () => {
+      await result.current.handleSubmit({ query: 'test' })
+    })
+
+    expect(mockRunPublishedPipeline).toHaveBeenCalledWith(
+      expect.objectContaining({
+        pipeline_id: 'pipeline-1',
+        is_preview: true,
+        start_node_id: 'node-1',
+      }),
+      expect.anything(),
+    )
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-options.spec.ts

@@ -0,0 +1,58 @@
+import type { DataSourceNodeType } from '@/app/components/workflow/nodes/data-source/types'
+import type { Node } from '@/app/components/workflow/types'
+import { renderHook } from '@testing-library/react'
+import { describe, expect, it, vi } from 'vitest'
+
+vi.mock('@/app/components/workflow/types', async () => {
+  const actual = await vi.importActual<Record<string, unknown>>('@/app/components/workflow/types')
+  const blockEnum = actual.BlockEnum as Record<string, string>
+  return {
+    ...actual,
+    BlockEnum: {
+      ...blockEnum,
+      DataSource: 'data-source',
+    },
+  }
+})
+
+const { useDatasourceOptions } = await import('./use-datasource-options')
+
+describe('useDatasourceOptions', () => {
+  const createNode = (id: string, title: string, type: string): Node<DataSourceNodeType> => ({
+    id,
+    position: { x: 0, y: 0 },
+    data: {
+      type,
+      title,
+      provider_type: 'local_file',
+    },
+  } as unknown as Node<DataSourceNodeType>)
+
+  it('should return empty array for no datasource nodes', () => {
+    const nodes = [
+      createNode('n1', 'LLM Node', 'llm'),
+    ]
+    const { result } = renderHook(() => useDatasourceOptions(nodes))
+    expect(result.current).toEqual([])
+  })
+
+  it('should return options for datasource nodes', () => {
+    const nodes = [
+      createNode('n1', 'File Upload', 'data-source'),
+      createNode('n2', 'Web Crawl', 'data-source'),
+      createNode('n3', 'LLM Node', 'llm'),
+    ]
+    const { result } = renderHook(() => useDatasourceOptions(nodes))
+    expect(result.current).toHaveLength(2)
+    expect(result.current[0]).toEqual({
+      label: 'File Upload',
+      value: 'n1',
+      data: expect.objectContaining({ title: 'File Upload' }),
+    })
+    expect(result.current[1]).toEqual({
+      label: 'Web Crawl',
+      value: 'n2',
+      data: expect.objectContaining({ title: 'Web Crawl' }),
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-store.spec.ts

@@ -0,0 +1,207 @@
+import type { ReactNode } from 'react'
+import type { DataSourceNotionWorkspace, NotionPage } from '@/models/common'
+import type { CrawlResultItem, CustomFile as File, FileItem } from '@/models/datasets'
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { act, renderHook } from '@testing-library/react'
+import * as React from 'react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { CrawlStep } from '@/models/datasets'
+import { createDataSourceStore } from '../data-source/store'
+import { DataSourceContext } from '../data-source/store/provider'
+import { useLocalFile, useOnlineDocument, useOnlineDrive, useWebsiteCrawl } from './use-datasource-store'
+
+const createWrapper = (store: ReturnType<typeof createDataSourceStore>) => {
+  return ({ children }: { children: ReactNode }) =>
+    React.createElement(DataSourceContext.Provider, { value: store }, children)
+}
+
+describe('useLocalFile', () => {
+  let store: ReturnType<typeof createDataSourceStore>
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    store = createDataSourceStore()
+  })
+
+  it('should return local file list and initial state', () => {
+    const { result } = renderHook(() => useLocalFile(), { wrapper: createWrapper(store) })
+
+    expect(result.current.localFileList).toEqual([])
+    expect(result.current.allFileLoaded).toBe(false)
+    expect(result.current.currentLocalFile).toBeUndefined()
+  })
+
+  it('should compute allFileLoaded when all files have ids', () => {
+    store.getState().setLocalFileList([
+      { file: { id: 'f1', name: 'a.pdf' } },
+      { file: { id: 'f2', name: 'b.pdf' } },
+    ] as unknown as FileItem[])
+
+    const { result } = renderHook(() => useLocalFile(), { wrapper: createWrapper(store) })
+    expect(result.current.allFileLoaded).toBe(true)
+  })
+
+  it('should compute allFileLoaded as false when some files lack ids', () => {
+    store.getState().setLocalFileList([
+      { file: { id: 'f1', name: 'a.pdf' } },
+      { file: { id: '', name: 'b.pdf' } },
+    ] as unknown as FileItem[])
+
+    const { result } = renderHook(() => useLocalFile(), { wrapper: createWrapper(store) })
+    expect(result.current.allFileLoaded).toBe(false)
+  })
+
+  it('should hide preview local file', () => {
+    store.getState().setCurrentLocalFile({ id: 'f1' } as unknown as File)
+    const { result } = renderHook(() => useLocalFile(), { wrapper: createWrapper(store) })
+
+    act(() => {
+      result.current.hidePreviewLocalFile()
+    })
+    expect(store.getState().currentLocalFile).toBeUndefined()
+  })
+})
+
+describe('useOnlineDocument', () => {
+  let store: ReturnType<typeof createDataSourceStore>
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    store = createDataSourceStore()
+  })
+
+  it('should return initial state', () => {
+    const { result } = renderHook(() => useOnlineDocument(), { wrapper: createWrapper(store) })
+
+    expect(result.current.onlineDocuments).toEqual([])
+    expect(result.current.currentDocument).toBeUndefined()
+    expect(result.current.currentWorkspace).toBeUndefined()
+  })
+
+  it('should build PagesMapAndSelectedPagesId from documentsData', () => {
+    store.getState().setDocumentsData([
+      { workspace_id: 'w1', pages: [{ page_id: 'p1', page_name: 'Page 1' }] },
+    ] as unknown as DataSourceNotionWorkspace[])
+
+    const { result } = renderHook(() => useOnlineDocument(), { wrapper: createWrapper(store) })
+    expect(result.current.PagesMapAndSelectedPagesId).toHaveProperty('p1')
+    expect(result.current.PagesMapAndSelectedPagesId.p1.workspace_id).toBe('w1')
+  })
+
+  it('should hide preview online document', () => {
+    store.getState().setCurrentDocument({ page_id: 'p1' } as unknown as NotionPage)
+    const { result } = renderHook(() => useOnlineDocument(), { wrapper: createWrapper(store) })
+
+    act(() => {
+      result.current.hidePreviewOnlineDocument()
+    })
+    expect(store.getState().currentDocument).toBeUndefined()
+  })
+
+  it('should clear online document data', () => {
+    store.getState().setDocumentsData([{ workspace_id: 'w1', pages: [] }] as unknown as DataSourceNotionWorkspace[])
+    store.getState().setSearchValue('test')
+    store.getState().setOnlineDocuments([{ page_id: 'p1' }] as unknown as NotionPage[])
+
+    const { result } = renderHook(() => useOnlineDocument(), { wrapper: createWrapper(store) })
+    act(() => {
+      result.current.clearOnlineDocumentData()
+    })
+
+    expect(store.getState().documentsData).toEqual([])
+    expect(store.getState().searchValue).toBe('')
+    expect(store.getState().onlineDocuments).toEqual([])
+  })
+})
+
+describe('useWebsiteCrawl', () => {
+  let store: ReturnType<typeof createDataSourceStore>
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    store = createDataSourceStore()
+  })
+
+  it('should return initial state', () => {
+    const { result } = renderHook(() => useWebsiteCrawl(), { wrapper: createWrapper(store) })
+
+    expect(result.current.websitePages).toEqual([])
+    expect(result.current.currentWebsite).toBeUndefined()
+  })
+
+  it('should hide website preview', () => {
+    store.getState().setCurrentWebsite({ title: 'Test' } as unknown as CrawlResultItem)
+    store.getState().setPreviewIndex(2)
+    const { result } = renderHook(() => useWebsiteCrawl(), { wrapper: createWrapper(store) })
+
+    act(() => {
+      result.current.hideWebsitePreview()
+    })
+
+    expect(store.getState().currentWebsite).toBeUndefined()
+    expect(store.getState().previewIndex).toBe(-1)
+  })
+
+  it('should clear website crawl data', () => {
+    store.getState().setStep(CrawlStep.running)
+    store.getState().setWebsitePages([{ title: 'Test' }] as unknown as CrawlResultItem[])
+
+    const { result } = renderHook(() => useWebsiteCrawl(), { wrapper: createWrapper(store) })
+    act(() => {
+      result.current.clearWebsiteCrawlData()
+    })
+
+    expect(store.getState().step).toBe(CrawlStep.init)
+    expect(store.getState().websitePages).toEqual([])
+    expect(store.getState().currentWebsite).toBeUndefined()
+  })
+})
+
+describe('useOnlineDrive', () => {
+  let store: ReturnType<typeof createDataSourceStore>
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+    store = createDataSourceStore()
+  })
+
+  it('should return initial state', () => {
+    const { result } = renderHook(() => useOnlineDrive(), { wrapper: createWrapper(store) })
+
+    expect(result.current.onlineDriveFileList).toEqual([])
+    expect(result.current.selectedFileIds).toEqual([])
+    expect(result.current.selectedOnlineDriveFileList).toEqual([])
+  })
+
+  it('should compute selected online drive file list', () => {
+    const files = [
+      { id: 'f1', name: 'a.pdf' },
+      { id: 'f2', name: 'b.pdf' },
+      { id: 'f3', name: 'c.pdf' },
+    ] as unknown as OnlineDriveFile[]
+    store.getState().setOnlineDriveFileList(files)
+    store.getState().setSelectedFileIds(['f1', 'f3'])
+
+    const { result } = renderHook(() => useOnlineDrive(), { wrapper: createWrapper(store) })
+    expect(result.current.selectedOnlineDriveFileList).toEqual([files[0], files[2]])
+  })
+
+  it('should clear online drive data', () => {
+    store.getState().setOnlineDriveFileList([{ id: 'f1' }] as unknown as OnlineDriveFile[])
+    store.getState().setBucket('b1')
+    store.getState().setPrefix(['p1'])
+    store.getState().setKeywords('kw')
+    store.getState().setSelectedFileIds(['f1'])
+
+    const { result } = renderHook(() => useOnlineDrive(), { wrapper: createWrapper(store) })
+    act(() => {
+      result.current.clearOnlineDriveData()
+    })
+
+    expect(store.getState().onlineDriveFileList).toEqual([])
+    expect(store.getState().bucket).toBe('')
+    expect(store.getState().prefix).toEqual([])
+    expect(store.getState().keywords).toBe('')
+    expect(store.getState().selectedFileIds).toEqual([])
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/hooks/use-datasource-ui-state.spec.ts

@@ -0,0 +1,205 @@
+import type { Datasource } from '@/app/components/rag-pipeline/components/panel/test-run/types'
+import type { OnlineDriveFile } from '@/models/pipeline'
+import { renderHook } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { DatasourceType, OnlineDriveFileType } from '@/models/pipeline'
+import { useDatasourceUIState } from './use-datasource-ui-state'
+
+describe('useDatasourceUIState', () => {
+  const defaultParams = {
+    datasource: { nodeData: { provider_type: DatasourceType.localFile } } as unknown as Datasource,
+    allFileLoaded: true,
+    localFileListLength: 3,
+    onlineDocumentsLength: 0,
+    websitePagesLength: 0,
+    selectedFileIdsLength: 0,
+    onlineDriveFileList: [] as OnlineDriveFile[],
+    isVectorSpaceFull: false,
+    enableBilling: false,
+    currentWorkspacePagesLength: 0,
+    fileUploadConfig: { file_size_limit: 50, batch_count_limit: 20 },
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  describe('datasourceType', () => {
+    it('should return provider_type from datasource', () => {
+      const { result } = renderHook(() => useDatasourceUIState(defaultParams))
+      expect(result.current.datasourceType).toBe(DatasourceType.localFile)
+    })
+
+    it('should return undefined when no datasource', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({ ...defaultParams, datasource: undefined }),
+      )
+      expect(result.current.datasourceType).toBeUndefined()
+    })
+  })
+
+  describe('isShowVectorSpaceFull', () => {
+    it('should be false when billing disabled', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({ ...defaultParams, isVectorSpaceFull: true }),
+      )
+      expect(result.current.isShowVectorSpaceFull).toBe(false)
+    })
+
+    it('should be true when billing enabled and space is full for local file', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          isVectorSpaceFull: true,
+          enableBilling: true,
+          allFileLoaded: true,
+        }),
+      )
+      expect(result.current.isShowVectorSpaceFull).toBe(true)
+    })
+
+    it('should be false when no datasource', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: undefined,
+          isVectorSpaceFull: true,
+          enableBilling: true,
+        }),
+      )
+      expect(result.current.isShowVectorSpaceFull).toBe(false)
+    })
+  })
+
+  describe('nextBtnDisabled', () => {
+    it('should be true when no datasource', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({ ...defaultParams, datasource: undefined }),
+      )
+      expect(result.current.nextBtnDisabled).toBe(true)
+    })
+
+    it('should be false when local files loaded', () => {
+      const { result } = renderHook(() => useDatasourceUIState(defaultParams))
+      expect(result.current.nextBtnDisabled).toBe(false)
+    })
+
+    it('should be true when local file list empty', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({ ...defaultParams, localFileListLength: 0 }),
+      )
+      expect(result.current.nextBtnDisabled).toBe(true)
+    })
+
+    it('should be true when files not all loaded', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({ ...defaultParams, allFileLoaded: false }),
+      )
+      expect(result.current.nextBtnDisabled).toBe(true)
+    })
+
+    it('should be false for online document with documents selected', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: { nodeData: { provider_type: DatasourceType.onlineDocument } } as unknown as Datasource,
+          onlineDocumentsLength: 2,
+        }),
+      )
+      expect(result.current.nextBtnDisabled).toBe(false)
+    })
+
+    it('should be true for online document with no documents', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: { nodeData: { provider_type: DatasourceType.onlineDocument } } as unknown as Datasource,
+          onlineDocumentsLength: 0,
+        }),
+      )
+      expect(result.current.nextBtnDisabled).toBe(true)
+    })
+  })
+
+  describe('showSelect', () => {
+    it('should be false for local file type', () => {
+      const { result } = renderHook(() => useDatasourceUIState(defaultParams))
+      expect(result.current.showSelect).toBe(false)
+    })
+
+    it('should be true for online document with workspace pages', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: { nodeData: { provider_type: DatasourceType.onlineDocument } } as unknown as Datasource,
+          currentWorkspacePagesLength: 5,
+        }),
+      )
+      expect(result.current.showSelect).toBe(true)
+    })
+
+    it('should be true for online drive with non-bucket files', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: { nodeData: { provider_type: DatasourceType.onlineDrive } } as unknown as Datasource,
+          onlineDriveFileList: [
+            { id: '1', name: 'file.txt', type: OnlineDriveFileType.file },
+          ],
+        }),
+      )
+      expect(result.current.showSelect).toBe(true)
+    })
+
+    it('should be false for online drive showing only buckets', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: { nodeData: { provider_type: DatasourceType.onlineDrive } } as unknown as Datasource,
+          onlineDriveFileList: [
+            { id: '1', name: 'bucket-1', type: OnlineDriveFileType.bucket },
+          ],
+        }),
+      )
+      expect(result.current.showSelect).toBe(false)
+    })
+  })
+
+  describe('totalOptions and selectedOptions', () => {
+    it('should return workspace pages count for online document', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: { nodeData: { provider_type: DatasourceType.onlineDocument } } as unknown as Datasource,
+          currentWorkspacePagesLength: 10,
+          onlineDocumentsLength: 3,
+        }),
+      )
+      expect(result.current.totalOptions).toBe(10)
+      expect(result.current.selectedOptions).toBe(3)
+    })
+
+    it('should return undefined for local file type', () => {
+      const { result } = renderHook(() => useDatasourceUIState(defaultParams))
+      expect(result.current.totalOptions).toBeUndefined()
+      expect(result.current.selectedOptions).toBeUndefined()
+    })
+  })
+
+  describe('tip', () => {
+    it('should return empty string for local file', () => {
+      const { result } = renderHook(() => useDatasourceUIState(defaultParams))
+      expect(result.current.tip).toBe('')
+    })
+
+    it('should return tip for online document', () => {
+      const { result } = renderHook(() =>
+        useDatasourceUIState({
+          ...defaultParams,
+          datasource: { nodeData: { provider_type: DatasourceType.onlineDocument } } as unknown as Datasource,
+        }),
+      )
+      expect(result.current.tip).toContain('selectOnlineDocumentTip')
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/left-header.spec.tsx

@@ -0,0 +1,115 @@
+import type { Step } from './step-indicator'
+import { render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import LeftHeader from './left-header'
+
+vi.mock('next/navigation', () => ({
+  useParams: () => ({ datasetId: 'test-ds-id' }),
+}))
+
+vi.mock('next/link', () => ({
+  default: ({ children, href }: { children: React.ReactNode, href: string }) => (
+    <a href={href} data-testid="back-link">{children}</a>
+  ),
+}))
+
+vi.mock('@remixicon/react', () => ({
+  RiArrowLeftLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="arrow-left-icon" {...props} />,
+}))
+
+vi.mock('./step-indicator', () => ({
+  default: ({ steps, currentStep }: { steps: Step[], currentStep: number }) => (
+    <div data-testid="step-indicator" data-steps={steps.length} data-current={currentStep} />
+  ),
+}))
+
+vi.mock('@/app/components/base/effect', () => ({
+  default: ({ className }: { className?: string }) => (
+    <div data-testid="effect" className={className} />
+  ),
+}))
+
+const createSteps = (): Step[] => [
+  { label: 'Data Source', value: 'data-source' },
+  { label: 'Processing', value: 'processing' },
+  { label: 'Complete', value: 'complete' },
+]
+
+describe('LeftHeader', () => {
+  const steps = createSteps()
+
+  const defaultProps = {
+    steps,
+    title: 'Add Documents',
+    currentStep: 1,
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering: title, step label, and step indicator
+  describe('Rendering', () => {
+    it('should render title text', () => {
+      render(<LeftHeader {...defaultProps} />)
+
+      expect(screen.getByText('Add Documents')).toBeInTheDocument()
+    })
+
+    it('should render current step label (steps[currentStep-1].label)', () => {
+      render(<LeftHeader {...defaultProps} currentStep={2} />)
+
+      expect(screen.getByText('Processing')).toBeInTheDocument()
+    })
+
+    it('should render step indicator component', () => {
+      render(<LeftHeader {...defaultProps} />)
+
+      expect(screen.getByTestId('step-indicator')).toBeInTheDocument()
+    })
+
+    it('should render separator between title and step indicator', () => {
+      render(<LeftHeader {...defaultProps} />)
+
+      expect(screen.getByText('/')).toBeInTheDocument()
+    })
+  })
+
+  // Back button visibility depends on currentStep vs total steps
+  describe('Back Button', () => {
+    it('should show back button when currentStep !== steps.length', () => {
+      render(<LeftHeader {...defaultProps} currentStep={1} />)
+
+      expect(screen.getByTestId('back-link')).toBeInTheDocument()
+      expect(screen.getByTestId('arrow-left-icon')).toBeInTheDocument()
+    })
+
+    it('should hide back button when currentStep === steps.length', () => {
+      render(<LeftHeader {...defaultProps} currentStep={steps.length} />)
+
+      expect(screen.queryByTestId('back-link')).not.toBeInTheDocument()
+    })
+
+    it('should link to correct URL using datasetId from params', () => {
+      render(<LeftHeader {...defaultProps} currentStep={1} />)
+
+      const link = screen.getByTestId('back-link')
+      expect(link).toHaveAttribute('href', '/datasets/test-ds-id/documents')
+    })
+  })
+
+  // Edge case: step label for boundary values
+  describe('Edge Cases', () => {
+    it('should render first step label when currentStep is 1', () => {
+      render(<LeftHeader {...defaultProps} currentStep={1} />)
+
+      expect(screen.getByText('Data Source')).toBeInTheDocument()
+    })
+
+    it('should render last step label when currentStep equals steps.length', () => {
+      render(<LeftHeader {...defaultProps} currentStep={3} />)
+
+      expect(screen.getByText('Complete')).toBeInTheDocument()
+    })
+  })
+})

web/app/components/datasets/documents/create-from-pipeline/preview/file-preview.spec.tsx

@@ -1,320 +1,77 @@
-import type { CustomFile as File } from '@/models/datasets'
+import type { CustomFile } from '@/models/datasets'
 import { fireEvent, render, screen } from '@testing-library/react'
-import * as React from 'react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
 import FilePreview from './file-preview'
 
-// Uses global react-i18next mock from web/vitest.setup.ts
-
-// Mock useFilePreview hook - needs to be mocked to control return values
-const mockUseFilePreview = vi.fn()
-vi.mock('@/service/use-common', () => ({
-  useFilePreview: (fileID: string) => mockUseFilePreview(fileID),
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
 }))
 
-// Test data factory
-const createMockFile = (overrides?: Partial<File>): File => ({
-  id: 'file-123',
-  name: 'test-document.pdf',
-  size: 2048,
-  type: 'application/pdf',
-  extension: 'pdf',
-  lastModified: Date.now(),
-  webkitRelativePath: '',
-  arrayBuffer: vi.fn() as () => Promise<ArrayBuffer>,
-  bytes: vi.fn() as () => Promise<Uint8Array>,
-  slice: vi.fn() as (start?: number, end?: number, contentType?: string) => Blob,
-  stream: vi.fn() as () => ReadableStream<Uint8Array>,
-  text: vi.fn() as () => Promise<string>,
-  ...overrides,
-} as File)
+vi.mock('@remixicon/react', () => ({
+  RiCloseLine: () => <span data-testid="close-icon" />,
+}))
 
-const createMockFilePreviewData = (content: string = 'This is the file content') => ({
-  content,
-})
+const mockFileData = { content: 'file content here with some text' }
+let mockIsFetching = false
 
-const defaultProps = {
-  file: createMockFile(),
-  hidePreview: vi.fn(),
-}
+vi.mock('@/service/use-common', () => ({
+  useFilePreview: () => ({
+    data: mockIsFetching ? undefined : mockFileData,
+    isFetching: mockIsFetching,
+  }),
+}))
+
+vi.mock('../../../common/document-file-icon', () => ({
+  default: () => <span data-testid="file-icon" />,
+}))
+
+vi.mock('./loading', () => ({
+  default: () => <div data-testid="loading" />,
+}))
 
 describe('FilePreview', () => {
+  const defaultProps = {
+    file: {
+      id: 'file-1',
+      name: 'document.pdf',
+      extension: 'pdf',
+      size: 1024,
+    } as CustomFile,
+    hidePreview: vi.fn(),
+  }
+
   beforeEach(() => {
     vi.clearAllMocks()
-    mockUseFilePreview.mockReturnValue({
-      data: undefined,
-      isFetching: false,
-    })
+    mockIsFetching = false
   })
 
-  describe('Rendering', () => {
-    it('should render the component with file information', () => {
-      render(<FilePreview {...defaultProps} />)
-
-      // i18n mock returns key by default
-      expect(screen.getByText('datasetPipeline.addDocuments.stepOne.preview')).toBeInTheDocument()
-      expect(screen.getByText('test-document.pdf')).toBeInTheDocument()
-    })
-
-    it('should display file extension in uppercase via CSS class', () => {
-      render(<FilePreview {...defaultProps} />)
-
-      // The extension is displayed in the info section (as uppercase via CSS class)
-      const extensionElement = screen.getByText('pdf')
-      expect(extensionElement).toBeInTheDocument()
-      expect(extensionElement).toHaveClass('uppercase')
-    })
-
-    it('should display formatted file size', () => {
-      render(<FilePreview {...defaultProps} />)
-
-      // Real formatFileSize: 2048 bytes => "2.00 KB"
-      expect(screen.getByText('2.00 KB')).toBeInTheDocument()
-    })
-
-    it('should render close button', () => {
-      render(<FilePreview {...defaultProps} />)
-
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should call useFilePreview with correct fileID', () => {
-      const file = createMockFile({ id: 'specific-file-id' })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      expect(mockUseFilePreview).toHaveBeenCalledWith('specific-file-id')
-    })
+  it('should render preview label', () => {
+    render(<FilePreview {...defaultProps} />)
+    expect(screen.getByText('addDocuments.stepOne.preview')).toBeInTheDocument()
   })
 
-  describe('File Name Processing', () => {
-    it('should extract file name without extension', () => {
-      const file = createMockFile({ name: 'my-document.pdf', extension: 'pdf' })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // The displayed text is `${fileName}.${extension}`, where fileName is name without ext
-      // my-document.pdf -> fileName = 'my-document', displayed as 'my-document.pdf'
-      expect(screen.getByText('my-document.pdf')).toBeInTheDocument()
-    })
-
-    it('should handle file name with multiple dots', () => {
-      const file = createMockFile({ name: 'my.file.name.pdf', extension: 'pdf' })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // fileName = arr.slice(0, -1).join() = 'my,file,name', then displayed as 'my,file,name.pdf'
-      expect(screen.getByText('my,file,name.pdf')).toBeInTheDocument()
-    })
-
-    it('should handle empty file name', () => {
-      const file = createMockFile({ name: '', extension: '' })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // fileName = '', displayed as '.'
-      expect(screen.getByText('.')).toBeInTheDocument()
-    })
-
-    it('should handle file without extension in name', () => {
-      const file = createMockFile({ name: 'noextension', extension: '' })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // fileName = '' (slice returns empty for single element array), displayed as '.'
-      expect(screen.getByText('.')).toBeInTheDocument()
-    })
+  it('should render file name', () => {
+    render(<FilePreview {...defaultProps} />)
+    expect(screen.getByText('document.pdf')).toBeInTheDocument()
   })
 
-  describe('Loading State', () => {
-    it('should render loading component when fetching', () => {
-      mockUseFilePreview.mockReturnValue({
-        data: undefined,
-        isFetching: true,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      // Loading component renders skeleton
-      expect(document.querySelector('.overflow-hidden')).toBeInTheDocument()
-    })
-
-    it('should not render content when loading', () => {
-      mockUseFilePreview.mockReturnValue({
-        data: createMockFilePreviewData('Some content'),
-        isFetching: true,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      expect(screen.queryByText('Some content')).not.toBeInTheDocument()
-    })
+  it('should render file content when loaded', () => {
+    render(<FilePreview {...defaultProps} />)
+    expect(screen.getByText('file content here with some text')).toBeInTheDocument()
   })
 
-  describe('Content Display', () => {
-    it('should render file content when loaded', () => {
-      mockUseFilePreview.mockReturnValue({
-        data: createMockFilePreviewData('This is the file content'),
-        isFetching: false,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      expect(screen.getByText('This is the file content')).toBeInTheDocument()
-    })
-
-    it('should display character count when data is available', () => {
-      mockUseFilePreview.mockReturnValue({
-        data: createMockFilePreviewData('Hello'), // 5 characters
-        isFetching: false,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      // Real formatNumberAbbreviated returns "5" for numbers < 1000
-      expect(screen.getByText(/5/)).toBeInTheDocument()
-    })
-
-    it('should format large character counts', () => {
-      const longContent = 'a'.repeat(2500)
-      mockUseFilePreview.mockReturnValue({
-        data: createMockFilePreviewData(longContent),
-        isFetching: false,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      // Real formatNumberAbbreviated uses lowercase 'k': "2.5k"
-      expect(screen.getByText(/2\.5k/)).toBeInTheDocument()
-    })
-
-    it('should not display character count when data is not available', () => {
-      mockUseFilePreview.mockReturnValue({
-        data: undefined,
-        isFetching: false,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      // No character text shown
-      expect(screen.queryByText(/datasetPipeline\.addDocuments\.characters/)).not.toBeInTheDocument()
-    })
+  it('should render loading state', () => {
+    mockIsFetching = true
+    render(<FilePreview {...defaultProps} />)
+    expect(screen.getByTestId('loading')).toBeInTheDocument()
   })
 
-  describe('User Interactions', () => {
-    it('should call hidePreview when close button is clicked', () => {
-      const hidePreview = vi.fn()
-
-      render(<FilePreview {...defaultProps} hidePreview={hidePreview} />)
-
-      const closeButton = screen.getByRole('button')
-      fireEvent.click(closeButton)
-
-      expect(hidePreview).toHaveBeenCalledTimes(1)
-    })
-  })
-
-  describe('File Size Formatting', () => {
-    it('should format small file sizes in bytes', () => {
-      const file = createMockFile({ size: 500 })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // Real formatFileSize: 500 => "500.00 bytes"
-      expect(screen.getByText('500.00 bytes')).toBeInTheDocument()
-    })
-
-    it('should format kilobyte file sizes', () => {
-      const file = createMockFile({ size: 5120 })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // Real formatFileSize: 5120 => "5.00 KB"
-      expect(screen.getByText('5.00 KB')).toBeInTheDocument()
-    })
-
-    it('should format megabyte file sizes', () => {
-      const file = createMockFile({ size: 2097152 })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // Real formatFileSize: 2097152 => "2.00 MB"
-      expect(screen.getByText('2.00 MB')).toBeInTheDocument()
-    })
-  })
-
-  describe('Edge Cases', () => {
-    it('should handle undefined file id', () => {
-      const file = createMockFile({ id: undefined })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      expect(mockUseFilePreview).toHaveBeenCalledWith('')
-    })
-
-    it('should handle empty extension', () => {
-      const file = createMockFile({ extension: undefined })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should handle zero file size', () => {
-      const file = createMockFile({ size: 0 })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      // Real formatFileSize returns 0 for falsy values
-      // The component still renders
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should handle very long file content', () => {
-      const veryLongContent = 'a'.repeat(1000000)
-      mockUseFilePreview.mockReturnValue({
-        data: createMockFilePreviewData(veryLongContent),
-        isFetching: false,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      // Real formatNumberAbbreviated: 1000000 => "1M"
-      expect(screen.getByText(/1M/)).toBeInTheDocument()
-    })
-
-    it('should handle empty content', () => {
-      mockUseFilePreview.mockReturnValue({
-        data: createMockFilePreviewData(''),
-        isFetching: false,
-      })
-
-      render(<FilePreview {...defaultProps} />)
-
-      // Real formatNumberAbbreviated: 0 => "0"
-      // Find the element that contains character count info
-      expect(screen.getByText(/0 datasetPipeline/)).toBeInTheDocument()
-    })
-  })
-
-  describe('useMemo for fileName', () => {
-    it('should extract file name when file exists', () => {
-      // When file exists, it should extract the name without extension
-      const file = createMockFile({ name: 'document.txt', extension: 'txt' })
-
-      render(<FilePreview {...defaultProps} file={file} />)
-
-      expect(screen.getByText('document.txt')).toBeInTheDocument()
-    })
-
-    it('should memoize fileName based on file prop', () => {
-      const file = createMockFile({ name: 'test.pdf', extension: 'pdf' })
-
-      const { rerender } = render(<FilePreview {...defaultProps} file={file} />)
-
-      // Same file should produce same result
-      rerender(<FilePreview {...defaultProps} file={file} />)
-
-      expect(screen.getByText('test.pdf')).toBeInTheDocument()
-    })
+  it('should call hidePreview when close button clicked', () => {
+    render(<FilePreview {...defaultProps} />)
+    const closeBtn = screen.getByTestId('close-icon').closest('button')!
+    fireEvent.click(closeBtn)
+    expect(defaultProps.hidePreview).toHaveBeenCalled()
   })
 })

web/app/components/datasets/documents/create-from-pipeline/preview/web-preview.spec.tsx

@@ -1,256 +1,58 @@
 import type { CrawlResultItem } from '@/models/datasets'
 import { fireEvent, render, screen } from '@testing-library/react'
-import * as React from 'react'
-import WebsitePreview from './web-preview'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import WebPreview from './web-preview'
 
-// Uses global react-i18next mock from web/vitest.setup.ts
+vi.mock('react-i18next', () => ({
+  useTranslation: () => ({
+    t: (key: string) => key,
+  }),
+}))
 
-// Test data factory
-const createMockCrawlResult = (overrides?: Partial<CrawlResultItem>): CrawlResultItem => ({
-  title: 'Test Website Title',
-  markdown: 'This is the **markdown** content of the website.',
-  description: 'Test description',
-  source_url: 'https://example.com/page',
-  ...overrides,
-})
+vi.mock('@remixicon/react', () => ({
+  RiCloseLine: () => <span data-testid="close-icon" />,
+  RiGlobalLine: () => <span data-testid="global-icon" />,
+}))
 
-const defaultProps = {
-  currentWebsite: createMockCrawlResult(),
-  hidePreview: vi.fn(),
-}
+describe('WebPreview', () => {
+  const defaultProps = {
+    currentWebsite: {
+      title: 'Test Page',
+      source_url: 'https://example.com',
+      markdown: 'Hello **markdown** content',
+      description: '',
+    } satisfies CrawlResultItem,
+    hidePreview: vi.fn(),
+  }
 
-describe('WebsitePreview', () => {
   beforeEach(() => {
     vi.clearAllMocks()
   })
 
-  describe('Rendering', () => {
-    it('should render the component with website information', () => {
-      render(<WebsitePreview {...defaultProps} />)
-
-      // i18n mock returns key by default
-      expect(screen.getByText('datasetPipeline.addDocuments.stepOne.preview')).toBeInTheDocument()
-      expect(screen.getByText('Test Website Title')).toBeInTheDocument()
-    })
-
-    it('should display the source URL', () => {
-      render(<WebsitePreview {...defaultProps} />)
-
-      expect(screen.getByText('https://example.com/page')).toBeInTheDocument()
-    })
-
-    it('should render close button', () => {
-      render(<WebsitePreview {...defaultProps} />)
-
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should render the markdown content', () => {
-      render(<WebsitePreview {...defaultProps} />)
-
-      expect(screen.getByText('This is the **markdown** content of the website.')).toBeInTheDocument()
-    })
+  it('should render preview label', () => {
+    render(<WebPreview {...defaultProps} />)
+    expect(screen.getByText('addDocuments.stepOne.preview')).toBeInTheDocument()
   })
 
-  describe('Character Count', () => {
-    it('should display character count for small content', () => {
-      const currentWebsite = createMockCrawlResult({ markdown: 'Hello' }) // 5 characters
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      // Real formatNumberAbbreviated returns "5" for numbers < 1000
-      expect(screen.getByText(/5/)).toBeInTheDocument()
-    })
-
-    it('should format character count in thousands', () => {
-      const longContent = 'a'.repeat(2500)
-      const currentWebsite = createMockCrawlResult({ markdown: longContent })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      // Real formatNumberAbbreviated uses lowercase 'k': "2.5k"
-      expect(screen.getByText(/2\.5k/)).toBeInTheDocument()
-    })
-
-    it('should format character count in millions', () => {
-      const veryLongContent = 'a'.repeat(1500000)
-      const currentWebsite = createMockCrawlResult({ markdown: veryLongContent })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByText(/1\.5M/)).toBeInTheDocument()
-    })
-
-    it('should show 0 characters for empty markdown', () => {
-      const currentWebsite = createMockCrawlResult({ markdown: '' })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByText(/0/)).toBeInTheDocument()
-    })
+  it('should render page title', () => {
+    render(<WebPreview {...defaultProps} />)
+    expect(screen.getByText('Test Page')).toBeInTheDocument()
   })
 
-  describe('User Interactions', () => {
-    it('should call hidePreview when close button is clicked', () => {
-      const hidePreview = vi.fn()
-
-      render(<WebsitePreview {...defaultProps} hidePreview={hidePreview} />)
-
-      const closeButton = screen.getByRole('button')
-      fireEvent.click(closeButton)
-
-      expect(hidePreview).toHaveBeenCalledTimes(1)
-    })
+  it('should render source URL', () => {
+    render(<WebPreview {...defaultProps} />)
+    expect(screen.getByText('https://example.com')).toBeInTheDocument()
   })
 
-  describe('URL Display', () => {
-    it('should display long URLs', () => {
-      const longUrl = 'https://example.com/very/long/path/to/page/with/many/segments'
-      const currentWebsite = createMockCrawlResult({ source_url: longUrl })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      const urlElement = screen.getByTitle(longUrl)
-      expect(urlElement).toBeInTheDocument()
-      expect(urlElement).toHaveTextContent(longUrl)
-    })
-
-    it('should display URL with title attribute', () => {
-      const currentWebsite = createMockCrawlResult({ source_url: 'https://test.com' })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByTitle('https://test.com')).toBeInTheDocument()
-    })
+  it('should render markdown content', () => {
+    render(<WebPreview {...defaultProps} />)
+    expect(screen.getByText('Hello **markdown** content')).toBeInTheDocument()
   })
 
-  describe('Content Display', () => {
-    it('should display the markdown content in content area', () => {
-      const currentWebsite = createMockCrawlResult({
-        markdown: 'Content with **bold** and *italic* text.',
-      })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByText('Content with **bold** and *italic* text.')).toBeInTheDocument()
-    })
-
-    it('should handle multiline content', () => {
-      const multilineContent = 'Line 1\nLine 2\nLine 3'
-      const currentWebsite = createMockCrawlResult({ markdown: multilineContent })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      // Multiline content is rendered as-is
-      expect(screen.getByText((content) => {
-        return content.includes('Line 1') && content.includes('Line 2') && content.includes('Line 3')
-      })).toBeInTheDocument()
-    })
-
-    it('should handle special characters in content', () => {
-      const specialContent = '<script>alert("xss")</script> & < > " \''
-      const currentWebsite = createMockCrawlResult({ markdown: specialContent })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByText(specialContent)).toBeInTheDocument()
-    })
-  })
-
-  describe('Edge Cases', () => {
-    it('should handle empty title', () => {
-      const currentWebsite = createMockCrawlResult({ title: '' })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should handle empty source URL', () => {
-      const currentWebsite = createMockCrawlResult({ source_url: '' })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByRole('button')).toBeInTheDocument()
-    })
-
-    it('should handle very long title', () => {
-      const longTitle = 'A'.repeat(500)
-      const currentWebsite = createMockCrawlResult({ title: longTitle })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByText(longTitle)).toBeInTheDocument()
-    })
-
-    it('should handle unicode characters in content', () => {
-      const unicodeContent = '你好世界 🌍 مرحبا こんにちは'
-      const currentWebsite = createMockCrawlResult({ markdown: unicodeContent })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByText(unicodeContent)).toBeInTheDocument()
-    })
-
-    it('should handle URL with query parameters', () => {
-      const urlWithParams = 'https://example.com/page?query=test&param=value'
-      const currentWebsite = createMockCrawlResult({ source_url: urlWithParams })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByTitle(urlWithParams)).toBeInTheDocument()
-    })
-
-    it('should handle URL with hash fragment', () => {
-      const urlWithHash = 'https://example.com/page#section-1'
-      const currentWebsite = createMockCrawlResult({ source_url: urlWithHash })
-
-      render(<WebsitePreview {...defaultProps} currentWebsite={currentWebsite} />)
-
-      expect(screen.getByTitle(urlWithHash)).toBeInTheDocument()
-    })
-  })
-
-  describe('Styling', () => {
-    it('should apply container styles', () => {
-      const { container } = render(<WebsitePreview {...defaultProps} />)
-
-      const mainContainer = container.firstChild as HTMLElement
-      expect(mainContainer).toHaveClass('flex', 'h-full', 'w-full', 'flex-col')
-    })
-  })
-
-  describe('Multiple Renders', () => {
-    it('should update when currentWebsite changes', () => {
-      const website1 = createMockCrawlResult({ title: 'Website 1', markdown: 'Content 1' })
-      const website2 = createMockCrawlResult({ title: 'Website 2', markdown: 'Content 2' })
-
-      const { rerender } = render(<WebsitePreview {...defaultProps} currentWebsite={website1} />)
-
-      expect(screen.getByText('Website 1')).toBeInTheDocument()
-      expect(screen.getByText('Content 1')).toBeInTheDocument()
-
-      rerender(<WebsitePreview {...defaultProps} currentWebsite={website2} />)
-
-      expect(screen.getByText('Website 2')).toBeInTheDocument()
-      expect(screen.getByText('Content 2')).toBeInTheDocument()
-    })
-
-    it('should call new hidePreview when prop changes', () => {
-      const hidePreview1 = vi.fn()
-      const hidePreview2 = vi.fn()
-
-      const { rerender } = render(<WebsitePreview {...defaultProps} hidePreview={hidePreview1} />)
-
-      const closeButton = screen.getByRole('button')
-      fireEvent.click(closeButton)
-      expect(hidePreview1).toHaveBeenCalledTimes(1)
-
-      rerender(<WebsitePreview {...defaultProps} hidePreview={hidePreview2} />)
-
-      fireEvent.click(closeButton)
-      expect(hidePreview2).toHaveBeenCalledTimes(1)
-      expect(hidePreview1).toHaveBeenCalledTimes(1)
-    })
+  it('should call hidePreview when close button clicked', () => {
+    render(<WebPreview {...defaultProps} />)
+    const closeBtn = screen.getByTestId('close-icon').closest('button')!
+    fireEvent.click(closeBtn)
+    expect(defaultProps.hidePreview).toHaveBeenCalled()
   })
 })

web/app/components/datasets/documents/create-from-pipeline/process-documents/actions.spec.tsx

@@ -0,0 +1,73 @@
+import { fireEvent, render, screen } from '@testing-library/react'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import Actions from './actions'
+
+vi.mock('@remixicon/react', () => ({
+  RiArrowLeftLine: (props: React.SVGProps<SVGSVGElement>) => <svg data-testid="arrow-left-icon" {...props} />,
+}))
+
+describe('Actions', () => {
+  const defaultProps = {
+    onBack: vi.fn(),
+    onProcess: vi.fn(),
+  }
+
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  // Rendering: verify both action buttons render with correct labels
+  describe('Rendering', () => {
+    it('should render back button and process button', () => {
+      render(<Actions {...defaultProps} />)
+
+      const buttons = screen.getAllByRole('button')
+      expect(buttons).toHaveLength(2)
+      expect(screen.getByText('datasetPipeline.operations.dataSource')).toBeInTheDocument()
+      expect(screen.getByText('datasetPipeline.operations.saveAndProcess')).toBeInTheDocument()
+    })
+  })
+
+  // User interactions: clicking back and process buttons
+  describe('User Interactions', () => {
+    it('should call onBack when back button clicked', () => {
+      render(<Actions {...defaultProps} />)
+
+      fireEvent.click(screen.getByText('datasetPipeline.operations.dataSource'))
+
+      expect(defaultProps.onBack).toHaveBeenCalledOnce()
+    })
+
+    it('should call onProcess when process button clicked', () => {
+      render(<Actions {...defaultProps} />)
+
+      fireEvent.click(screen.getByText('datasetPipeline.operations.saveAndProcess'))
+
+      expect(defaultProps.onProcess).toHaveBeenCalledOnce()
+    })
+  })
+
+  // Props: disabled state for the process button
+  describe('Props', () => {
+    it('should disable process button when runDisabled is true', () => {
+      render(<Actions {...defaultProps} runDisabled />)
+
+      const processButton = screen.getByText('datasetPipeline.operations.saveAndProcess').closest('button')
+      expect(processButton).toBeDisabled()
+    })
+
+    it('should enable process button when runDisabled is false', () => {
+      render(<Actions {...defaultProps} runDisabled={false} />)
+
+      const processButton = screen.getByText('datasetPipeline.operations.saveAndProcess').closest('button')
+      expect(processButton).not.toBeDisabled()
+    })
+
+    it('should enable process button when runDisabled is undefined', () => {
+      render(<Actions {...defaultProps} />)
+
+      const processButton = screen.getByText('datasetPipeline.operations.saveAndProcess').closest('button')
+      expect(processButton).not.toBeDisabled()
+    })
+  })
+})