youngkingdom/dify
1
0
Fork 0
mirror of https://github.com/langgenius/dify.git synced 2026-05-03 17:08:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

merge main

Browse Source
This commit is contained in:
Joel
2024-10-28 10:51:02 +08:00
parent c777d55a1c aa11141660
commit 765eb282f3
858 changed files with 16206 additions and 17932 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
api/core/agent/base_agent_runner.py
View File

@ -16,13 +16,14 @@ from core.app.entities.app_invoke_entities import (
)
from core.callback_handler.agent_tool_callback_handler import DifyAgentCallbackHandler
from core.callback_handler.index_tool_callback_handler import DatasetIndexToolCallbackHandler
from core.file.message_file_parser import MessageFileParser
from core.file import file_manager
from core.memory.token_buffer_memory import TokenBufferMemory
from core.model_manager import ModelInstance
from core.model_runtime.entities.llm_entities import LLMUsage
from core.model_runtime.entities.message_entities import (
from core.model_runtime.entities import (
AssistantPromptMessage,
LLMUsage,
PromptMessage,
PromptMessageContent,
PromptMessageTool,
SystemPromptMessage,
TextPromptMessageContent,
@ -40,9 +41,9 @@ from core.tools.entities.tool_entities import (
from core.tools.tool.dataset_retriever_tool import DatasetRetrieverTool
from core.tools.tool.tool import Tool
from core.tools.tool_manager import ToolManager
from core.tools.utils.tool_parameter_converter import ToolParameterConverter
from extensions.ext_database import db
from models.model import Conversation, Message, MessageAgentThought
from factories import file_factory
from models.model import Conversation, Message, MessageAgentThought, MessageFile
from models.tools import ToolConversationVariables
logger = logging.getLogger(__name__)
@ -66,23 +67,6 @@ class BaseAgentRunner(AppRunner):
db_variables: Optional[ToolConversationVariables] = None,
model_instance: ModelInstance = None,
) -> None:
"""
Agent runner
:param tenant_id: tenant id
:param application_generate_entity: application generate entity
:param conversation: conversation
:param app_config: app generate entity
:param model_config: model config
:param config: dataset config
:param queue_manager: queue manager
:param message: message
:param user_id: user id
:param memory: memory
:param prompt_messages: prompt messages
:param variables_pool: variables pool
:param db_variables: db variables
:param model_instance: model instance
"""
self.tenant_id = tenant_id
self.application_generate_entity = application_generate_entity
self.conversation = conversation
@ -180,7 +164,7 @@ class BaseAgentRunner(AppRunner):
if parameter.form != ToolParameter.ToolParameterForm.LLM:
continue
parameter_type = ToolParameterConverter.get_parameter_type(parameter.type)
parameter_type = parameter.type.as_normal_type()
enum = []
if parameter.type == ToolParameter.ToolParameterType.SELECT:
enum = [option.value for option in parameter.options]
@ -265,7 +249,7 @@ class BaseAgentRunner(AppRunner):
if parameter.form != ToolParameter.ToolParameterForm.LLM:
continue
parameter_type = ToolParameterConverter.get_parameter_type(parameter.type)
parameter_type = parameter.type.as_normal_type()
enum = []
if parameter.type == ToolParameter.ToolParameterType.SELECT:
enum = [option.value for option in parameter.options]
@ -511,26 +495,24 @@ class BaseAgentRunner(AppRunner):
return result
def organize_agent_user_prompt(self, message: Message) -> UserPromptMessage:
message_file_parser = MessageFileParser(
tenant_id=self.tenant_id,
app_id=self.app_config.app_id,
)
files = message.message_files
files = db.session.query(MessageFile).filter(MessageFile.message_id == message.id).all()
if files:
file_extra_config = FileUploadConfigManager.convert(message.app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.transform_message_files(files, file_extra_config)
file_objs = file_factory.build_from_message_files(
message_files=files, tenant_id=self.tenant_id, config=file_extra_config
)
else:
file_objs = []
if not file_objs:
return UserPromptMessage(content=message.query)
else:
prompt_message_contents = [TextPromptMessageContent(data=message.query)]
prompt_message_contents: list[PromptMessageContent] = []
prompt_message_contents.append(TextPromptMessageContent(data=message.query))
for file_obj in file_objs:
prompt_message_contents.append(file_obj.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file_obj))
return UserPromptMessage(content=prompt_message_contents)
else:

9
api/core/agent/cot_chat_agent_runner.py
View File

@ -1,9 +1,11 @@
import json
from core.agent.cot_agent_runner import CotAgentRunner
from core.model_runtime.entities.message_entities import (
from core.file import file_manager
from core.model_runtime.entities import (
AssistantPromptMessage,
PromptMessage,
PromptMessageContent,
SystemPromptMessage,
TextPromptMessageContent,
UserPromptMessage,
@ -32,9 +34,10 @@ class CotChatAgentRunner(CotAgentRunner):
Organize user query
"""
if self.files:
prompt_message_contents = [TextPromptMessageContent(data=query)]
prompt_message_contents: list[PromptMessageContent] = []
prompt_message_contents.append(TextPromptMessageContent(data=query))
for file_obj in self.files:
prompt_message_contents.append(file_obj.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file_obj))
prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
else:

14
api/core/agent/fc_agent_runner.py
View File

@ -7,10 +7,15 @@ from typing import Any, Optional, Union
from core.agent.base_agent_runner import BaseAgentRunner
from core.app.apps.base_app_queue_manager import PublishFrom
from core.app.entities.queue_entities import QueueAgentThoughtEvent, QueueMessageEndEvent, QueueMessageFileEvent
from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta, LLMUsage
from core.model_runtime.entities.message_entities import (
from core.file import file_manager
from core.model_runtime.entities import (
AssistantPromptMessage,
LLMResult,
LLMResultChunk,
LLMResultChunkDelta,
LLMUsage,
PromptMessage,
PromptMessageContent,
PromptMessageContentType,
SystemPromptMessage,
TextPromptMessageContent,
@ -390,9 +395,10 @@ class FunctionCallAgentRunner(BaseAgentRunner):
Organize user query
"""
if self.files:
prompt_message_contents = [TextPromptMessageContent(data=query)]
prompt_message_contents: list[PromptMessageContent] = []
prompt_message_contents.append(TextPromptMessageContent(data=query))
for file_obj in self.files:
prompt_message_contents.append(file_obj.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file_obj))
prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
else:

5
api/core/app/app_config/easy_ui_based_app/variables/manager.py
View File

@ -53,12 +53,11 @@ class BasicVariablesConfigManager:
VariableEntity(
type=variable_type,
variable=variable.get("variable"),
description=variable.get("description"),
description=variable.get("description") or "",
label=variable.get("label"),
required=variable.get("required", False),
max_length=variable.get("max_length"),
options=variable.get("options"),
default=variable.get("default"),
options=variable.get("options") or [],
)
)

32
api/core/app/app_config/entities.py
View File

@ -1,11 +1,12 @@
from collections.abc import Sequence
from enum import Enum
from typing import Any, Optional
from pydantic import BaseModel
from pydantic import BaseModel, Field, field_validator
from core.file.file_obj import FileExtraConfig
from core.file import FileExtraConfig, FileTransferMethod, FileType
from core.model_runtime.entities.message_entities import PromptMessageRole
from models import AppMode
from models.model import AppMode
class ModelConfigEntity(BaseModel):
@ -69,7 +70,7 @@ class PromptTemplateEntity(BaseModel):
ADVANCED = "advanced"
@classmethod
def value_of(cls, value: str) -> "PromptType":
def value_of(cls, value: str):
"""
Get value of given mode.
@ -93,6 +94,8 @@ class VariableEntityType(str, Enum):
PARAGRAPH = "paragraph"
NUMBER = "number"
EXTERNAL_DATA_TOOL = "external_data_tool"
FILE = "file"
FILE_LIST = "file-list"
class VariableEntity(BaseModel):
@ -102,13 +105,24 @@ class VariableEntity(BaseModel):
variable: str
label: str
description: Optional[str] = None
description: str = ""
type: VariableEntityType
required: bool = False
max_length: Optional[int] = None
options: Optional[list[str]] = None
default: Optional[str] = None
hint: Optional[str] = None
options: Sequence[str] = Field(default_factory=list)
allowed_file_types: Sequence[FileType] = Field(default_factory=list)
allowed_file_extensions: Sequence[str] = Field(default_factory=list)
allowed_file_upload_methods: Sequence[FileTransferMethod] = Field(default_factory=list)
@field_validator("description", mode="before")
@classmethod
def convert_none_description(cls, v: Any) -> str:
return v or ""
@field_validator("options", mode="before")
@classmethod
def convert_none_options(cls, v: Any) -> Sequence[str]:
return v or []
class ExternalDataVariableEntity(BaseModel):
@ -136,7 +150,7 @@ class DatasetRetrieveConfigEntity(BaseModel):
MULTIPLE = "multiple"
@classmethod
def value_of(cls, value: str) -> "RetrieveStrategy":
def value_of(cls, value: str):
"""
Get value of given mode.

55
api/core/app/app_config/features/file_upload/manager.py
View File

@ -1,12 +1,13 @@
from collections.abc import Mapping
from typing import Any, Optional
from typing import Any
from core.file.file_obj import FileExtraConfig
from core.file.models import FileExtraConfig
from models import FileUploadConfig
class FileUploadConfigManager:
@classmethod
def convert(cls, config: Mapping[str, Any], is_vision: bool = True) -> Optional[FileExtraConfig]:
def convert(cls, config: Mapping[str, Any], is_vision: bool = True):
"""
Convert model config to model config
@ -15,19 +16,21 @@ class FileUploadConfigManager:
"""
file_upload_dict = config.get("file_upload")
if file_upload_dict:
if file_upload_dict.get("image"):
if "enabled" in file_upload_dict["image"] and file_upload_dict["image"]["enabled"]:
image_config = {
"number_limits": file_upload_dict["image"]["number_limits"],
"transfer_methods": file_upload_dict["image"]["transfer_methods"],
if file_upload_dict.get("enabled"):
transform_methods = file_upload_dict.get("allowed_file_upload_methods") or file_upload_dict.get(
"allowed_upload_methods", []
)
data = {
"image_config": {
"number_limits": file_upload_dict["number_limits"],
"transfer_methods": transform_methods,
}
}
if is_vision:
image_config["detail"] = file_upload_dict["image"]["detail"]
if is_vision:
data["image_config"]["detail"] = file_upload_dict.get("image", {}).get("detail", "low")
return FileExtraConfig(image_config=image_config)
return None
return FileExtraConfig.model_validate(data)
@classmethod
def validate_and_set_defaults(cls, config: dict, is_vision: bool = True) -> tuple[dict, list[str]]:
@ -39,29 +42,7 @@ class FileUploadConfigManager:
"""
if not config.get("file_upload"):
config["file_upload"] = {}
if not isinstance(config["file_upload"], dict):
raise ValueError("file_upload must be of dict type")
# check image config
if not config["file_upload"].get("image"):
config["file_upload"]["image"] = {"enabled": False}
if config["file_upload"]["image"]["enabled"]:
number_limits = config["file_upload"]["image"]["number_limits"]
if number_limits < 1 or number_limits > 6:
raise ValueError("number_limits must be in [1, 6]")
if is_vision:
detail = config["file_upload"]["image"]["detail"]
if detail not in {"high", "low"}:
raise ValueError("detail must be in ['high', 'low']")
transfer_methods = config["file_upload"]["image"]["transfer_methods"]
if not isinstance(transfer_methods, list):
raise ValueError("transfer_methods must be of list type")
for method in transfer_methods:
if method not in {"remote_url", "local_file"}:
raise ValueError("transfer_methods must be in ['remote_url', 'local_file']")
else:
FileUploadConfig.model_validate(config["file_upload"])
return config, ["file_upload"]

2
api/core/app/app_config/workflow_ui_based_app/variables/manager.py
View File

@ -17,6 +17,6 @@ class WorkflowVariablesConfigManager:
# variables
for variable in user_input_form:
variables.append(VariableEntity(**variable))
variables.append(VariableEntity.model_validate(variable))
return variables

22
api/core/app/apps/advanced_chat/app_generator.py
View File

@ -21,11 +21,12 @@ from core.app.apps.message_based_app_generator import MessageBasedAppGenerator
from core.app.apps.message_based_app_queue_manager import MessageBasedAppQueueManager
from core.app.entities.app_invoke_entities import AdvancedChatAppGenerateEntity, InvokeFrom
from core.app.entities.task_entities import ChatbotAppBlockingResponse, ChatbotAppStreamResponse
from core.file.message_file_parser import MessageFileParser
from core.model_runtime.errors.invoke import InvokeAuthorizationError, InvokeError
from core.ops.ops_trace_manager import TraceQueueManager
from extensions.ext_database import db
from factories import file_factory
from models.account import Account
from models.enums import CreatedByRole
from models.model import App, Conversation, EndUser, Message
from models.workflow import Workflow
@ -96,10 +97,16 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
# parse files
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(workflow.features_dict, is_vision=False)
role = CreatedByRole.ACCOUNT if isinstance(user, Account) else CreatedByRole.END_USER
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
file_objs = file_factory.build_from_mappings(
mappings=files,
tenant_id=app_model.tenant_id,
user_id=user.id,
role=role,
config=file_extra_config,
)
else:
file_objs = []
@ -107,8 +114,9 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
app_config = AdvancedChatAppConfigManager.get_app_config(app_model=app_model, workflow=workflow)
# get tracing instance
user_id = user.id if isinstance(user, Account) else user.session_id
trace_manager = TraceQueueManager(app_model.id, user_id)
trace_manager = TraceQueueManager(
app_id=app_model.id, user_id=user.id if isinstance(user, Account) else user.session_id
)
if invoke_from == InvokeFrom.DEBUGGER:
# always enable retriever resource in debugger mode
@ -120,7 +128,9 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
task_id=str(uuid.uuid4()),
app_config=app_config,
conversation_id=conversation.id if conversation else None,
inputs=conversation.inputs if conversation else self._get_cleaned_inputs(inputs, app_config),
inputs=conversation.inputs
if conversation
else self._prepare_user_inputs(user_inputs=inputs, app_config=app_config, user_id=user.id, role=role),
query=query,
files=file_objs,
parent_message_id=args.get("parent_message_id") if invoke_from != InvokeFrom.SERVICE_API else UUID_NIL,

43
api/core/app/apps/advanced_chat/app_runner.py
View File

@ -1,31 +1,27 @@
import logging
import os
from collections.abc import Mapping
from typing import Any, cast
from sqlalchemy import select
from sqlalchemy.orm import Session
from configs import dify_config
from core.app.apps.advanced_chat.app_config_manager import AdvancedChatAppConfig
from core.app.apps.base_app_queue_manager import AppQueueManager
from core.app.apps.workflow_app_runner import WorkflowBasedAppRunner
from core.app.apps.workflow_logging_callback import WorkflowLoggingCallback
from core.app.entities.app_invoke_entities import (
AdvancedChatAppGenerateEntity,
InvokeFrom,
)
from core.app.entities.app_invoke_entities import AdvancedChatAppGenerateEntity, InvokeFrom
from core.app.entities.queue_entities import (
QueueAnnotationReplyEvent,
QueueStopEvent,
QueueTextChunkEvent,
)
from core.moderation.base import ModerationError
from core.workflow.callbacks.base_workflow_callback import WorkflowCallback
from core.workflow.entities.node_entities import UserFrom
from core.workflow.callbacks import WorkflowCallback, WorkflowLoggingCallback
from core.workflow.entities.variable_pool import VariablePool
from core.workflow.enums import SystemVariableKey
from core.workflow.workflow_entry import WorkflowEntry
from extensions.ext_database import db
from models.enums import UserFrom
from models.model import App, Conversation, EndUser, Message
from models.workflow import ConversationVariable, WorkflowType
@ -44,12 +40,6 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
conversation: Conversation,
message: Message,
) -> None:
"""
:param application_generate_entity: application generate entity
:param queue_manager: application queue manager
:param conversation: conversation
:param message: message
"""
super().__init__(queue_manager)
self.application_generate_entity = application_generate_entity
@ -57,10 +47,6 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
self.message = message
def run(self) -> None:
"""
Run application
:return:
"""
app_config = self.application_generate_entity.app_config
app_config = cast(AdvancedChatAppConfig, app_config)
@ -81,7 +67,7 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
user_id = self.application_generate_entity.user_id
workflow_callbacks: list[WorkflowCallback] = []
if bool(os.environ.get("DEBUG", "False").lower() == "true"):
if dify_config.DEBUG:
workflow_callbacks.append(WorkflowLoggingCallback())
if self.application_generate_entity.single_iteration_run:
@ -201,15 +187,6 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
query: str,
message_id: str,
) -> bool:
"""
Handle input moderation
:param app_record: app record
:param app_generate_entity: application generate entity
:param inputs: inputs
:param query: query
:param message_id: message id
:return:
"""
try:
# process sensitive_word_avoidance
_, inputs, query = self.moderation_for_inputs(
@ -229,14 +206,6 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
def handle_annotation_reply(
self, app_record: App, message: Message, query: str, app_generate_entity: AdvancedChatAppGenerateEntity
) -> bool:
"""
Handle annotation reply
:param app_record: app record
:param message: message
:param query: query
:param app_generate_entity: application generate entity
"""
# annotation reply
annotation_reply = self.query_app_annotations_to_reply(
app_record=app_record,
message=message,
@ -258,8 +227,6 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
def _complete_with_stream_output(self, text: str, stopped_by: QueueStopEvent.StopBy) -> None:
"""
Direct output
:param text: text
:return:
"""
self._publish_event(QueueTextChunkEvent(text=text))

36
api/core/app/apps/advanced_chat/generate_task_pipeline.py
View File

@ -1,7 +1,7 @@
import json
import logging
import time
from collections.abc import Generator
from collections.abc import Generator, Mapping
from typing import Any, Optional, Union
from constants.tts_auto_play_timeout import TTS_AUTO_PLAY_TIMEOUT, TTS_AUTO_PLAY_YIELD_CPU_TIME
@ -9,6 +9,7 @@ from core.app.apps.advanced_chat.app_generator_tts_publisher import AppGenerator
from core.app.apps.base_app_queue_manager import AppQueueManager, PublishFrom
from core.app.entities.app_invoke_entities import (
AdvancedChatAppGenerateEntity,
InvokeFrom,
)
from core.app.entities.queue_entities import (
QueueAdvancedChatMessageEndEvent,
@ -50,10 +51,12 @@ from core.model_runtime.utils.encoders import jsonable_encoder
from core.ops.ops_trace_manager import TraceQueueManager
from core.workflow.enums import SystemVariableKey
from core.workflow.graph_engine.entities.graph_runtime_state import GraphRuntimeState
from core.workflow.nodes import NodeType
from events.message_event import message_was_created
from extensions.ext_database import db
from models import Conversation, EndUser, Message, MessageFile
from models.account import Account
from models.model import Conversation, EndUser, Message
from models.enums import CreatedByRole
from models.workflow import (
Workflow,
WorkflowNodeExecution,
@ -120,6 +123,7 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
self._wip_workflow_node_executions = {}
self._conversation_name_generate_thread = None
self._recorded_files: list[Mapping[str, Any]] = []
def process(self):
"""
@ -298,6 +302,10 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
elif isinstance(event, QueueNodeSucceededEvent):
workflow_node_execution = self._handle_workflow_node_execution_success(event)
# Record files if it's an answer node or end node
if event.node_type in [NodeType.ANSWER, NodeType.END]:
self._recorded_files.extend(self._fetch_files_from_node_outputs(event.outputs or {}))
response = self._workflow_node_finish_to_stream_response(
event=event,
task_id=self._application_generate_entity.task_id,
@ -364,7 +372,7 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
start_at=graph_runtime_state.start_at,
total_tokens=graph_runtime_state.total_tokens,
total_steps=graph_runtime_state.node_run_steps,
outputs=json.dumps(event.outputs) if event.outputs else None,
outputs=event.outputs,
conversation_id=self._conversation.id,
trace_manager=trace_manager,
)
@ -490,10 +498,6 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
self._conversation_name_generate_thread.join()
def _save_message(self, graph_runtime_state: Optional[GraphRuntimeState] = None) -> None:
"""
Save message.
:return:
"""
self._refetch_message()
self._message.answer = self._task_state.answer
@ -501,6 +505,22 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
self._message.message_metadata = (
json.dumps(jsonable_encoder(self._task_state.metadata)) if self._task_state.metadata else None
)
message_files = [
MessageFile(
message_id=self._message.id,
type=file["type"],
transfer_method=file["transfer_method"],
url=file["remote_url"],
belongs_to="assistant",
upload_file_id=file["related_id"],
created_by_role=CreatedByRole.ACCOUNT
if self._message.invoke_from in {InvokeFrom.EXPLORE, InvokeFrom.DEBUGGER}
else CreatedByRole.END_USER,
created_by=self._message.from_account_id or self._message.from_end_user_id or "",
)
for file in self._recorded_files
]
db.session.add_all(message_files)
if graph_runtime_state and graph_runtime_state.llm_usage:
usage = graph_runtime_state.llm_usage
@ -540,7 +560,7 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
del extras["metadata"]["annotation_reply"]
return MessageEndStreamResponse(
task_id=self._application_generate_entity.task_id, id=self._message.id, **extras
task_id=self._application_generate_entity.task_id, id=self._message.id, files=self._recorded_files, **extras
)
def _handle_output_moderation_chunk(self, text: str) -> bool:

33
api/core/app/apps/agent_chat/app_generator.py
View File

@ -18,12 +18,12 @@ from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskSt
from core.app.apps.message_based_app_generator import MessageBasedAppGenerator
from core.app.apps.message_based_app_queue_manager import MessageBasedAppQueueManager
from core.app.entities.app_invoke_entities import AgentChatAppGenerateEntity, InvokeFrom
from core.file.message_file_parser import MessageFileParser
from core.model_runtime.errors.invoke import InvokeAuthorizationError, InvokeError
from core.ops.ops_trace_manager import TraceQueueManager
from extensions.ext_database import db
from models.account import Account
from models.model import App, EndUser
from factories import file_factory
from models import Account, App, EndUser
from models.enums import CreatedByRole
logger = logging.getLogger(__name__)
@ -50,7 +50,12 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
) -> dict: ...
def generate(
self, app_model: App, user: Union[Account, EndUser], args: Any, invoke_from: InvokeFrom, stream: bool = True
self,
app_model: App,
user: Union[Account, EndUser],
args: Any,
invoke_from: InvokeFrom,
stream: bool = True,
) -> Union[dict, Generator[dict, None, None]]:
"""
Generate App response.
@ -98,12 +103,19 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
# always enable retriever resource in debugger mode
override_model_config_dict["retriever_resource"] = {"enabled": True}
role = CreatedByRole.ACCOUNT if isinstance(user, Account) else CreatedByRole.END_USER
# parse files
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
files = args.get("files") or []
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
file_objs = file_factory.build_from_mappings(
mappings=files,
tenant_id=app_model.tenant_id,
user_id=user.id,
role=role,
config=file_extra_config,
)
else:
file_objs = []
@ -116,8 +128,7 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
)
# get tracing instance
user_id = user.id if isinstance(user, Account) else user.session_id
trace_manager = TraceQueueManager(app_model.id, user_id)
trace_manager = TraceQueueManager(app_model.id, user.id if isinstance(user, Account) else user.session_id)
# init application generate entity
application_generate_entity = AgentChatAppGenerateEntity(
@ -125,7 +136,9 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
app_config=app_config,
model_conf=ModelConfigConverter.convert(app_config),
conversation_id=conversation.id if conversation else None,
inputs=conversation.inputs if conversation else self._get_cleaned_inputs(inputs, app_config),
inputs=conversation.inputs
if conversation
else self._prepare_user_inputs(user_inputs=inputs, app_config=app_config, user_id=user.id, role=role),
query=query,
files=file_objs,
parent_message_id=args.get("parent_message_id") if invoke_from != InvokeFrom.SERVICE_API else UUID_NIL,

117
api/core/app/apps/base_app_generator.py
View File

@ -1,35 +1,92 @@
from collections.abc import Mapping
from typing import Any, Optional
from typing import TYPE_CHECKING, Any, Optional
from core.app.app_config.entities import AppConfig, VariableEntity, VariableEntityType
from core.app.app_config.entities import VariableEntityType
from core.file import File, FileExtraConfig
from factories import file_factory
if TYPE_CHECKING:
from core.app.app_config.entities import AppConfig, VariableEntity
from models.enums import CreatedByRole
class BaseAppGenerator:
def _get_cleaned_inputs(self, user_inputs: Optional[Mapping[str, Any]], app_config: AppConfig) -> Mapping[str, Any]:
def _prepare_user_inputs(
self,
*,
user_inputs: Optional[Mapping[str, Any]],
app_config: "AppConfig",
user_id: str,
role: "CreatedByRole",
) -> Mapping[str, Any]:
user_inputs = user_inputs or {}
# Filter input variables from form configuration, handle required fields, default values, and option values
variables = app_config.variables
filtered_inputs = {var.variable: self._validate_input(inputs=user_inputs, var=var) for var in variables}
filtered_inputs = {k: self._sanitize_value(v) for k, v in filtered_inputs.items()}
return filtered_inputs
user_inputs = {var.variable: self._validate_input(inputs=user_inputs, var=var) for var in variables}
user_inputs = {k: self._sanitize_value(v) for k, v in user_inputs.items()}
# Convert files in inputs to File
entity_dictionary = {item.variable: item for item in app_config.variables}
# Convert single file to File
files_inputs = {
k: file_factory.build_from_mapping(
mapping=v,
tenant_id=app_config.tenant_id,
user_id=user_id,
role=role,
config=FileExtraConfig(
allowed_file_types=entity_dictionary[k].allowed_file_types,
allowed_extensions=entity_dictionary[k].allowed_file_extensions,
allowed_upload_methods=entity_dictionary[k].allowed_file_upload_methods,
),
)
for k, v in user_inputs.items()
if isinstance(v, dict) and entity_dictionary[k].type == VariableEntityType.FILE
}
# Convert list of files to File
file_list_inputs = {
k: file_factory.build_from_mappings(
mappings=v,
tenant_id=app_config.tenant_id,
user_id=user_id,
role=role,
config=FileExtraConfig(
allowed_file_types=entity_dictionary[k].allowed_file_types,
allowed_extensions=entity_dictionary[k].allowed_file_extensions,
allowed_upload_methods=entity_dictionary[k].allowed_file_upload_methods,
),
)
for k, v in user_inputs.items()
if isinstance(v, list)
# Ensure skip List<File>
and all(isinstance(item, dict) for item in v)
and entity_dictionary[k].type == VariableEntityType.FILE_LIST
}
# Merge all inputs
user_inputs = {**user_inputs, **files_inputs, **file_list_inputs}
def _validate_input(self, *, inputs: Mapping[str, Any], var: VariableEntity):
user_input_value = inputs.get(var.variable)
if var.required and not user_input_value:
raise ValueError(f"{var.variable} is required in input form")
if not var.required and not user_input_value:
# TODO: should we return None here if the default value is None?
return var.default or ""
if (
var.type
in {
VariableEntityType.TEXT_INPUT,
VariableEntityType.SELECT,
VariableEntityType.PARAGRAPH,
}
and user_input_value
and not isinstance(user_input_value, str)
# Check if all files are converted to File
if any(filter(lambda v: isinstance(v, dict), user_inputs.values())):
raise ValueError("Invalid input type")
if any(
filter(lambda v: isinstance(v, dict), filter(lambda item: isinstance(item, list), user_inputs.values()))
):
raise ValueError("Invalid input type")
return user_inputs
def _validate_input(self, *, inputs: Mapping[str, Any], var: "VariableEntity"):
user_input_value = inputs.get(var.variable)
if not user_input_value:
if var.required:
raise ValueError(f"{var.variable} is required in input form")
else:
return None
if var.type in {
VariableEntityType.TEXT_INPUT,
VariableEntityType.SELECT,
VariableEntityType.PARAGRAPH,
} and not isinstance(user_input_value, str):
raise ValueError(f"(type '{var.type}') {var.variable} in input form must be a string")
if var.type == VariableEntityType.NUMBER and isinstance(user_input_value, str):
# may raise ValueError if user_input_value is not a valid number
@ -41,12 +98,24 @@ class BaseAppGenerator:
except ValueError:
raise ValueError(f"{var.variable} in input form must be a valid number")
if var.type == VariableEntityType.SELECT:
options = var.options or []
options = var.options
if user_input_value not in options:
raise ValueError(f"{var.variable} in input form must be one of the following: {options}")
elif var.type in {VariableEntityType.TEXT_INPUT, VariableEntityType.PARAGRAPH}:
if var.max_length and user_input_value and len(user_input_value) > var.max_length:
if var.max_length and len(user_input_value) > var.max_length:
raise ValueError(f"{var.variable} in input form must be less than {var.max_length} characters")
elif var.type == VariableEntityType.FILE:
if not isinstance(user_input_value, dict) and not isinstance(user_input_value, File):
raise ValueError(f"{var.variable} in input form must be a file")
elif var.type == VariableEntityType.FILE_LIST:
if not (
isinstance(user_input_value, list)
and (
all(isinstance(item, dict) for item in user_input_value)
or all(isinstance(item, File) for item in user_input_value)
)
):
raise ValueError(f"{var.variable} in input form must be a list of files")
return user_input_value

6
api/core/app/apps/base_app_runner.py
View File

@ -27,7 +27,7 @@ from core.prompt.simple_prompt_transform import ModelMode, SimplePromptTransform
from models.model import App, AppMode, Message, MessageAnnotation
if TYPE_CHECKING:
from core.file.file_obj import FileVar
from core.file.models import File
class AppRunner:
@ -37,7 +37,7 @@ class AppRunner:
model_config: ModelConfigWithCredentialsEntity,
prompt_template_entity: PromptTemplateEntity,
inputs: dict[str, str],
files: list["FileVar"],
files: list["File"],
query: Optional[str] = None,
) -> int:
"""
@ -137,7 +137,7 @@ class AppRunner:
model_config: ModelConfigWithCredentialsEntity,
prompt_template_entity: PromptTemplateEntity,
inputs: dict[str, str],
files: list["FileVar"],
files: list["File"],
query: Optional[str] = None,
context: Optional[str] = None,
memory: Optional[TokenBufferMemory] = None,

22
api/core/app/apps/chat/app_generator.py
View File

@ -18,11 +18,12 @@ from core.app.apps.chat.generate_response_converter import ChatAppGenerateRespon
from core.app.apps.message_based_app_generator import MessageBasedAppGenerator
from core.app.apps.message_based_app_queue_manager import MessageBasedAppQueueManager
from core.app.entities.app_invoke_entities import ChatAppGenerateEntity, InvokeFrom
from core.file.message_file_parser import MessageFileParser
from core.model_runtime.errors.invoke import InvokeAuthorizationError, InvokeError
from core.ops.ops_trace_manager import TraceQueueManager
from extensions.ext_database import db
from factories import file_factory
from models.account import Account
from models.enums import CreatedByRole
from models.model import App, EndUser
logger = logging.getLogger(__name__)
@ -100,12 +101,19 @@ class ChatAppGenerator(MessageBasedAppGenerator):
# always enable retriever resource in debugger mode
override_model_config_dict["retriever_resource"] = {"enabled": True}
role = CreatedByRole.ACCOUNT if isinstance(user, Account) else CreatedByRole.END_USER
# parse files
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
file_objs = file_factory.build_from_mappings(
mappings=files,
tenant_id=app_model.tenant_id,
user_id=user.id,
role=role,
config=file_extra_config,
)
else:
file_objs = []
@ -118,7 +126,7 @@ class ChatAppGenerator(MessageBasedAppGenerator):
)
# get tracing instance
trace_manager = TraceQueueManager(app_model.id)
trace_manager = TraceQueueManager(app_id=app_model.id)
# init application generate entity
application_generate_entity = ChatAppGenerateEntity(
@ -126,15 +134,17 @@ class ChatAppGenerator(MessageBasedAppGenerator):
app_config=app_config,
model_conf=ModelConfigConverter.convert(app_config),
conversation_id=conversation.id if conversation else None,
inputs=conversation.inputs if conversation else self._get_cleaned_inputs(inputs, app_config),
inputs=conversation.inputs
if conversation
else self._prepare_user_inputs(user_inputs=inputs, app_config=app_config, user_id=user.id, role=role),
query=query,
files=file_objs,
parent_message_id=args.get("parent_message_id") if invoke_from != InvokeFrom.SERVICE_API else UUID_NIL,
user_id=user.id,
stream=stream,
invoke_from=invoke_from,
extras=extras,
trace_manager=trace_manager,
stream=stream,
)
# init generate records

32
api/core/app/apps/completion/app_generator.py
View File

@ -17,12 +17,12 @@ from core.app.apps.completion.generate_response_converter import CompletionAppGe
from core.app.apps.message_based_app_generator import MessageBasedAppGenerator
from core.app.apps.message_based_app_queue_manager import MessageBasedAppQueueManager
from core.app.entities.app_invoke_entities import CompletionAppGenerateEntity, InvokeFrom
from core.file.message_file_parser import MessageFileParser
from core.model_runtime.errors.invoke import InvokeAuthorizationError, InvokeError
from core.ops.ops_trace_manager import TraceQueueManager
from extensions.ext_database import db
from models.account import Account
from models.model import App, EndUser, Message
from factories import file_factory
from models import Account, App, EndUser, Message
from models.enums import CreatedByRole
from services.errors.app import MoreLikeThisDisabledError
from services.errors.message import MessageNotExistsError
@ -88,12 +88,19 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
tenant_id=app_model.tenant_id, config=args.get("model_config")
)
role = CreatedByRole.ACCOUNT if isinstance(user, Account) else CreatedByRole.END_USER
# parse files
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
file_objs = file_factory.build_from_mappings(
mappings=files,
tenant_id=app_model.tenant_id,
user_id=user.id,
role=role,
config=file_extra_config,
)
else:
file_objs = []
@ -103,6 +110,7 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
)
# get tracing instance
user_id = user.id if isinstance(user, Account) else user.session_id
trace_manager = TraceQueueManager(app_model.id)
# init application generate entity
@ -110,7 +118,7 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
task_id=str(uuid.uuid4()),
app_config=app_config,
model_conf=ModelConfigConverter.convert(app_config),
inputs=self._get_cleaned_inputs(inputs, app_config),
inputs=self._prepare_user_inputs(user_inputs=inputs, app_config=app_config, user_id=user.id, role=role),
query=query,
files=file_objs,
user_id=user.id,
@ -251,10 +259,16 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
override_model_config_dict["model"] = model_dict
# parse files
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
role = CreatedByRole.ACCOUNT if isinstance(user, Account) else CreatedByRole.END_USER
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict)
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(message.files, file_extra_config, user)
file_objs = file_factory.build_from_mappings(
mappings=message.message_files,
tenant_id=app_model.tenant_id,
user_id=user.id,
role=role,
config=file_extra_config,
)
else:
file_objs = []

13
api/core/app/apps/message_based_app_generator.py
View File

@ -26,7 +26,8 @@ from core.app.entities.task_entities import (
from core.app.task_pipeline.easy_ui_based_generate_task_pipeline import EasyUIBasedGenerateTaskPipeline
from core.prompt.utils.prompt_template_parser import PromptTemplateParser
from extensions.ext_database import db
from models.account import Account
from models import Account
from models.enums import CreatedByRole
from models.model import App, AppMode, AppModelConfig, Conversation, EndUser, Message, MessageFile
from services.errors.app_model_config import AppModelConfigBrokenError
from services.errors.conversation import ConversationCompletedError, ConversationNotExistsError
@ -235,13 +236,13 @@ class MessageBasedAppGenerator(BaseAppGenerator):
for file in application_generate_entity.files:
message_file = MessageFile(
message_id=message.id,
type=file.type.value,
transfer_method=file.transfer_method.value,
type=file.type,
transfer_method=file.transfer_method,
belongs_to="user",
url=file.url,
url=file.remote_url,
upload_file_id=file.related_id,
created_by_role=("account" if account_id else "end_user"),
created_by=account_id or end_user_id,
created_by_role=(CreatedByRole.ACCOUNT if account_id else CreatedByRole.END_USER),
created_by=account_id or end_user_id or "",
)
db.session.add(message_file)
db.session.commit()

54
api/core/app/apps/workflow/app_generator.py
View File

@ -3,7 +3,7 @@ import logging
import os
import threading
import uuid
from collections.abc import Generator
from collections.abc import Generator, Mapping, Sequence
from typing import Any, Literal, Optional, Union, overload
from flask import Flask, current_app
@ -20,13 +20,12 @@ from core.app.apps.workflow.generate_response_converter import WorkflowAppGenera
from core.app.apps.workflow.generate_task_pipeline import WorkflowAppGenerateTaskPipeline
from core.app.entities.app_invoke_entities import InvokeFrom, WorkflowAppGenerateEntity
from core.app.entities.task_entities import WorkflowAppBlockingResponse, WorkflowAppStreamResponse
from core.file.message_file_parser import MessageFileParser
from core.model_runtime.errors.invoke import InvokeAuthorizationError, InvokeError
from core.ops.ops_trace_manager import TraceQueueManager
from extensions.ext_database import db
from models.account import Account
from models.model import App, EndUser
from models.workflow import Workflow
from factories import file_factory
from models import Account, App, EndUser, Workflow
from models.enums import CreatedByRole
logger = logging.getLogger(__name__)
@ -63,49 +62,46 @@ class WorkflowAppGenerator(BaseAppGenerator):
app_model: App,
workflow: Workflow,
user: Union[Account, EndUser],
args: dict,
args: Mapping[str, Any],
invoke_from: InvokeFrom,
stream: bool = True,
call_depth: int = 0,
workflow_thread_pool_id: Optional[str] = None,
):
"""
Generate App response.
files: Sequence[Mapping[str, Any]] = args.get("files") or []
:param app_model: App
:param workflow: Workflow
:param user: account or end user
:param args: request args
:param invoke_from: invoke from source
:param stream: is stream
:param call_depth: call depth
:param workflow_thread_pool_id: workflow thread pool id
"""
inputs = args["inputs"]
role = CreatedByRole.ACCOUNT if isinstance(user, Account) else CreatedByRole.END_USER
# parse files
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(workflow.features_dict, is_vision=False)
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
else:
file_objs = []
system_files = file_factory.build_from_mappings(
mappings=files,
tenant_id=app_model.tenant_id,
user_id=user.id,
role=role,
config=file_extra_config,
)
# convert to app config
app_config = WorkflowAppConfigManager.get_app_config(app_model=app_model, workflow=workflow)
app_config = WorkflowAppConfigManager.get_app_config(
app_model=app_model,
workflow=workflow,
)
# get tracing instance
user_id = user.id if isinstance(user, Account) else user.session_id
trace_manager = TraceQueueManager(app_model.id, user_id)
trace_manager = TraceQueueManager(
app_id=app_model.id,
user_id=user.id if isinstance(user, Account) else user.session_id,
)
inputs: Mapping[str, Any] = args["inputs"]
workflow_run_id = str(uuid.uuid4())
# init application generate entity
application_generate_entity = WorkflowAppGenerateEntity(
task_id=str(uuid.uuid4()),
app_config=app_config,
inputs=self._get_cleaned_inputs(inputs, app_config),
files=file_objs,
inputs=self._prepare_user_inputs(user_inputs=inputs, app_config=app_config, user_id=user.id, role=role),
files=system_files,
user_id=user.id,
stream=stream,
invoke_from=invoke_from,

9
api/core/app/apps/workflow/app_runner.py
View File

@ -1,21 +1,20 @@
import logging
import os
from typing import Optional, cast
from configs import dify_config
from core.app.apps.base_app_queue_manager import AppQueueManager
from core.app.apps.workflow.app_config_manager import WorkflowAppConfig
from core.app.apps.workflow_app_runner import WorkflowBasedAppRunner
from core.app.apps.workflow_logging_callback import WorkflowLoggingCallback
from core.app.entities.app_invoke_entities import (
InvokeFrom,
WorkflowAppGenerateEntity,
)
from core.workflow.callbacks.base_workflow_callback import WorkflowCallback
from core.workflow.entities.node_entities import UserFrom
from core.workflow.callbacks import WorkflowCallback, WorkflowLoggingCallback
from core.workflow.entities.variable_pool import VariablePool
from core.workflow.enums import SystemVariableKey
from core.workflow.workflow_entry import WorkflowEntry
from extensions.ext_database import db
from models.enums import UserFrom
from models.model import App, EndUser
from models.workflow import WorkflowType
@ -71,7 +70,7 @@ class WorkflowAppRunner(WorkflowBasedAppRunner):
db.session.close()
workflow_callbacks: list[WorkflowCallback] = []
if bool(os.environ.get("DEBUG", "False").lower() == "true"):
if dify_config.DEBUG:
workflow_callbacks.append(WorkflowLoggingCallback())
# if only single iteration run is requested

5
api/core/app/apps/workflow/generate_task_pipeline.py
View File

@ -1,4 +1,3 @@
import json
import logging
import time
from collections.abc import Generator
@ -334,9 +333,7 @@ class WorkflowAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCycleMa
start_at=graph_runtime_state.start_at,
total_tokens=graph_runtime_state.total_tokens,
total_steps=graph_runtime_state.node_run_steps,
outputs=json.dumps(event.outputs)
if isinstance(event, QueueWorkflowSucceededEvent) and event.outputs
else None,
outputs=event.outputs,
conversation_id=None,
trace_manager=trace_manager,
)

12
api/core/app/apps/workflow_app_runner.py
View File

@ -20,7 +20,6 @@ from core.app.entities.queue_entities import (
QueueWorkflowStartedEvent,
QueueWorkflowSucceededEvent,
)
from core.workflow.entities.node_entities import NodeType
from core.workflow.entities.variable_pool import VariablePool
from core.workflow.graph_engine.entities.event import (
GraphEngineEvent,
@ -41,9 +40,9 @@ from core.workflow.graph_engine.entities.event import (
ParallelBranchRunSucceededEvent,
)
from core.workflow.graph_engine.entities.graph import Graph
from core.workflow.nodes.base_node import BaseNode
from core.workflow.nodes.iteration.entities import IterationNodeData
from core.workflow.nodes.node_mapping import node_classes
from core.workflow.nodes import NodeType
from core.workflow.nodes.iteration import IterationNodeData
from core.workflow.nodes.node_mapping import node_type_classes_mapping
from core.workflow.workflow_entry import WorkflowEntry
from extensions.ext_database import db
from models.model import App
@ -137,9 +136,8 @@ class WorkflowBasedAppRunner(AppRunner):
raise ValueError("iteration node id not found in workflow graph")
# Get node class
node_type = NodeType.value_of(iteration_node_config.get("data", {}).get("type"))
node_cls = node_classes.get(node_type)
node_cls = cast(type[BaseNode], node_cls)
node_type = NodeType(iteration_node_config.get("data", {}).get("type"))
node_cls = node_type_classes_mapping[node_type]
# init variable pool
variable_pool = VariablePool(

220
api/core/app/apps/workflow_logging_callback.py
View File

@ -1,220 +0,0 @@
from typing import Optional
from core.model_runtime.utils.encoders import jsonable_encoder
from core.workflow.callbacks.base_workflow_callback import WorkflowCallback
from core.workflow.graph_engine.entities.event import (
GraphEngineEvent,
GraphRunFailedEvent,
GraphRunStartedEvent,
GraphRunSucceededEvent,
IterationRunFailedEvent,
IterationRunNextEvent,
IterationRunStartedEvent,
IterationRunSucceededEvent,
NodeRunFailedEvent,
NodeRunStartedEvent,
NodeRunStreamChunkEvent,
NodeRunSucceededEvent,
ParallelBranchRunFailedEvent,
ParallelBranchRunStartedEvent,
ParallelBranchRunSucceededEvent,
)
_TEXT_COLOR_MAPPING = {
"blue": "36;1",
"yellow": "33;1",
"pink": "38;5;200",
"green": "32;1",
"red": "31;1",
}
class WorkflowLoggingCallback(WorkflowCallback):
def __init__(self) -> None:
self.current_node_id = None
def on_event(self, event: GraphEngineEvent) -> None:
if isinstance(event, GraphRunStartedEvent):
self.print_text("\n[GraphRunStartedEvent]", color="pink")
elif isinstance(event, GraphRunSucceededEvent):
self.print_text("\n[GraphRunSucceededEvent]", color="green")
elif isinstance(event, GraphRunFailedEvent):
self.print_text(f"\n[GraphRunFailedEvent] reason: {event.error}", color="red")
elif isinstance(event, NodeRunStartedEvent):
self.on_workflow_node_execute_started(event=event)
elif isinstance(event, NodeRunSucceededEvent):
self.on_workflow_node_execute_succeeded(event=event)
elif isinstance(event, NodeRunFailedEvent):
self.on_workflow_node_execute_failed(event=event)
elif isinstance(event, NodeRunStreamChunkEvent):
self.on_node_text_chunk(event=event)
elif isinstance(event, ParallelBranchRunStartedEvent):
self.on_workflow_parallel_started(event=event)
elif isinstance(event, ParallelBranchRunSucceededEvent | ParallelBranchRunFailedEvent):
self.on_workflow_parallel_completed(event=event)
elif isinstance(event, IterationRunStartedEvent):
self.on_workflow_iteration_started(event=event)
elif isinstance(event, IterationRunNextEvent):
self.on_workflow_iteration_next(event=event)
elif isinstance(event, IterationRunSucceededEvent | IterationRunFailedEvent):
self.on_workflow_iteration_completed(event=event)
else:
self.print_text(f"\n[{event.__class__.__name__}]", color="blue")
def on_workflow_node_execute_started(self, event: NodeRunStartedEvent) -> None:
"""
Workflow node execute started
"""
self.print_text("\n[NodeRunStartedEvent]", color="yellow")
self.print_text(f"Node ID: {event.node_id}", color="yellow")
self.print_text(f"Node Title: {event.node_data.title}", color="yellow")
self.print_text(f"Type: {event.node_type.value}", color="yellow")
def on_workflow_node_execute_succeeded(self, event: NodeRunSucceededEvent) -> None:
"""
Workflow node execute succeeded
"""
route_node_state = event.route_node_state
self.print_text("\n[NodeRunSucceededEvent]", color="green")
self.print_text(f"Node ID: {event.node_id}", color="green")
self.print_text(f"Node Title: {event.node_data.title}", color="green")
self.print_text(f"Type: {event.node_type.value}", color="green")
if route_node_state.node_run_result:
node_run_result = route_node_state.node_run_result
self.print_text(
f"Inputs: {jsonable_encoder(node_run_result.inputs) if node_run_result.inputs else ''}",
color="green",
)
self.print_text(
f"Process Data: "
f"{jsonable_encoder(node_run_result.process_data) if node_run_result.process_data else ''}",
color="green",
)
self.print_text(
f"Outputs: {jsonable_encoder(node_run_result.outputs) if node_run_result.outputs else ''}",
color="green",
)
self.print_text(
f"Metadata: {jsonable_encoder(node_run_result.metadata) if node_run_result.metadata else ''}",
color="green",
)
def on_workflow_node_execute_failed(self, event: NodeRunFailedEvent) -> None:
"""
Workflow node execute failed
"""
route_node_state = event.route_node_state
self.print_text("\n[NodeRunFailedEvent]", color="red")
self.print_text(f"Node ID: {event.node_id}", color="red")
self.print_text(f"Node Title: {event.node_data.title}", color="red")
self.print_text(f"Type: {event.node_type.value}", color="red")
if route_node_state.node_run_result:
node_run_result = route_node_state.node_run_result
self.print_text(f"Error: {node_run_result.error}", color="red")
self.print_text(
f"Inputs: {jsonable_encoder(node_run_result.inputs) if node_run_result.inputs else ''}",
color="red",
)
self.print_text(
f"Process Data: "
f"{jsonable_encoder(node_run_result.process_data) if node_run_result.process_data else ''}",
color="red",
)
self.print_text(
f"Outputs: {jsonable_encoder(node_run_result.outputs) if node_run_result.outputs else ''}",
color="red",
)
def on_node_text_chunk(self, event: NodeRunStreamChunkEvent) -> None:
"""
Publish text chunk
"""
route_node_state = event.route_node_state
if not self.current_node_id or self.current_node_id != route_node_state.node_id:
self.current_node_id = route_node_state.node_id
self.print_text("\n[NodeRunStreamChunkEvent]")
self.print_text(f"Node ID: {route_node_state.node_id}")
node_run_result = route_node_state.node_run_result
if node_run_result:
self.print_text(
f"Metadata: {jsonable_encoder(node_run_result.metadata) if node_run_result.metadata else ''}"
)
self.print_text(event.chunk_content, color="pink", end="")
def on_workflow_parallel_started(self, event: ParallelBranchRunStartedEvent) -> None:
"""
Publish parallel started
"""
self.print_text("\n[ParallelBranchRunStartedEvent]", color="blue")
self.print_text(f"Parallel ID: {event.parallel_id}", color="blue")
self.print_text(f"Branch ID: {event.parallel_start_node_id}", color="blue")
if event.in_iteration_id:
self.print_text(f"Iteration ID: {event.in_iteration_id}", color="blue")
def on_workflow_parallel_completed(
self, event: ParallelBranchRunSucceededEvent | ParallelBranchRunFailedEvent
) -> None:
"""
Publish parallel completed
"""
if isinstance(event, ParallelBranchRunSucceededEvent):
color = "blue"
elif isinstance(event, ParallelBranchRunFailedEvent):
color = "red"
self.print_text(
"\n[ParallelBranchRunSucceededEvent]"
if isinstance(event, ParallelBranchRunSucceededEvent)
else "\n[ParallelBranchRunFailedEvent]",
color=color,
)
self.print_text(f"Parallel ID: {event.parallel_id}", color=color)
self.print_text(f"Branch ID: {event.parallel_start_node_id}", color=color)
if event.in_iteration_id:
self.print_text(f"Iteration ID: {event.in_iteration_id}", color=color)
if isinstance(event, ParallelBranchRunFailedEvent):
self.print_text(f"Error: {event.error}", color=color)
def on_workflow_iteration_started(self, event: IterationRunStartedEvent) -> None:
"""
Publish iteration started
"""
self.print_text("\n[IterationRunStartedEvent]", color="blue")
self.print_text(f"Iteration Node ID: {event.iteration_id}", color="blue")
def on_workflow_iteration_next(self, event: IterationRunNextEvent) -> None:
"""
Publish iteration next
"""
self.print_text("\n[IterationRunNextEvent]", color="blue")
self.print_text(f"Iteration Node ID: {event.iteration_id}", color="blue")
self.print_text(f"Iteration Index: {event.index}", color="blue")
def on_workflow_iteration_completed(self, event: IterationRunSucceededEvent | IterationRunFailedEvent) -> None:
"""
Publish iteration completed
"""
self.print_text(
"\n[IterationRunSucceededEvent]"
if isinstance(event, IterationRunSucceededEvent)
else "\n[IterationRunFailedEvent]",
color="blue",
)
self.print_text(f"Node ID: {event.iteration_id}", color="blue")
def print_text(self, text: str, color: Optional[str] = None, end: str = "\n") -> None:
"""Print text with highlighting and no end characters."""
text_to_print = self._get_colored_text(text, color) if color else text
print(f"{text_to_print}", end=end)
def _get_colored_text(self, text: str, color: str) -> str:
"""Get colored text."""
color_str = _TEXT_COLOR_MAPPING[color]
return f"\u001b[{color_str}m\033[1;3m{text}\u001b[0m"

8
api/core/app/entities/app_invoke_entities.py
View File

@ -1,4 +1,4 @@
from collections.abc import Mapping
from collections.abc import Mapping, Sequence
from enum import Enum
from typing import Any, Optional
@ -7,7 +7,7 @@ from pydantic import BaseModel, ConfigDict, Field, ValidationInfo, field_validat
from constants import UUID_NIL
from core.app.app_config.entities import AppConfig, EasyUIBasedAppConfig, WorkflowUIBasedAppConfig
from core.entities.provider_configuration import ProviderModelBundle
from core.file.file_obj import FileVar
from core.file.models import File
from core.model_runtime.entities.model_entities import AIModelEntity
from core.ops.ops_trace_manager import TraceQueueManager
@ -23,7 +23,7 @@ class InvokeFrom(Enum):
DEBUGGER = "debugger"
@classmethod
def value_of(cls, value: str) -> "InvokeFrom":
def value_of(cls, value: str):
"""
Get value of given mode.
@ -82,7 +82,7 @@ class AppGenerateEntity(BaseModel):
app_config: AppConfig
inputs: Mapping[str, Any]
files: list[FileVar] = []
files: Sequence[File]
user_id: str
# extras

5
api/core/app/entities/queue_entities.py
View File

@ -5,9 +5,10 @@ from typing import Any, Optional
from pydantic import BaseModel, field_validator
from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk
from core.workflow.entities.base_node_data_entities import BaseNodeData
from core.workflow.entities.node_entities import NodeRunMetadataKey, NodeType
from core.workflow.entities.node_entities import NodeRunMetadataKey
from core.workflow.graph_engine.entities.graph_runtime_state import GraphRuntimeState
from core.workflow.nodes import NodeType
from core.workflow.nodes.base import BaseNodeData
class QueueEvent(str, Enum):

6
api/core/app/entities/task_entities.py
View File

@ -1,3 +1,4 @@
from collections.abc import Mapping, Sequence
from enum import Enum
from typing import Any, Optional
@ -119,6 +120,7 @@ class MessageEndStreamResponse(StreamResponse):
event: StreamEvent = StreamEvent.MESSAGE_END
id: str
metadata: dict = {}
files: Optional[Sequence[Mapping[str, Any]]] = None
class MessageFileStreamResponse(StreamResponse):
@ -211,7 +213,7 @@ class WorkflowFinishStreamResponse(StreamResponse):
created_by: Optional[dict] = None
created_at: int
finished_at: int
files: Optional[list[dict]] = []
files: Optional[Sequence[Mapping[str, Any]]] = []
event: StreamEvent = StreamEvent.WORKFLOW_FINISHED
workflow_run_id: str
@ -296,7 +298,7 @@ class NodeFinishStreamResponse(StreamResponse):
execution_metadata: Optional[dict] = None
created_at: int
finished_at: int
files: Optional[list[dict]] = []
files: Optional[Sequence[Mapping[str, Any]]] = []
parallel_id: Optional[str] = None
parallel_start_node_id: Optional[str] = None
parent_parallel_id: Optional[str] = None

49
api/core/app/segments/__init__.py
View File

@ -1,49 +0,0 @@
from .segment_group import SegmentGroup
from .segments import (
ArrayAnySegment,
ArraySegment,
FloatSegment,
IntegerSegment,
NoneSegment,
ObjectSegment,
Segment,
StringSegment,
)
from .types import SegmentType
from .variables import (
ArrayAnyVariable,
ArrayNumberVariable,
ArrayObjectVariable,
ArrayStringVariable,
FloatVariable,
IntegerVariable,
NoneVariable,
ObjectVariable,
SecretVariable,
StringVariable,
Variable,
)
__all__ = [
"IntegerVariable",
"FloatVariable",
"ObjectVariable",
"SecretVariable",
"StringVariable",
"ArrayAnyVariable",
"Variable",
"SegmentType",
"SegmentGroup",
"Segment",
"NoneSegment",
"NoneVariable",
"IntegerSegment",
"FloatSegment",
"ObjectSegment",
"ArrayAnySegment",
"StringSegment",
"ArrayStringVariable",
"ArrayNumberVariable",
"ArrayObjectVariable",
"ArraySegment",
]

2
api/core/app/segments/exc.py
View File

@ -1,2 +0,0 @@
class VariableError(ValueError):
pass

76
api/core/app/segments/factory.py
View File

@ -1,76 +0,0 @@
from collections.abc import Mapping
from typing import Any
from configs import dify_config
from .exc import VariableError
from .segments import (
ArrayAnySegment,
FloatSegment,
IntegerSegment,
NoneSegment,
ObjectSegment,
Segment,
StringSegment,
)
from .types import SegmentType
from .variables import (
ArrayNumberVariable,
ArrayObjectVariable,
ArrayStringVariable,
FloatVariable,
IntegerVariable,
ObjectVariable,
SecretVariable,
StringVariable,
Variable,
)
def build_variable_from_mapping(mapping: Mapping[str, Any], /) -> Variable:
if (value_type := mapping.get("value_type")) is None:
raise VariableError("missing value type")
if not mapping.get("name"):
raise VariableError("missing name")
if (value := mapping.get("value")) is None:
raise VariableError("missing value")
match value_type:
case SegmentType.STRING:
result = StringVariable.model_validate(mapping)
case SegmentType.SECRET:
result = SecretVariable.model_validate(mapping)
case SegmentType.NUMBER if isinstance(value, int):
result = IntegerVariable.model_validate(mapping)
case SegmentType.NUMBER if isinstance(value, float):
result = FloatVariable.model_validate(mapping)
case SegmentType.NUMBER if not isinstance(value, float | int):
raise VariableError(f"invalid number value {value}")
case SegmentType.OBJECT if isinstance(value, dict):
result = ObjectVariable.model_validate(mapping)
case SegmentType.ARRAY_STRING if isinstance(value, list):
result = ArrayStringVariable.model_validate(mapping)
case SegmentType.ARRAY_NUMBER if isinstance(value, list):
result = ArrayNumberVariable.model_validate(mapping)
case SegmentType.ARRAY_OBJECT if isinstance(value, list):
result = ArrayObjectVariable.model_validate(mapping)
case _:
raise VariableError(f"not supported value type {value_type}")
if result.size > dify_config.MAX_VARIABLE_SIZE:
raise VariableError(f"variable size {result.size} exceeds limit {dify_config.MAX_VARIABLE_SIZE}")
return result
def build_segment(value: Any, /) -> Segment:
if value is None:
return NoneSegment()
if isinstance(value, str):
return StringSegment(value=value)
if isinstance(value, int):
return IntegerSegment(value=value)
if isinstance(value, float):
return FloatSegment(value=value)
if isinstance(value, dict):
return ObjectSegment(value=value)
if isinstance(value, list):
return ArrayAnySegment(value=value)
raise ValueError(f"not supported value {value}")

18
api/core/app/segments/parser.py
View File

@ -1,18 +0,0 @@
import re
from core.workflow.entities.variable_pool import VariablePool
from . import SegmentGroup, factory
VARIABLE_PATTERN = re.compile(r"\{\{#([a-zA-Z0-9_]{1,50}(?:\.[a-zA-Z_][a-zA-Z0-9_]{0,29}){1,10})#\}\}")
def convert_template(*, template: str, variable_pool: VariablePool):
parts = re.split(VARIABLE_PATTERN, template)
segments = []
for part in filter(lambda x: x, parts):
if "." in part and (value := variable_pool.get(part.split("."))):
segments.append(value)
else:
segments.append(factory.build_segment(part))
return SegmentGroup(value=segments)

22
api/core/app/segments/segment_group.py
View File

@ -1,22 +0,0 @@
from .segments import Segment
from .types import SegmentType
class SegmentGroup(Segment):
value_type: SegmentType = SegmentType.GROUP
value: list[Segment]
@property
def text(self):
return "".join([segment.text for segment in self.value])
@property
def log(self):
return "".join([segment.log for segment in self.value])
@property
def markdown(self):
return "".join([segment.markdown for segment in self.value])
def to_object(self):
return [segment.to_object() for segment in self.value]

126
api/core/app/segments/segments.py
View File

@ -1,126 +0,0 @@
import json
import sys
from collections.abc import Mapping, Sequence
from typing import Any
from pydantic import BaseModel, ConfigDict, field_validator
from .types import SegmentType
class Segment(BaseModel):
model_config = ConfigDict(frozen=True)
value_type: SegmentType
value: Any
@field_validator("value_type")
@classmethod
def validate_value_type(cls, value):
"""
This validator checks if the provided value is equal to the default value of the 'value_type' field.
If the value is different, a ValueError is raised.
"""
if value != cls.model_fields["value_type"].default:
raise ValueError("Cannot modify 'value_type'")
return value
@property
def text(self) -> str:
return str(self.value)
@property
def log(self) -> str:
return str(self.value)
@property
def markdown(self) -> str:
return str(self.value)
@property
def size(self) -> int:
return sys.getsizeof(self.value)
def to_object(self) -> Any:
return self.value
class NoneSegment(Segment):
value_type: SegmentType = SegmentType.NONE
value: None = None
@property
def text(self) -> str:
return "null"
@property
def log(self) -> str:
return "null"
@property
def markdown(self) -> str:
return "null"
class StringSegment(Segment):
value_type: SegmentType = SegmentType.STRING
value: str
class FloatSegment(Segment):
value_type: SegmentType = SegmentType.NUMBER
value: float
class IntegerSegment(Segment):
value_type: SegmentType = SegmentType.NUMBER
value: int
class ObjectSegment(Segment):
value_type: SegmentType = SegmentType.OBJECT
value: Mapping[str, Any]
@property
def text(self) -> str:
return json.dumps(self.model_dump()["value"], ensure_ascii=False)
@property
def log(self) -> str:
return json.dumps(self.model_dump()["value"], ensure_ascii=False, indent=2)
@property
def markdown(self) -> str:
return json.dumps(self.model_dump()["value"], ensure_ascii=False, indent=2)
class ArraySegment(Segment):
@property
def markdown(self) -> str:
items = []
for item in self.value:
if hasattr(item, "to_markdown"):
items.append(item.to_markdown())
else:
items.append(str(item))
return "\n".join(items)
class ArrayAnySegment(ArraySegment):
value_type: SegmentType = SegmentType.ARRAY_ANY
value: Sequence[Any]
class ArrayStringSegment(ArraySegment):
value_type: SegmentType = SegmentType.ARRAY_STRING
value: Sequence[str]
class ArrayNumberSegment(ArraySegment):
value_type: SegmentType = SegmentType.ARRAY_NUMBER
value: Sequence[float | int]
class ArrayObjectSegment(ArraySegment):
value_type: SegmentType = SegmentType.ARRAY_OBJECT
value: Sequence[Mapping[str, Any]]

15
api/core/app/segments/types.py
View File

@ -1,15 +0,0 @@
from enum import Enum
class SegmentType(str, Enum):
NONE = "none"
NUMBER = "number"
STRING = "string"
SECRET = "secret"
ARRAY_ANY = "array[any]"
ARRAY_STRING = "array[string]"
ARRAY_NUMBER = "array[number]"
ARRAY_OBJECT = "array[object]"
OBJECT = "object"
GROUP = "group"

75
api/core/app/segments/variables.py
View File

@ -1,75 +0,0 @@
from pydantic import Field
from core.helper import encrypter
from .segments import (
ArrayAnySegment,
ArrayNumberSegment,
ArrayObjectSegment,
ArrayStringSegment,
FloatSegment,
IntegerSegment,
NoneSegment,
ObjectSegment,
Segment,
StringSegment,
)
from .types import SegmentType
class Variable(Segment):
"""
A variable is a segment that has a name.
"""
id: str = Field(
default="",
description="Unique identity for variable. It's only used by environment variables now.",
)
name: str
description: str = Field(default="", description="Description of the variable.")
class StringVariable(StringSegment, Variable):
pass
class FloatVariable(FloatSegment, Variable):
pass
class IntegerVariable(IntegerSegment, Variable):
pass
class ObjectVariable(ObjectSegment, Variable):
pass
class ArrayAnyVariable(ArrayAnySegment, Variable):
pass
class ArrayStringVariable(ArrayStringSegment, Variable):
pass
class ArrayNumberVariable(ArrayNumberSegment, Variable):
pass
class ArrayObjectVariable(ArrayObjectSegment, Variable):
pass
class SecretVariable(StringVariable):
value_type: SegmentType = SegmentType.SECRET
@property
def log(self) -> str:
return encrypter.obfuscated_token(self.value)
class NoneVariable(NoneSegment, Variable):
value_type: SegmentType = SegmentType.NONE
value: None = None

4
api/core/app/task_pipeline/based_generate_task_pipeline.py
View File

@ -53,7 +53,7 @@ class BasedGenerateTaskPipeline:
self._output_moderation_handler = self._init_output_moderation()
self._stream = stream
def _handle_error(self, event: QueueErrorEvent, message: Optional[Message] = None) -> Exception:
def _handle_error(self, event: QueueErrorEvent, message: Optional[Message] = None):
"""
Handle error event.
:param event: event
@ -100,7 +100,7 @@ class BasedGenerateTaskPipeline:
return message
def _error_to_stream_response(self, e: Exception) -> ErrorStreamResponse:
def _error_to_stream_response(self, e: Exception):
"""
Error to stream response.
:param e: exception

107
api/core/app/task_pipeline/workflow_cycle_manage.py
View File

@ -1,8 +1,11 @@
import json
import time
from collections.abc import Mapping, Sequence
from datetime import datetime, timezone
from typing import Any, Optional, Union, cast
from sqlalchemy.orm import Session
from core.app.entities.app_invoke_entities import AdvancedChatAppGenerateEntity, InvokeFrom, WorkflowAppGenerateEntity
from core.app.entities.queue_entities import (
QueueIterationCompletedEvent,
@ -27,27 +30,26 @@ from core.app.entities.task_entities import (
WorkflowStartStreamResponse,
WorkflowTaskState,
)
from core.file.file_obj import FileVar
from core.file import FILE_MODEL_IDENTITY, File
from core.model_runtime.utils.encoders import jsonable_encoder
from core.ops.entities.trace_entity import TraceTaskName
from core.ops.ops_trace_manager import TraceQueueManager, TraceTask
from core.tools.tool_manager import ToolManager
from core.workflow.entities.node_entities import NodeType
from core.workflow.enums import SystemVariableKey
from core.workflow.nodes import NodeType
from core.workflow.nodes.tool.entities import ToolNodeData
from core.workflow.workflow_entry import WorkflowEntry
from extensions.ext_database import db
from models.account import Account
from models.enums import CreatedByRole, WorkflowRunTriggeredFrom
from models.model import EndUser
from models.workflow import (
CreatedByRole,
Workflow,
WorkflowNodeExecution,
WorkflowNodeExecutionStatus,
WorkflowNodeExecutionTriggeredFrom,
WorkflowRun,
WorkflowRunStatus,
WorkflowRunTriggeredFrom,
)
@ -117,7 +119,7 @@ class WorkflowCycleManage:
start_at: float,
total_tokens: int,
total_steps: int,
outputs: Optional[str] = None,
outputs: Mapping[str, Any] | None = None,
conversation_id: Optional[str] = None,
trace_manager: Optional[TraceQueueManager] = None,
) -> WorkflowRun:
@ -133,8 +135,10 @@ class WorkflowCycleManage:
"""
workflow_run = self._refetch_workflow_run(workflow_run.id)
outputs = WorkflowEntry.handle_special_values(outputs)
workflow_run.status = WorkflowRunStatus.SUCCEEDED.value
workflow_run.outputs = outputs
workflow_run.outputs = json.dumps(outputs or {})
workflow_run.elapsed_time = time.perf_counter() - start_at
workflow_run.total_tokens = total_tokens
workflow_run.total_steps = total_steps
@ -230,30 +234,30 @@ class WorkflowCycleManage:
self, workflow_run: WorkflowRun, event: QueueNodeStartedEvent
) -> WorkflowNodeExecution:
# init workflow node execution
workflow_node_execution = WorkflowNodeExecution()
workflow_node_execution.tenant_id = workflow_run.tenant_id
workflow_node_execution.app_id = workflow_run.app_id
workflow_node_execution.workflow_id = workflow_run.workflow_id
workflow_node_execution.triggered_from = WorkflowNodeExecutionTriggeredFrom.WORKFLOW_RUN.value
workflow_node_execution.workflow_run_id = workflow_run.id
workflow_node_execution.predecessor_node_id = event.predecessor_node_id
workflow_node_execution.index = event.node_run_index
workflow_node_execution.node_execution_id = event.node_execution_id
workflow_node_execution.node_id = event.node_id
workflow_node_execution.node_type = event.node_type.value
workflow_node_execution.title = event.node_data.title
workflow_node_execution.status = WorkflowNodeExecutionStatus.RUNNING.value
workflow_node_execution.created_by_role = workflow_run.created_by_role
workflow_node_execution.created_by = workflow_run.created_by
workflow_node_execution.created_at = datetime.now(timezone.utc).replace(tzinfo=None)
db.session.add(workflow_node_execution)
db.session.commit()
db.session.refresh(workflow_node_execution)
db.session.close()
with Session(db.engine, expire_on_commit=False) as session:
workflow_node_execution = WorkflowNodeExecution()
workflow_node_execution.tenant_id = workflow_run.tenant_id
workflow_node_execution.app_id = workflow_run.app_id
workflow_node_execution.workflow_id = workflow_run.workflow_id
workflow_node_execution.triggered_from = WorkflowNodeExecutionTriggeredFrom.WORKFLOW_RUN.value
workflow_node_execution.workflow_run_id = workflow_run.id
workflow_node_execution.predecessor_node_id = event.predecessor_node_id
workflow_node_execution.index = event.node_run_index
workflow_node_execution.node_execution_id = event.node_execution_id
workflow_node_execution.node_id = event.node_id
workflow_node_execution.node_type = event.node_type.value
workflow_node_execution.title = event.node_data.title
workflow_node_execution.status = WorkflowNodeExecutionStatus.RUNNING.value
workflow_node_execution.created_by_role = workflow_run.created_by_role
workflow_node_execution.created_by = workflow_run.created_by
workflow_node_execution.created_at = datetime.now(timezone.utc).replace(tzinfo=None)
session.add(workflow_node_execution)
session.commit()
session.refresh(workflow_node_execution)
self._wip_workflow_node_executions[workflow_node_execution.node_execution_id] = workflow_node_execution
return workflow_node_execution
def _handle_workflow_node_execution_success(self, event: QueueNodeSucceededEvent) -> WorkflowNodeExecution:
@ -265,6 +269,7 @@ class WorkflowCycleManage:
workflow_node_execution = self._refetch_workflow_node_execution(event.node_execution_id)
inputs = WorkflowEntry.handle_special_values(event.inputs)
process_data = WorkflowEntry.handle_special_values(event.process_data)
outputs = WorkflowEntry.handle_special_values(event.outputs)
execution_metadata = (
json.dumps(jsonable_encoder(event.execution_metadata)) if event.execution_metadata else None
@ -276,7 +281,7 @@ class WorkflowCycleManage:
{
WorkflowNodeExecution.status: WorkflowNodeExecutionStatus.SUCCEEDED.value,
WorkflowNodeExecution.inputs: json.dumps(inputs) if inputs else None,
WorkflowNodeExecution.process_data: json.dumps(event.process_data) if event.process_data else None,
WorkflowNodeExecution.process_data: json.dumps(process_data) if event.process_data else None,
WorkflowNodeExecution.outputs: json.dumps(outputs) if outputs else None,
WorkflowNodeExecution.execution_metadata: execution_metadata,
WorkflowNodeExecution.finished_at: finished_at,
@ -286,10 +291,11 @@ class WorkflowCycleManage:
db.session.commit()
db.session.close()
process_data = WorkflowEntry.handle_special_values(event.process_data)
workflow_node_execution.status = WorkflowNodeExecutionStatus.SUCCEEDED.value
workflow_node_execution.inputs = json.dumps(inputs) if inputs else None
workflow_node_execution.process_data = json.dumps(event.process_data) if event.process_data else None
workflow_node_execution.process_data = json.dumps(process_data) if process_data else None
workflow_node_execution.outputs = json.dumps(outputs) if outputs else None
workflow_node_execution.execution_metadata = execution_metadata
workflow_node_execution.finished_at = finished_at
@ -308,6 +314,7 @@ class WorkflowCycleManage:
workflow_node_execution = self._refetch_workflow_node_execution(event.node_execution_id)
inputs = WorkflowEntry.handle_special_values(event.inputs)
process_data = WorkflowEntry.handle_special_values(event.process_data)
outputs = WorkflowEntry.handle_special_values(event.outputs)
finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
elapsed_time = (finished_at - event.start_at).total_seconds()
@ -317,7 +324,7 @@ class WorkflowCycleManage:
WorkflowNodeExecution.status: WorkflowNodeExecutionStatus.FAILED.value,
WorkflowNodeExecution.error: event.error,
WorkflowNodeExecution.inputs: json.dumps(inputs) if inputs else None,
WorkflowNodeExecution.process_data: json.dumps(event.process_data) if event.process_data else None,
WorkflowNodeExecution.process_data: json.dumps(process_data) if event.process_data else None,
WorkflowNodeExecution.outputs: json.dumps(outputs) if outputs else None,
WorkflowNodeExecution.finished_at: finished_at,
WorkflowNodeExecution.elapsed_time: elapsed_time,
@ -326,11 +333,12 @@ class WorkflowCycleManage:
db.session.commit()
db.session.close()
process_data = WorkflowEntry.handle_special_values(event.process_data)
workflow_node_execution.status = WorkflowNodeExecutionStatus.FAILED.value
workflow_node_execution.error = event.error
workflow_node_execution.inputs = json.dumps(inputs) if inputs else None
workflow_node_execution.process_data = json.dumps(event.process_data) if event.process_data else None
workflow_node_execution.process_data = json.dumps(process_data) if process_data else None
workflow_node_execution.outputs = json.dumps(outputs) if outputs else None
workflow_node_execution.finished_at = finished_at
workflow_node_execution.elapsed_time = elapsed_time
@ -637,7 +645,7 @@ class WorkflowCycleManage:
),
)
def _fetch_files_from_node_outputs(self, outputs_dict: dict) -> list[dict]:
def _fetch_files_from_node_outputs(self, outputs_dict: dict) -> Sequence[Mapping[str, Any]]:
"""
Fetch files from node outputs
:param outputs_dict: node outputs dict
@ -646,15 +654,15 @@ class WorkflowCycleManage:
if not outputs_dict:
return []
files = []
for output_var, output_value in outputs_dict.items():
file_vars = self._fetch_files_from_variable_value(output_value)
if file_vars:
files.extend(file_vars)
files = [self._fetch_files_from_variable_value(output_value) for output_value in outputs_dict.values()]
# Remove None
files = [file for file in files if file]
# Flatten list
files = [file for sublist in files for file in sublist]
return files
def _fetch_files_from_variable_value(self, value: Union[dict, list]) -> list[dict]:
def _fetch_files_from_variable_value(self, value: Union[dict, list]) -> Sequence[Mapping[str, Any]]:
"""
Fetch files from variable value
:param value: variable value
@ -666,17 +674,17 @@ class WorkflowCycleManage:
files = []
if isinstance(value, list):
for item in value:
file_var = self._get_file_var_from_value(item)
if file_var:
files.append(file_var)
file = self._get_file_var_from_value(item)
if file:
files.append(file)
elif isinstance(value, dict):
file_var = self._get_file_var_from_value(value)
if file_var:
files.append(file_var)
file = self._get_file_var_from_value(value)
if file:
files.append(file)
return files
def _get_file_var_from_value(self, value: Union[dict, list]) -> Optional[dict]:
def _get_file_var_from_value(self, value: Union[dict, list]) -> Mapping[str, Any] | None:
"""
Get file var from value
:param value: variable value
@ -685,14 +693,11 @@ class WorkflowCycleManage:
if not value:
return None
if isinstance(value, dict):
if "__variant" in value and value["__variant"] == FileVar.__name__:
return value
elif isinstance(value, FileVar):
if isinstance(value, dict) and value.get("dify_model_identity") == FILE_MODEL_IDENTITY:
return value
elif isinstance(value, File):
return value.to_dict()
return None
def _refetch_workflow_run(self, workflow_run_id: str) -> WorkflowRun:
"""
Refetch workflow run

29
api/core/entities/message_entities.py
View File

@ -1,29 +0,0 @@
import enum
from typing import Any
from pydantic import BaseModel
class PromptMessageFileType(enum.Enum):
IMAGE = "image"
@staticmethod
def value_of(value):
for member in PromptMessageFileType:
if member.value == value:
return member
raise ValueError(f"No matching enum found for value '{value}'")
class PromptMessageFile(BaseModel):
type: PromptMessageFileType
data: Any = None
class ImagePromptMessageFile(PromptMessageFile):
class DETAIL(enum.Enum):
LOW = "low"
HIGH = "high"
type: PromptMessageFileType = PromptMessageFileType.IMAGE
detail: DETAIL = DETAIL.LOW

19
api/core/file/__init__.py
View File

@ -0,0 +1,19 @@
from .constants import FILE_MODEL_IDENTITY
from .enums import ArrayFileAttribute, FileAttribute, FileBelongsTo, FileTransferMethod, FileType
from .models import (
File,
FileExtraConfig,
ImageConfig,
)
__all__ = [
"FileType",
"FileExtraConfig",
"FileTransferMethod",
"FileBelongsTo",
"File",
"ImageConfig",
"FileAttribute",
"ArrayFileAttribute",
"FILE_MODEL_IDENTITY",
]

145
api/core/file/file_obj.py
View File

@ -1,145 +0,0 @@
import enum
from typing import Any, Optional
from pydantic import BaseModel
from core.file.tool_file_parser import ToolFileParser
from core.file.upload_file_parser import UploadFileParser
from core.model_runtime.entities.message_entities import ImagePromptMessageContent
from extensions.ext_database import db
class FileExtraConfig(BaseModel):
"""
File Upload Entity.
"""
image_config: Optional[dict[str, Any]] = None
class FileType(enum.Enum):
IMAGE = "image"
@staticmethod
def value_of(value):
for member in FileType:
if member.value == value:
return member
raise ValueError(f"No matching enum found for value '{value}'")
class FileTransferMethod(enum.Enum):
REMOTE_URL = "remote_url"
LOCAL_FILE = "local_file"
TOOL_FILE = "tool_file"
@staticmethod
def value_of(value):
for member in FileTransferMethod:
if member.value == value:
return member
raise ValueError(f"No matching enum found for value '{value}'")
class FileBelongsTo(enum.Enum):
USER = "user"
ASSISTANT = "assistant"
@staticmethod
def value_of(value):
for member in FileBelongsTo:
if member.value == value:
return member
raise ValueError(f"No matching enum found for value '{value}'")
class FileVar(BaseModel):
id: Optional[str] = None # message file id
tenant_id: str
type: FileType
transfer_method: FileTransferMethod
url: Optional[str] = None # remote url
related_id: Optional[str] = None
extra_config: Optional[FileExtraConfig] = None
filename: Optional[str] = None
extension: Optional[str] = None
mime_type: Optional[str] = None
def to_dict(self) -> dict:
return {
"__variant": self.__class__.__name__,
"tenant_id": self.tenant_id,
"type": self.type.value,
"transfer_method": self.transfer_method.value,
"url": self.preview_url,
"remote_url": self.url,
"related_id": self.related_id,
"filename": self.filename,
"extension": self.extension,
"mime_type": self.mime_type,
}
def to_markdown(self) -> str:
"""
Convert file to markdown
:return:
"""
preview_url = self.preview_url
if self.type == FileType.IMAGE:
text = f'![{self.filename or ""}]({preview_url})'
else:
text = f"[{self.filename or preview_url}]({preview_url})"
return text
@property
def data(self) -> Optional[str]:
"""
Get image data, file signed url or base64 data
depending on config MULTIMODAL_SEND_IMAGE_FORMAT
:return:
"""
return self._get_data()
@property
def preview_url(self) -> Optional[str]:
"""
Get signed preview url
:return:
"""
return self._get_data(force_url=True)
@property
def prompt_message_content(self) -> ImagePromptMessageContent:
if self.type == FileType.IMAGE:
image_config = self.extra_config.image_config
return ImagePromptMessageContent(
data=self.data,
detail=ImagePromptMessageContent.DETAIL.HIGH
if image_config.get("detail") == "high"
else ImagePromptMessageContent.DETAIL.LOW,
)
def _get_data(self, force_url: bool = False) -> Optional[str]:
from models.model import UploadFile
if self.type == FileType.IMAGE:
if self.transfer_method == FileTransferMethod.REMOTE_URL:
return self.url
elif self.transfer_method == FileTransferMethod.LOCAL_FILE:
upload_file = (
db.session.query(UploadFile)
.filter(UploadFile.id == self.related_id, UploadFile.tenant_id == self.tenant_id)
.first()
)
return UploadFileParser.get_image_data(upload_file=upload_file, force_url=force_url)
elif self.transfer_method == FileTransferMethod.TOOL_FILE:
extension = self.extension
# add sign url
return ToolFileParser.get_tool_file_manager().sign_file(
tool_file_id=self.related_id, extension=extension
)
return None

243
api/core/file/message_file_parser.py
View File

@ -1,243 +0,0 @@
import re
from collections.abc import Mapping, Sequence
from typing import Any, Union
from urllib.parse import parse_qs, urlparse
import requests
from core.file.file_obj import FileBelongsTo, FileExtraConfig, FileTransferMethod, FileType, FileVar
from extensions.ext_database import db
from models.account import Account
from models.model import EndUser, MessageFile, UploadFile
from services.file_service import IMAGE_EXTENSIONS
class MessageFileParser:
def __init__(self, tenant_id: str, app_id: str) -> None:
self.tenant_id = tenant_id
self.app_id = app_id
def validate_and_transform_files_arg(
self, files: Sequence[Mapping[str, Any]], file_extra_config: FileExtraConfig, user: Union[Account, EndUser]
) -> list[FileVar]:
"""
validate and transform files arg
:param files:
:param file_extra_config:
:param user:
:return:
"""
for file in files:
if not isinstance(file, dict):
raise ValueError("Invalid file format, must be dict")
if not file.get("type"):
raise ValueError("Missing file type")
FileType.value_of(file.get("type"))
if not file.get("transfer_method"):
raise ValueError("Missing file transfer method")
FileTransferMethod.value_of(file.get("transfer_method"))
if file.get("transfer_method") == FileTransferMethod.REMOTE_URL.value:
if not file.get("url"):
raise ValueError("Missing file url")
if not file.get("url").startswith("http"):
raise ValueError("Invalid file url")
if file.get("transfer_method") == FileTransferMethod.LOCAL_FILE.value and not file.get("upload_file_id"):
raise ValueError("Missing file upload_file_id")
if file.get("transform_method") == FileTransferMethod.TOOL_FILE.value and not file.get("tool_file_id"):
raise ValueError("Missing file tool_file_id")
# transform files to file objs
type_file_objs = self._to_file_objs(files, file_extra_config)
# validate files
new_files = []
for file_type, file_objs in type_file_objs.items():
if file_type == FileType.IMAGE:
# parse and validate files
image_config = file_extra_config.image_config
# check if image file feature is enabled
if not image_config:
continue
# Validate number of files
if len(files) > image_config["number_limits"]:
raise ValueError(f"Number of image files exceeds the maximum limit {image_config['number_limits']}")
for file_obj in file_objs:
# Validate transfer method
if file_obj.transfer_method.value not in image_config["transfer_methods"]:
raise ValueError(f"Invalid transfer method: {file_obj.transfer_method.value}")
# Validate file type
if file_obj.type != FileType.IMAGE:
raise ValueError(f"Invalid file type: {file_obj.type}")
if file_obj.transfer_method == FileTransferMethod.REMOTE_URL:
# check remote url valid and is image
result, error = self._check_image_remote_url(file_obj.url)
if result is False:
raise ValueError(error)
elif file_obj.transfer_method == FileTransferMethod.LOCAL_FILE:
# get upload file from upload_file_id
upload_file = (
db.session.query(UploadFile)
.filter(
UploadFile.id == file_obj.related_id,
UploadFile.tenant_id == self.tenant_id,
UploadFile.created_by == user.id,
UploadFile.created_by_role == ("account" if isinstance(user, Account) else "end_user"),
UploadFile.extension.in_(IMAGE_EXTENSIONS),
)
.first()
)
# check upload file is belong to tenant and user
if not upload_file:
raise ValueError("Invalid upload file")
new_files.append(file_obj)
# return all file objs
return new_files
def transform_message_files(self, files: list[MessageFile], file_extra_config: FileExtraConfig):
"""
transform message files
:param files:
:param file_extra_config:
:return:
"""
# transform files to file objs
type_file_objs = self._to_file_objs(files, file_extra_config)
# return all file objs
return [file_obj for file_objs in type_file_objs.values() for file_obj in file_objs]
def _to_file_objs(
self, files: list[Union[dict, MessageFile]], file_extra_config: FileExtraConfig
) -> dict[FileType, list[FileVar]]:
"""
transform files to file objs
:param files:
:param file_extra_config:
:return:
"""
type_file_objs: dict[FileType, list[FileVar]] = {
# Currently only support image
FileType.IMAGE: []
}
if not files:
return type_file_objs
# group by file type and convert file args or message files to FileObj
for file in files:
if isinstance(file, MessageFile):
if file.belongs_to == FileBelongsTo.ASSISTANT.value:
continue
file_obj = self._to_file_obj(file, file_extra_config)
if file_obj.type not in type_file_objs:
continue
type_file_objs[file_obj.type].append(file_obj)
return type_file_objs
def _to_file_obj(self, file: Union[dict, MessageFile], file_extra_config: FileExtraConfig):
"""
transform file to file obj
:param file:
:return:
"""
if isinstance(file, dict):
transfer_method = FileTransferMethod.value_of(file.get("transfer_method"))
if transfer_method != FileTransferMethod.TOOL_FILE:
return FileVar(
tenant_id=self.tenant_id,
type=FileType.value_of(file.get("type")),
transfer_method=transfer_method,
url=file.get("url") if transfer_method == FileTransferMethod.REMOTE_URL else None,
related_id=file.get("upload_file_id") if transfer_method == FileTransferMethod.LOCAL_FILE else None,
extra_config=file_extra_config,
)
return FileVar(
tenant_id=self.tenant_id,
type=FileType.value_of(file.get("type")),
transfer_method=transfer_method,
url=None,
related_id=file.get("tool_file_id"),
extra_config=file_extra_config,
)
else:
return FileVar(
id=file.id,
tenant_id=self.tenant_id,
type=FileType.value_of(file.type),
transfer_method=FileTransferMethod.value_of(file.transfer_method),
url=file.url,
related_id=file.upload_file_id or None,
extra_config=file_extra_config,
)
def _check_image_remote_url(self, url):
try:
headers = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko)"
" Chrome/91.0.4472.124 Safari/537.36"
}
def is_s3_presigned_url(url):
try:
parsed_url = urlparse(url)
if "amazonaws.com" not in parsed_url.netloc:
return False
query_params = parse_qs(parsed_url.query)
def check_presign_v2(query_params):
required_params = ["Signature", "Expires"]
for param in required_params:
if param not in query_params:
return False
if not query_params["Expires"][0].isdigit():
return False
signature = query_params["Signature"][0]
if not re.match(r"^[A-Za-z0-9+/]+={0,2}$", signature):
return False
return True
def check_presign_v4(query_params):
required_params = ["X-Amz-Signature", "X-Amz-Expires"]
for param in required_params:
if param not in query_params:
return False
if not query_params["X-Amz-Expires"][0].isdigit():
return False
signature = query_params["X-Amz-Signature"][0]
if not re.match(r"^[A-Za-z0-9+/]+={0,2}$", signature):
return False
return True
return check_presign_v4(query_params) or check_presign_v2(query_params)
except Exception:
return False
if is_s3_presigned_url(url):
response = requests.get(url, headers=headers, allow_redirects=True)
if response.status_code in {200, 304}:
return True, ""
response = requests.head(url, headers=headers, allow_redirects=True)
if response.status_code in {200, 304}:
return True, ""
else:
return False, "URL does not exist."
except requests.RequestException as e:
return False, f"Error checking URL: {e}"

7
api/core/file/tool_file_parser.py
View File

@ -1,4 +1,9 @@
tool_file_manager = {"manager": None}
from typing import TYPE_CHECKING, Any
if TYPE_CHECKING:
from core.tools.tool_file_manager import ToolFileManager
tool_file_manager: dict[str, Any] = {"manager": None}
class ToolFileParser:

79
api/core/file/upload_file_parser.py
View File

@ -1,79 +0,0 @@
import base64
import hashlib
import hmac
import logging
import os
import time
from typing import Optional
from configs import dify_config
from extensions.ext_storage import storage
IMAGE_EXTENSIONS = ["jpg", "jpeg", "png", "webp", "gif", "svg"]
IMAGE_EXTENSIONS.extend([ext.upper() for ext in IMAGE_EXTENSIONS])
class UploadFileParser:
@classmethod
def get_image_data(cls, upload_file, force_url: bool = False) -> Optional[str]:
if not upload_file:
return None
if upload_file.extension not in IMAGE_EXTENSIONS:
return None
if dify_config.MULTIMODAL_SEND_IMAGE_FORMAT == "url" or force_url:
return cls.get_signed_temp_image_url(upload_file.id)
else:
# get image file base64
try:
data = storage.load(upload_file.key)
except FileNotFoundError:
logging.error(f"File not found: {upload_file.key}")
return None
encoded_string = base64.b64encode(data).decode("utf-8")
return f"data:{upload_file.mime_type};base64,{encoded_string}"
@classmethod
def get_signed_temp_image_url(cls, upload_file_id) -> str:
"""
get signed url from upload file
:param upload_file: UploadFile object
:return:
"""
base_url = dify_config.FILES_URL
image_preview_url = f"{base_url}/files/{upload_file_id}/image-preview"
timestamp = str(int(time.time()))
nonce = os.urandom(16).hex()
data_to_sign = f"image-preview|{upload_file_id}|{timestamp}|{nonce}"
secret_key = dify_config.SECRET_KEY.encode()
sign = hmac.new(secret_key, data_to_sign.encode(), hashlib.sha256).digest()
encoded_sign = base64.urlsafe_b64encode(sign).decode()
return f"{image_preview_url}?timestamp={timestamp}&nonce={nonce}&sign={encoded_sign}"
@classmethod
def verify_image_file_signature(cls, upload_file_id: str, timestamp: str, nonce: str, sign: str) -> bool:
"""
verify signature
:param upload_file_id: file id
:param timestamp: timestamp
:param nonce: nonce
:param sign: signature
:return:
"""
data_to_sign = f"image-preview|{upload_file_id}|{timestamp}|{nonce}"
secret_key = dify_config.SECRET_KEY.encode()
recalculated_sign = hmac.new(secret_key, data_to_sign.encode(), hashlib.sha256).digest()
recalculated_encoded_sign = base64.urlsafe_b64encode(recalculated_sign).decode()
# verify signature
if sign != recalculated_encoded_sign:
return False
current_time = int(time.time())
return current_time - int(timestamp) <= dify_config.FILES_ACCESS_TIMEOUT

18
api/core/helper/ssrf_proxy.py
View File

@ -13,8 +13,11 @@ SSRF_PROXY_HTTP_URL = os.getenv("SSRF_PROXY_HTTP_URL", "")
SSRF_PROXY_HTTPS_URL = os.getenv("SSRF_PROXY_HTTPS_URL", "")
SSRF_DEFAULT_MAX_RETRIES = int(os.getenv("SSRF_DEFAULT_MAX_RETRIES", "3"))
proxies = (
{"http://": SSRF_PROXY_HTTP_URL, "https://": SSRF_PROXY_HTTPS_URL}
proxy_mounts = (
{
"http://": httpx.HTTPTransport(proxy=SSRF_PROXY_HTTP_URL),
"https://": httpx.HTTPTransport(proxy=SSRF_PROXY_HTTPS_URL),
}
if SSRF_PROXY_HTTP_URL and SSRF_PROXY_HTTPS_URL
else None
)
@ -33,11 +36,14 @@ def make_request(method, url, max_retries=SSRF_DEFAULT_MAX_RETRIES, **kwargs):
while retries <= max_retries:
try:
if SSRF_PROXY_ALL_URL:
response = httpx.request(method=method, url=url, proxy=SSRF_PROXY_ALL_URL, **kwargs)
elif proxies:
response = httpx.request(method=method, url=url, proxies=proxies, **kwargs)
with httpx.Client(proxy=SSRF_PROXY_ALL_URL) as client:
response = client.request(method=method, url=url, **kwargs)
elif proxy_mounts:
with httpx.Client(mounts=proxy_mounts) as client:
response = client.request(method=method, url=url, **kwargs)
else:
response = httpx.request(method=method, url=url, **kwargs)
with httpx.Client() as client:
response = client.request(method=method, url=url, **kwargs)
if response.status_code not in STATUS_FORCELIST:
return response

91
api/core/hosting_configuration.py
View File

@ -1,8 +1,9 @@
from typing import Optional
from flask import Config, Flask
from flask import Flask
from pydantic import BaseModel
from configs import dify_config
from core.entities.provider_entities import QuotaUnit, RestrictModel
from core.model_runtime.entities.model_entities import ModelType
from models.provider import ProviderQuotaType
@ -44,32 +45,30 @@ class HostingConfiguration:
moderation_config: HostedModerationConfig = None
def init_app(self, app: Flask) -> None:
config = app.config
if config.get("EDITION") != "CLOUD":
if dify_config.EDITION != "CLOUD":
return
self.provider_map["azure_openai"] = self.init_azure_openai(config)
self.provider_map["openai"] = self.init_openai(config)
self.provider_map["anthropic"] = self.init_anthropic(config)
self.provider_map["minimax"] = self.init_minimax(config)
self.provider_map["spark"] = self.init_spark(config)
self.provider_map["zhipuai"] = self.init_zhipuai(config)
self.provider_map["azure_openai"] = self.init_azure_openai()
self.provider_map["openai"] = self.init_openai()
self.provider_map["anthropic"] = self.init_anthropic()
self.provider_map["minimax"] = self.init_minimax()
self.provider_map["spark"] = self.init_spark()
self.provider_map["zhipuai"] = self.init_zhipuai()
self.moderation_config = self.init_moderation_config(config)
self.moderation_config = self.init_moderation_config()
@staticmethod
def init_azure_openai(app_config: Config) -> HostingProvider:
def init_azure_openai() -> HostingProvider:
quota_unit = QuotaUnit.TIMES
if app_config.get("HOSTED_AZURE_OPENAI_ENABLED"):
if dify_config.HOSTED_AZURE_OPENAI_ENABLED:
credentials = {
"openai_api_key": app_config.get("HOSTED_AZURE_OPENAI_API_KEY"),
"openai_api_base": app_config.get("HOSTED_AZURE_OPENAI_API_BASE"),
"openai_api_key": dify_config.HOSTED_AZURE_OPENAI_API_KEY,
"openai_api_base": dify_config.HOSTED_AZURE_OPENAI_API_BASE,
"base_model_name": "gpt-35-turbo",
}
quotas = []
hosted_quota_limit = int(app_config.get("HOSTED_AZURE_OPENAI_QUOTA_LIMIT", "1000"))
hosted_quota_limit = dify_config.HOSTED_AZURE_OPENAI_QUOTA_LIMIT
trial_quota = TrialHostingQuota(
quota_limit=hosted_quota_limit,
restrict_models=[
@ -122,31 +121,31 @@ class HostingConfiguration:
quota_unit=quota_unit,
)
def init_openai(self, app_config: Config) -> HostingProvider:
def init_openai(self) -> HostingProvider:
quota_unit = QuotaUnit.CREDITS
quotas = []
if app_config.get("HOSTED_OPENAI_TRIAL_ENABLED"):
hosted_quota_limit = int(app_config.get("HOSTED_OPENAI_QUOTA_LIMIT", "200"))
trial_models = self.parse_restrict_models_from_env(app_config, "HOSTED_OPENAI_TRIAL_MODELS")
if dify_config.HOSTED_OPENAI_TRIAL_ENABLED:
hosted_quota_limit = dify_config.HOSTED_OPENAI_QUOTA_LIMIT
trial_models = self.parse_restrict_models_from_env("HOSTED_OPENAI_TRIAL_MODELS")
trial_quota = TrialHostingQuota(quota_limit=hosted_quota_limit, restrict_models=trial_models)
quotas.append(trial_quota)
if app_config.get("HOSTED_OPENAI_PAID_ENABLED"):
paid_models = self.parse_restrict_models_from_env(app_config, "HOSTED_OPENAI_PAID_MODELS")
if dify_config.HOSTED_OPENAI_PAID_ENABLED:
paid_models = self.parse_restrict_models_from_env("HOSTED_OPENAI_PAID_MODELS")
paid_quota = PaidHostingQuota(restrict_models=paid_models)
quotas.append(paid_quota)
if len(quotas) > 0:
credentials = {
"openai_api_key": app_config.get("HOSTED_OPENAI_API_KEY"),
"openai_api_key": dify_config.HOSTED_OPENAI_API_KEY,
}
if app_config.get("HOSTED_OPENAI_API_BASE"):
credentials["openai_api_base"] = app_config.get("HOSTED_OPENAI_API_BASE")
if dify_config.HOSTED_OPENAI_API_BASE:
credentials["openai_api_base"] = dify_config.HOSTED_OPENAI_API_BASE
if app_config.get("HOSTED_OPENAI_API_ORGANIZATION"):
credentials["openai_organization"] = app_config.get("HOSTED_OPENAI_API_ORGANIZATION")
if dify_config.HOSTED_OPENAI_API_ORGANIZATION:
credentials["openai_organization"] = dify_config.HOSTED_OPENAI_API_ORGANIZATION
return HostingProvider(enabled=True, credentials=credentials, quota_unit=quota_unit, quotas=quotas)
@ -156,26 +155,26 @@ class HostingConfiguration:
)
@staticmethod
def init_anthropic(app_config: Config) -> HostingProvider:
def init_anthropic() -> HostingProvider:
quota_unit = QuotaUnit.TOKENS
quotas = []
if app_config.get("HOSTED_ANTHROPIC_TRIAL_ENABLED"):
hosted_quota_limit = int(app_config.get("HOSTED_ANTHROPIC_QUOTA_LIMIT", "0"))
if dify_config.HOSTED_ANTHROPIC_TRIAL_ENABLED:
hosted_quota_limit = dify_config.HOSTED_ANTHROPIC_QUOTA_LIMIT
trial_quota = TrialHostingQuota(quota_limit=hosted_quota_limit)
quotas.append(trial_quota)
if app_config.get("HOSTED_ANTHROPIC_PAID_ENABLED"):
if dify_config.HOSTED_ANTHROPIC_PAID_ENABLED:
paid_quota = PaidHostingQuota()
quotas.append(paid_quota)
if len(quotas) > 0:
credentials = {
"anthropic_api_key": app_config.get("HOSTED_ANTHROPIC_API_KEY"),
"anthropic_api_key": dify_config.HOSTED_ANTHROPIC_API_KEY,
}
if app_config.get("HOSTED_ANTHROPIC_API_BASE"):
credentials["anthropic_api_url"] = app_config.get("HOSTED_ANTHROPIC_API_BASE")
if dify_config.HOSTED_ANTHROPIC_API_BASE:
credentials["anthropic_api_url"] = dify_config.HOSTED_ANTHROPIC_API_BASE
return HostingProvider(enabled=True, credentials=credentials, quota_unit=quota_unit, quotas=quotas)
@ -185,9 +184,9 @@ class HostingConfiguration:
)
@staticmethod
def init_minimax(app_config: Config) -> HostingProvider:
def init_minimax() -> HostingProvider:
quota_unit = QuotaUnit.TOKENS
if app_config.get("HOSTED_MINIMAX_ENABLED"):
if dify_config.HOSTED_MINIMAX_ENABLED:
quotas = [FreeHostingQuota()]
return HostingProvider(
@ -203,9 +202,9 @@ class HostingConfiguration:
)
@staticmethod
def init_spark(app_config: Config) -> HostingProvider:
def init_spark() -> HostingProvider:
quota_unit = QuotaUnit.TOKENS
if app_config.get("HOSTED_SPARK_ENABLED"):
if dify_config.HOSTED_SPARK_ENABLED:
quotas = [FreeHostingQuota()]
return HostingProvider(
@ -221,9 +220,9 @@ class HostingConfiguration:
)
@staticmethod
def init_zhipuai(app_config: Config) -> HostingProvider:
def init_zhipuai() -> HostingProvider:
quota_unit = QuotaUnit.TOKENS
if app_config.get("HOSTED_ZHIPUAI_ENABLED"):
if dify_config.HOSTED_ZHIPUAI_ENABLED:
quotas = [FreeHostingQuota()]
return HostingProvider(
@ -239,17 +238,15 @@ class HostingConfiguration:
)
@staticmethod
def init_moderation_config(app_config: Config) -> HostedModerationConfig:
if app_config.get("HOSTED_MODERATION_ENABLED") and app_config.get("HOSTED_MODERATION_PROVIDERS"):
return HostedModerationConfig(
enabled=True, providers=app_config.get("HOSTED_MODERATION_PROVIDERS").split(",")
)
def init_moderation_config() -> HostedModerationConfig:
if dify_config.HOSTED_MODERATION_ENABLED and dify_config.HOSTED_MODERATION_PROVIDERS:
return HostedModerationConfig(enabled=True, providers=dify_config.HOSTED_MODERATION_PROVIDERS.split(","))
return HostedModerationConfig(enabled=False)
@staticmethod
def parse_restrict_models_from_env(app_config: Config, env_var: str) -> list[RestrictModel]:
models_str = app_config.get(env_var)
def parse_restrict_models_from_env(env_var: str) -> list[RestrictModel]:
models_str = dify_config.model_dump().get(env_var)
models_list = models_str.split(",") if models_str else []
return [
RestrictModel(model=model_name.strip(), model_type=ModelType.LLM)

50
api/core/llm_generator/llm_generator.py
View File

@ -8,6 +8,8 @@ from core.llm_generator.output_parser.suggested_questions_after_answer import Su
from core.llm_generator.prompts import (
CONVERSATION_TITLE_PROMPT,
GENERATOR_QA_PROMPT,
JAVASCRIPT_CODE_GENERATOR_PROMPT_TEMPLATE,
PYTHON_CODE_GENERATOR_PROMPT_TEMPLATE,
WORKFLOW_RULE_CONFIG_PROMPT_GENERATE_TEMPLATE,
)
from core.model_manager import ModelManager
@ -239,6 +241,54 @@ class LLMGenerator:
return rule_config
@classmethod
def generate_code(
cls,
tenant_id: str,
instruction: str,
model_config: dict,
code_language: str = "javascript",
max_tokens: int = 1000,
) -> dict:
if code_language == "python":
prompt_template = PromptTemplateParser(PYTHON_CODE_GENERATOR_PROMPT_TEMPLATE)
else:
prompt_template = PromptTemplateParser(JAVASCRIPT_CODE_GENERATOR_PROMPT_TEMPLATE)
prompt = prompt_template.format(
inputs={
"INSTRUCTION": instruction,
"CODE_LANGUAGE": code_language,
},
remove_template_variables=False,
)
model_manager = ModelManager()
model_instance = model_manager.get_model_instance(
tenant_id=tenant_id,
model_type=ModelType.LLM,
provider=model_config.get("provider") if model_config else None,
model=model_config.get("name") if model_config else None,
)
prompt_messages = [UserPromptMessage(content=prompt)]
model_parameters = {"max_tokens": max_tokens, "temperature": 0.01}
try:
response = model_instance.invoke_llm(
prompt_messages=prompt_messages, model_parameters=model_parameters, stream=False
)
generated_code = response.message.content
return {"code": generated_code, "language": code_language, "error": ""}
except InvokeError as e:
error = str(e)
return {"code": "", "language": code_language, "error": f"Failed to generate code. Error: {error}"}
except Exception as e:
logging.exception(e)
return {"code": "", "language": code_language, "error": f"An unexpected error occurred: {str(e)}"}
@classmethod
def generate_qa_document(cls, tenant_id: str, query, document_language: str):
prompt = GENERATOR_QA_PROMPT.format(language=document_language)

67
api/core/llm_generator/prompts.py
View File

@ -61,6 +61,73 @@ User Input: yo, 你今天咋样？
User Input:
""" # noqa: E501
PYTHON_CODE_GENERATOR_PROMPT_TEMPLATE = (
"You are an expert programmer. Generate code based on the following instructions:\n\n"
"Instructions: {{INSTRUCTION}}\n\n"
"Write the code in {{CODE_LANGUAGE}}.\n\n"
"Please ensure that you meet the following requirements:\n"
"1. Define a function named 'main'.\n"
"2. The 'main' function must return a dictionary (dict).\n"
"3. You may modify the arguments of the 'main' function, but include appropriate type hints.\n"
"4. The returned dictionary should contain at least one key-value pair.\n\n"
"5. You may ONLY use the following libraries in your code: \n"
"- json\n"
"- datetime\n"
"- math\n"
"- random\n"
"- re\n"
"- string\n"
"- sys\n"
"- time\n"
"- traceback\n"
"- uuid\n"
"- os\n"
"- base64\n"
"- hashlib\n"
"- hmac\n"
"- binascii\n"
"- collections\n"
"- functools\n"
"- operator\n"
"- itertools\n\n"
"Example:\n"
"def main(arg1: str, arg2: int) -> dict:\n"
" return {\n"
' "result": arg1 * arg2,\n'
" }\n\n"
"IMPORTANT:\n"
"- Provide ONLY the code without any additional explanations, comments, or markdown formatting.\n"
"- DO NOT use markdown code blocks (``` or ``` python). Return the raw code directly.\n"
"- The code should start immediately after this instruction, without any preceding newlines or spaces.\n"
"- The code should be complete, functional, and follow best practices for {{CODE_LANGUAGE}}.\n\n"
"- Always use the format return {'result': ...} for the output.\n\n"
"Generated Code:\n"
)
JAVASCRIPT_CODE_GENERATOR_PROMPT_TEMPLATE = (
"You are an expert programmer. Generate code based on the following instructions:\n\n"
"Instructions: {{INSTRUCTION}}\n\n"
"Write the code in {{CODE_LANGUAGE}}.\n\n"
"Please ensure that you meet the following requirements:\n"
"1. Define a function named 'main'.\n"
"2. The 'main' function must return an object.\n"
"3. You may modify the arguments of the 'main' function, but include appropriate JSDoc annotations.\n"
"4. The returned object should contain at least one key-value pair.\n\n"
"5. The returned object should always be in the format: {result: ...}\n\n"
"Example:\n"
"function main(arg1, arg2) {\n"
" return {\n"
" result: arg1 * arg2\n"
" };\n"
"}\n\n"
"IMPORTANT:\n"
"- Provide ONLY the code without any additional explanations, comments, or markdown formatting.\n"
"- DO NOT use markdown code blocks (``` or ``` javascript). Return the raw code directly.\n"
"- The code should start immediately after this instruction, without any preceding newlines or spaces.\n"
"- The code should be complete, functional, and follow best practices for {{CODE_LANGUAGE}}.\n\n"
"Generated Code:\n"
)
SUGGESTED_QUESTIONS_AFTER_ANSWER_INSTRUCTION_PROMPT = (
"Please help me predict the three most likely questions that human would ask, "
"and keeping each question under 20 characters.\n"

23
api/core/memory/token_buffer_memory.py
View File

@ -1,18 +1,21 @@
from typing import Optional
from core.app.app_config.features.file_upload.manager import FileUploadConfigManager
from core.file.message_file_parser import MessageFileParser
from core.file import file_manager
from core.file.models import FileType
from core.model_manager import ModelInstance
from core.model_runtime.entities.message_entities import (
from core.model_runtime.entities import (
AssistantPromptMessage,
ImagePromptMessageContent,
PromptMessage,
PromptMessageContent,
PromptMessageRole,
TextPromptMessageContent,
UserPromptMessage,
)
from core.prompt.utils.extract_thread_messages import extract_thread_messages
from extensions.ext_database import db
from factories import file_factory
from models.model import AppMode, Conversation, Message, MessageFile
from models.workflow import WorkflowRun
@ -65,13 +68,12 @@ class TokenBufferMemory:
messages = list(reversed(thread_messages))
message_file_parser = MessageFileParser(tenant_id=app_record.tenant_id, app_id=app_record.id)
prompt_messages = []
for message in messages:
files = db.session.query(MessageFile).filter(MessageFile.message_id == message.id).all()
if files:
file_extra_config = None
if self.conversation.mode not in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}:
if self.conversation.mode not in {AppMode.ADVANCED_CHAT, AppMode.WORKFLOW}:
file_extra_config = FileUploadConfigManager.convert(self.conversation.model_config)
else:
if message.workflow_run_id:
@ -84,17 +86,22 @@ class TokenBufferMemory:
workflow_run.workflow.features_dict, is_vision=False
)
if file_extra_config:
file_objs = message_file_parser.transform_message_files(files, file_extra_config)
if file_extra_config and app_record:
file_objs = file_factory.build_from_message_files(
message_files=files, tenant_id=app_record.tenant_id, config=file_extra_config
)
else:
file_objs = []
if not file_objs:
prompt_messages.append(UserPromptMessage(content=message.query))
else:
prompt_message_contents = [TextPromptMessageContent(data=message.query)]
prompt_message_contents: list[PromptMessageContent] = []
prompt_message_contents.append(TextPromptMessageContent(data=message.query))
for file_obj in file_objs:
prompt_message_contents.append(file_obj.prompt_message_content)
if file_obj.type in {FileType.IMAGE, FileType.AUDIO}:
prompt_message = file_manager.to_prompt_message_content(file_obj)
prompt_message_contents.append(prompt_message)
prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
else:

8
api/core/model_manager.py
View File

@ -1,7 +1,7 @@
import logging
import os
from collections.abc import Callable, Generator, Sequence
from typing import IO, Optional, Union, cast
from collections.abc import Callable, Generator, Iterable, Sequence
from typing import IO, Any, Optional, Union, cast
from core.entities.embedding_type import EmbeddingInputType
from core.entities.provider_configuration import ProviderConfiguration, ProviderModelBundle
@ -274,7 +274,7 @@ class ModelInstance:
user=user,
)
def invoke_tts(self, content_text: str, tenant_id: str, voice: str, user: Optional[str] = None) -> str:
def invoke_tts(self, content_text: str, tenant_id: str, voice: str, user: Optional[str] = None) -> Iterable[bytes]:
"""
Invoke large language tts model
@ -298,7 +298,7 @@ class ModelInstance:
voice=voice,
)
def _round_robin_invoke(self, function: Callable, *args, **kwargs):
def _round_robin_invoke(self, function: Callable[..., Any], *args, **kwargs):
"""
Round-robin invoke
:param function: function to invoke

2
api/core/model_runtime/docs/en_US/customizable_model_scale_out.md
View File

@ -218,7 +218,7 @@ For instance, Xinference supports `max_tokens`, `temperature`, and `top_p` param
However, some vendors may support different parameters for different models. For example, the `OpenLLM` vendor supports `top_k`, but not all models provided by this vendor support `top_k`. Let's say model A supports `top_k` but model B does not. In such cases, we need to dynamically generate the model parameter schema, as illustrated below:
```python
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/docs/zh_Hans/customizable_model_scale_out.md
View File

@ -205,7 +205,7 @@ provider_credential_schema:
但是有的供应商根据不同的模型支持不同的参数，如供应商`OpenLLM`支持`top_k`，但是并不是这个供应商提供的所有模型都支持`top_k`，我们这里举例A模型支持`top_k`，B模型不支持`top_k`，那么我们需要在这里动态生成模型参数的Schema，如下所示：
```python
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

38
api/core/model_runtime/entities/__init__.py
View File

@ -0,0 +1,38 @@
from .llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta, LLMUsage
from .message_entities import (
AssistantPromptMessage,
AudioPromptMessageContent,
ImagePromptMessageContent,
PromptMessage,
PromptMessageContent,
PromptMessageContentType,
PromptMessageRole,
PromptMessageTool,
SystemPromptMessage,
TextPromptMessageContent,
ToolPromptMessage,
UserPromptMessage,
)
from .model_entities import ModelPropertyKey
__all__ = [
"ImagePromptMessageContent",
"PromptMessage",
"PromptMessageRole",
"LLMUsage",
"ModelPropertyKey",
"AssistantPromptMessage",
"PromptMessage",
"PromptMessageContent",
"PromptMessageRole",
"SystemPromptMessage",
"TextPromptMessageContent",
"UserPromptMessage",
"PromptMessageTool",
"ToolPromptMessage",
"PromptMessageContentType",
"LLMResult",
"LLMResultChunk",
"LLMResultChunkDelta",
"AudioPromptMessageContent",
]

1
api/core/model_runtime/entities/llm_entities.py
View File

@ -105,6 +105,7 @@ class LLMResult(BaseModel):
Model class for llm result.
"""
id: Optional[str] = None
model: str
prompt_messages: list[PromptMessage]
message: AssistantPromptMessage

11
api/core/model_runtime/entities/message_entities.py
View File

@ -2,7 +2,7 @@ from abc import ABC
from enum import Enum
from typing import Optional
from pydantic import BaseModel, field_validator
from pydantic import BaseModel, Field, field_validator
class PromptMessageRole(Enum):
@ -55,6 +55,7 @@ class PromptMessageContentType(Enum):
TEXT = "text"
IMAGE = "image"
AUDIO = "audio"
class PromptMessageContent(BaseModel):
@ -74,12 +75,18 @@ class TextPromptMessageContent(PromptMessageContent):
type: PromptMessageContentType = PromptMessageContentType.TEXT
class AudioPromptMessageContent(PromptMessageContent):
type: PromptMessageContentType = PromptMessageContentType.AUDIO
data: str = Field(..., description="Base64 encoded audio data")
format: str = Field(..., description="Audio format")
class ImagePromptMessageContent(PromptMessageContent):
"""
Model class for image prompt message content.
"""
class DETAIL(Enum):
class DETAIL(str, Enum):
LOW = "low"
HIGH = "high"

15
api/core/model_runtime/model_providers/__base/large_language_model.py
View File

@ -1,5 +1,4 @@
import logging
import os
import re
import time
from abc import abstractmethod
@ -8,6 +7,7 @@ from typing import Optional, Union
from pydantic import ConfigDict
from configs import dify_config
from core.model_runtime.callbacks.base_callback import Callback
from core.model_runtime.callbacks.logging_callback import LoggingCallback
from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, LLMResultChunk, LLMResultChunkDelta, LLMUsage
@ -77,7 +77,7 @@ class LargeLanguageModel(AIModel):
callbacks = callbacks or []
if bool(os.environ.get("DEBUG", "False").lower() == "true"):
if dify_config.DEBUG:
callbacks.append(LoggingCallback())
# trigger before invoke callbacks
@ -107,7 +107,16 @@ class LargeLanguageModel(AIModel):
callbacks=callbacks,
)
else:
result = self._invoke(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
result = self._invoke(
model=model,
credentials=credentials,
prompt_messages=prompt_messages,
model_parameters=model_parameters,
tools=tools,
stop=stop,
stream=stream,
user=user,
)
except Exception as e:
self._trigger_invoke_error_callbacks(
model=model,

69
api/core/model_runtime/model_providers/__base/tts_model.py
View File

@ -1,6 +1,7 @@
import logging
import re
from abc import abstractmethod
from collections.abc import Iterable
from typing import Any, Optional
from pydantic import ConfigDict
@ -22,8 +23,14 @@ class TTSModel(AIModel):
model_config = ConfigDict(protected_namespaces=())
def invoke(
self, model: str, tenant_id: str, credentials: dict, content_text: str, voice: str, user: Optional[str] = None
):
self,
model: str,
tenant_id: str,
credentials: dict,
content_text: str,
voice: str,
user: Optional[str] = None,
) -> Iterable[bytes]:
"""
Invoke large language model
@ -50,8 +57,14 @@ class TTSModel(AIModel):
@abstractmethod
def _invoke(
self, model: str, tenant_id: str, credentials: dict, content_text: str, voice: str, user: Optional[str] = None
):
self,
model: str,
tenant_id: str,
credentials: dict,
content_text: str,
voice: str,
user: Optional[str] = None,
) -> Iterable[bytes]:
"""
Invoke large language model
@ -68,25 +81,25 @@ class TTSModel(AIModel):
def get_tts_model_voices(self, model: str, credentials: dict, language: Optional[str] = None) -> list:
"""
Get voice for given tts model voices
Retrieves the list of voices supported by a given text-to-speech (TTS) model.
:param language: tts language
:param model: model name
:param credentials: model credentials
:return: voices lists
:param language: The language for which the voices are requested.
:param model: The name of the TTS model.
:param credentials: The credentials required to access the TTS model.
:return: A list of voices supported by the TTS model.
"""
model_schema = self.get_model_schema(model, credentials)
if model_schema and ModelPropertyKey.VOICES in model_schema.model_properties:
voices = model_schema.model_properties[ModelPropertyKey.VOICES]
if language:
return [
{"name": d["name"], "value": d["mode"]}
for d in voices
if language and language in d.get("language")
]
else:
return [{"name": d["name"], "value": d["mode"]} for d in voices]
if not model_schema or ModelPropertyKey.VOICES not in model_schema.model_properties:
raise ValueError("this model does not support voice")
voices = model_schema.model_properties[ModelPropertyKey.VOICES]
if language:
return [
{"name": d["name"], "value": d["mode"]} for d in voices if language and language in d.get("language")
]
else:
return [{"name": d["name"], "value": d["mode"]} for d in voices]
def _get_model_default_voice(self, model: str, credentials: dict) -> Any:
"""
@ -111,8 +124,10 @@ class TTSModel(AIModel):
"""
model_schema = self.get_model_schema(model, credentials)
if model_schema and ModelPropertyKey.AUDIO_TYPE in model_schema.model_properties:
return model_schema.model_properties[ModelPropertyKey.AUDIO_TYPE]
if not model_schema or ModelPropertyKey.AUDIO_TYPE not in model_schema.model_properties:
raise ValueError("this model does not support audio type")
return model_schema.model_properties[ModelPropertyKey.AUDIO_TYPE]
def _get_model_word_limit(self, model: str, credentials: dict) -> int:
"""
@ -121,8 +136,10 @@ class TTSModel(AIModel):
"""
model_schema = self.get_model_schema(model, credentials)
if model_schema and ModelPropertyKey.WORD_LIMIT in model_schema.model_properties:
return model_schema.model_properties[ModelPropertyKey.WORD_LIMIT]
if not model_schema or ModelPropertyKey.WORD_LIMIT not in model_schema.model_properties:
raise ValueError("this model does not support word limit")
return model_schema.model_properties[ModelPropertyKey.WORD_LIMIT]
def _get_model_workers_limit(self, model: str, credentials: dict) -> int:
"""
@ -131,8 +148,10 @@ class TTSModel(AIModel):
"""
model_schema = self.get_model_schema(model, credentials)
if model_schema and ModelPropertyKey.MAX_WORKERS in model_schema.model_properties:
return model_schema.model_properties[ModelPropertyKey.MAX_WORKERS]
if not model_schema or ModelPropertyKey.MAX_WORKERS not in model_schema.model_properties:
raise ValueError("this model does not support max workers")
return model_schema.model_properties[ModelPropertyKey.MAX_WORKERS]
@staticmethod
def _split_text_into_sentences(org_text, max_length=2000, pattern=r"[。.!?]"):

1
api/core/model_runtime/model_providers/anthropic/llm/_position.yaml
View File

@ -1,3 +1,4 @@
- claude-3-5-sonnet-20241022
- claude-3-5-sonnet-20240620
- claude-3-haiku-20240307
- claude-3-opus-20240229

2
api/core/model_runtime/model_providers/azure_ai_studio/llm/llm.py
View File

@ -294,7 +294,7 @@ class AzureAIStudioLargeLanguageModel(LargeLanguageModel):
],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
Used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/azure_ai_studio/rerank/rerank.py
View File

@ -148,7 +148,7 @@ class AzureRerankModel(RerankModel):
InvokeBadRequestError: [InvokeBadRequestError, KeyError, ValueError, json.JSONDecodeError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

3
api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml
View File

@ -53,6 +53,9 @@ model_credential_schema:
type: select
required: true
options:
- label:
en_US: 2024-10-01-preview
value: 2024-10-01-preview
- label:
en_US: 2024-09-01-preview
value: 2024-09-01-preview

27
api/core/model_runtime/model_providers/azure_openai/llm/llm.py
View File

@ -45,9 +45,7 @@ class AzureOpenAILargeLanguageModel(_CommonAzureOpenAI, LargeLanguageModel):
stream: bool = True,
user: Optional[str] = None,
) -> Union[LLMResult, Generator]:
base_model_name = credentials.get("base_model_name")
if not base_model_name:
raise ValueError("Base Model Name is required")
base_model_name = self._get_base_model_name(credentials)
ai_model_entity = self._get_ai_model_entity(base_model_name=base_model_name, model=model)
if ai_model_entity and ai_model_entity.entity.model_properties.get(ModelPropertyKey.MODE) == LLMMode.CHAT.value:
@ -81,9 +79,7 @@ class AzureOpenAILargeLanguageModel(_CommonAzureOpenAI, LargeLanguageModel):
prompt_messages: list[PromptMessage],
tools: Optional[list[PromptMessageTool]] = None,
) -> int:
base_model_name = credentials.get("base_model_name")
if not base_model_name:
raise ValueError("Base Model Name is required")
base_model_name = self._get_base_model_name(credentials)
model_entity = self._get_ai_model_entity(base_model_name=base_model_name, model=model)
if not model_entity:
raise ValueError(f"Base Model Name {base_model_name} is invalid")
@ -108,9 +104,7 @@ class AzureOpenAILargeLanguageModel(_CommonAzureOpenAI, LargeLanguageModel):
if "base_model_name" not in credentials:
raise CredentialsValidateFailedError("Base Model Name is required")
base_model_name = credentials.get("base_model_name")
if not base_model_name:
raise CredentialsValidateFailedError("Base Model Name is required")
base_model_name = self._get_base_model_name(credentials)
ai_model_entity = self._get_ai_model_entity(base_model_name=base_model_name, model=model)
if not ai_model_entity:
@ -149,9 +143,7 @@ class AzureOpenAILargeLanguageModel(_CommonAzureOpenAI, LargeLanguageModel):
raise CredentialsValidateFailedError(str(ex))
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
base_model_name = credentials.get("base_model_name")
if not base_model_name:
raise ValueError("Base Model Name is required")
base_model_name = self._get_base_model_name(credentials)
ai_model_entity = self._get_ai_model_entity(base_model_name=base_model_name, model=model)
return ai_model_entity.entity if ai_model_entity else None
@ -308,11 +300,6 @@ class AzureOpenAILargeLanguageModel(_CommonAzureOpenAI, LargeLanguageModel):
if tools:
extra_model_kwargs["tools"] = [helper.dump_model(PromptMessageFunction(function=tool)) for tool in tools]
# extra_model_kwargs['functions'] = [{
# "name": tool.name,
# "description": tool.description,
# "parameters": tool.parameters
# } for tool in tools]
if stop:
extra_model_kwargs["stop"] = stop
@ -769,3 +756,9 @@ class AzureOpenAILargeLanguageModel(_CommonAzureOpenAI, LargeLanguageModel):
ai_model_entity_copy.entity.label.en_US = model
ai_model_entity_copy.entity.label.zh_Hans = model
return ai_model_entity_copy
def _get_base_model_name(self, credentials: dict) -> str:
base_model_name = credentials.get("base_model_name")
if not base_model_name:
raise ValueError("Base Model Name is required")
return base_model_name

1
api/core/model_runtime/model_providers/groq/groq.yaml
View File

@ -18,6 +18,7 @@ help:
en_US: https://console.groq.com/
supported_model_types:
- llm
- speech2text
configurate_methods:
- predefined-model
provider_credential_schema:

2
api/core/model_runtime/model_providers/huggingface_tei/rerank/rerank.py
View File

@ -118,7 +118,7 @@ class HuggingfaceTeiRerankModel(RerankModel):
InvokeBadRequestError: [InvokeBadRequestError, KeyError, ValueError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py
View File

@ -189,7 +189,7 @@ class HuggingfaceTeiTextEmbeddingModel(TextEmbeddingModel):
return usage
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

4
api/core/model_runtime/model_providers/localai/llm/llm.py
View File

@ -1,5 +1,5 @@
from collections.abc import Generator
from typing import cast
from typing import Optional, cast
from httpx import Timeout
from openai import (
@ -212,7 +212,7 @@ class LocalAILanguageModel(LargeLanguageModel):
except Exception as ex:
raise CredentialsValidateFailedError(f"Invalid credentials {str(ex)}")
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
completion_model = None
if credentials["completion_type"] == "chat_completion":
completion_model = LLMMode.CHAT.value

2
api/core/model_runtime/model_providers/localai/speech2text/speech2text.py
View File

@ -73,7 +73,7 @@ class LocalAISpeech2text(Speech2TextModel):
InvokeBadRequestError: [InvokeBadRequestError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py
View File

@ -115,7 +115,7 @@ class LocalAITextEmbeddingModel(TextEmbeddingModel):
num_tokens += self._get_num_tokens_by_gpt2(text)
return num_tokens
def _get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def _get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
Get customizable model schema

5
api/core/model_runtime/model_providers/moonshot/llm/llm.py
View File

@ -44,13 +44,16 @@ class MoonshotLargeLanguageModel(OAIAPICompatLargeLanguageModel):
self._add_custom_parameters(credentials)
self._add_function_call(model, credentials)
user = user[:32] if user else None
# {"response_format": "json_object"} need convert to {"response_format": {"type": "json_object"}}
if "response_format" in model_parameters:
model_parameters["response_format"] = {"type": model_parameters.get("response_format")}
return super()._invoke(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
def validate_credentials(self, model: str, credentials: dict) -> None:
self._add_custom_parameters(credentials)
super().validate_credentials(model, credentials)
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
return AIModelEntity(
model=model,
label=I18nObject(en_US=model, zh_Hans=model),

1
api/core/model_runtime/model_providers/openai/llm/_position.yaml
View File

@ -1,3 +1,4 @@
- gpt-4o-audio-preview
- gpt-4
- gpt-4o
- gpt-4o-2024-05-13

29
api/core/model_runtime/model_providers/openai/llm/llm.py
View File

@ -1,7 +1,7 @@
import json
import logging
from collections.abc import Generator
from typing import Optional, Union, cast
from typing import Any, Optional, Union, cast
import tiktoken
from openai import OpenAI, Stream
@ -11,9 +11,9 @@ from openai.types.chat.chat_completion_chunk import ChoiceDeltaFunctionCall, Cho
from openai.types.chat.chat_completion_message import FunctionCall
from core.model_runtime.callbacks.base_callback import Callback
from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, LLMResultChunk, LLMResultChunkDelta
from core.model_runtime.entities.message_entities import (
from core.model_runtime.entities import (
AssistantPromptMessage,
AudioPromptMessageContent,
ImagePromptMessageContent,
PromptMessage,
PromptMessageContentType,
@ -23,6 +23,7 @@ from core.model_runtime.entities.message_entities import (
ToolPromptMessage,
UserPromptMessage,
)
from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, LLMResultChunk, LLMResultChunkDelta
from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, I18nObject, ModelType, PriceConfig
from core.model_runtime.errors.validate import CredentialsValidateFailedError
from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
@ -613,6 +614,7 @@ class OpenAILargeLanguageModel(_CommonOpenAI, LargeLanguageModel):
# clear illegal prompt messages
prompt_messages = self._clear_illegal_prompt_messages(model, prompt_messages)
# o1 compatibility
block_as_stream = False
if model.startswith("o1"):
if stream:
@ -626,8 +628,9 @@ class OpenAILargeLanguageModel(_CommonOpenAI, LargeLanguageModel):
del extra_model_kwargs["stop"]
# chat model
messages: Any = [self._convert_prompt_message_to_dict(m) for m in prompt_messages]
response = client.chat.completions.create(
messages=[self._convert_prompt_message_to_dict(m) for m in prompt_messages],
messages=messages,
model=model,
stream=stream,
**model_parameters,
@ -946,23 +949,29 @@ class OpenAILargeLanguageModel(_CommonOpenAI, LargeLanguageModel):
Convert PromptMessage to dict for OpenAI API
"""
if isinstance(message, UserPromptMessage):
message = cast(UserPromptMessage, message)
if isinstance(message.content, str):
message_dict = {"role": "user", "content": message.content}
else:
elif isinstance(message.content, list):
sub_messages = []
for message_content in message.content:
if message_content.type == PromptMessageContentType.TEXT:
message_content = cast(TextPromptMessageContent, message_content)
if isinstance(message_content, TextPromptMessageContent):
sub_message_dict = {"type": "text", "text": message_content.data}
sub_messages.append(sub_message_dict)
elif message_content.type == PromptMessageContentType.IMAGE:
message_content = cast(ImagePromptMessageContent, message_content)
elif isinstance(message_content, ImagePromptMessageContent):
sub_message_dict = {
"type": "image_url",
"image_url": {"url": message_content.data, "detail": message_content.detail.value},
}
sub_messages.append(sub_message_dict)
elif isinstance(message_content, AudioPromptMessageContent):
sub_message_dict = {
"type": "input_audio",
"input_audio": {
"data": message_content.data,
"format": message_content.format,
},
}
sub_messages.append(sub_message_dict)
message_dict = {"role": "user", "content": sub_messages}
elif isinstance(message, AssistantPromptMessage):

2
api/core/model_runtime/model_providers/openai/speech2text/speech2text.py
View File

@ -61,7 +61,7 @@ class OpenAISpeech2TextModel(_CommonOpenAI, Speech2TextModel):
return response.text
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

33
api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py
View File

@ -397,16 +397,21 @@ class OAIAPICompatLargeLanguageModel(_CommonOaiApiCompat, LargeLanguageModel):
chunk_index = 0
def create_final_llm_result_chunk(
index: int, message: AssistantPromptMessage, finish_reason: str
id: Optional[str], index: int, message: AssistantPromptMessage, finish_reason: str, usage: dict
) -> LLMResultChunk:
# calculate num tokens
prompt_tokens = self._num_tokens_from_string(model, prompt_messages[0].content)
completion_tokens = self._num_tokens_from_string(model, full_assistant_content)
prompt_tokens = usage and usage.get("prompt_tokens")
if prompt_tokens is None:
prompt_tokens = self._num_tokens_from_string(model, prompt_messages[0].content)
completion_tokens = usage and usage.get("completion_tokens")
if completion_tokens is None:
completion_tokens = self._num_tokens_from_string(model, full_assistant_content)
# transform usage
usage = self._calc_response_usage(model, credentials, prompt_tokens, completion_tokens)
return LLMResultChunk(
id=id,
model=model,
prompt_messages=prompt_messages,
delta=LLMResultChunkDelta(index=index, message=message, finish_reason=finish_reason, usage=usage),
@ -450,7 +455,7 @@ class OAIAPICompatLargeLanguageModel(_CommonOaiApiCompat, LargeLanguageModel):
tool_call.function.arguments += new_tool_call.function.arguments
finish_reason = None # The default value of finish_reason is None
message_id, usage = None, None
for chunk in response.iter_lines(decode_unicode=True, delimiter=delimiter):
chunk = chunk.strip()
if chunk:
@ -462,20 +467,26 @@ class OAIAPICompatLargeLanguageModel(_CommonOaiApiCompat, LargeLanguageModel):
continue
try:
chunk_json = json.loads(decoded_chunk)
chunk_json: dict = json.loads(decoded_chunk)
# stream ended
except json.JSONDecodeError as e:
yield create_final_llm_result_chunk(
id=message_id,
index=chunk_index + 1,
message=AssistantPromptMessage(content=""),
finish_reason="Non-JSON encountered.",
usage=usage,
)
break
if chunk_json:
if u := chunk_json.get("usage"):
usage = u
if not chunk_json or len(chunk_json["choices"]) == 0:
continue
choice = chunk_json["choices"][0]
finish_reason = chunk_json["choices"][0].get("finish_reason")
message_id = chunk_json.get("id")
chunk_index += 1
if "delta" in choice:
@ -524,6 +535,7 @@ class OAIAPICompatLargeLanguageModel(_CommonOaiApiCompat, LargeLanguageModel):
continue
yield LLMResultChunk(
id=message_id,
model=model,
prompt_messages=prompt_messages,
delta=LLMResultChunkDelta(
@ -536,6 +548,7 @@ class OAIAPICompatLargeLanguageModel(_CommonOaiApiCompat, LargeLanguageModel):
if tools_calls:
yield LLMResultChunk(
id=message_id,
model=model,
prompt_messages=prompt_messages,
delta=LLMResultChunkDelta(
@ -545,17 +558,22 @@ class OAIAPICompatLargeLanguageModel(_CommonOaiApiCompat, LargeLanguageModel):
)
yield create_final_llm_result_chunk(
index=chunk_index, message=AssistantPromptMessage(content=""), finish_reason=finish_reason
id=message_id,
index=chunk_index,
message=AssistantPromptMessage(content=""),
finish_reason=finish_reason,
usage=usage,
)
def _handle_generate_response(
self, model: str, credentials: dict, response: requests.Response, prompt_messages: list[PromptMessage]
) -> LLMResult:
response_json = response.json()
response_json: dict = response.json()
completion_type = LLMMode.value_of(credentials["mode"])
output = response_json["choices"][0]
message_id = response_json.get("id")
response_content = ""
tool_calls = None
@ -593,6 +611,7 @@ class OAIAPICompatLargeLanguageModel(_CommonOaiApiCompat, LargeLanguageModel):
# transform response
result = LLMResult(
id=message_id,
model=response_json["model"],
prompt_messages=prompt_messages,
message=assistant_message,

14
api/core/model_runtime/model_providers/openai_api_compatible/openai_api_compatible.yaml
View File

@ -8,6 +8,7 @@ supported_model_types:
- llm
- text-embedding
- speech2text
- rerank
configurate_methods:
- customizable-model
model_credential_schema:
@ -83,6 +84,19 @@ model_credential_schema:
placeholder:
zh_Hans: 在此输入您的模型上下文长度
en_US: Enter your Model context size
- variable: context_size
label:
zh_Hans: 模型上下文长度
en_US: Model context size
required: true
show_on:
- variable: __model_type
value: rerank
type: text-input
default: '4096'
placeholder:
zh_Hans: 在此输入您的模型上下文长度
en_US: Enter your Model context size
- variable: max_tokens_to_sample
label:
zh_Hans: 最大 token 上限

2
api/core/model_runtime/model_providers/openai_api_compatible/speech2text/speech2text.py
View File

@ -62,7 +62,7 @@ class OAICompatSpeech2TextModel(_CommonOaiApiCompat, Speech2TextModel):
except Exception as ex:
raise CredentialsValidateFailedError(str(ex))
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

3
api/core/model_runtime/model_providers/openllm/llm/llm.py
View File

@ -1,4 +1,5 @@
from collections.abc import Generator
from typing import Optional
from core.model_runtime.entities.common_entities import I18nObject
from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, LLMResultChunk, LLMResultChunkDelta
@ -193,7 +194,7 @@ class OpenLLMLargeLanguageModel(LargeLanguageModel):
),
)
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/sagemaker/llm/llm.py
View File

@ -408,7 +408,7 @@ class SageMakerLargeLanguageModel(LargeLanguageModel):
InvokeBadRequestError: [InvokeBadRequestError, KeyError, ValueError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/sagemaker/rerank/rerank.py
View File

@ -157,7 +157,7 @@ class SageMakerRerankModel(RerankModel):
InvokeBadRequestError: [InvokeBadRequestError, KeyError, ValueError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/sagemaker/speech2text/speech2text.py
View File

@ -111,7 +111,7 @@ class SageMakerSpeech2TextModel(Speech2TextModel):
InvokeBadRequestError: [InvokeBadRequestError, KeyError, ValueError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py
View File

@ -180,7 +180,7 @@ class SageMakerEmbeddingModel(TextEmbeddingModel):
InvokeBadRequestError: [KeyError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/sagemaker/tts/tts.py
View File

@ -159,7 +159,7 @@ class SageMakerText2SpeechModel(TTSModel):
return self._tts_invoke_streaming(model_type, payload, sagemaker_endpoint)
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/siliconflow/llm/llm.py
View File

@ -40,7 +40,7 @@ class SiliconflowLargeLanguageModel(OAIAPICompatLargeLanguageModel):
credentials["mode"] = "chat"
credentials["endpoint_url"] = "https://api.siliconflow.cn/v1"
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
return AIModelEntity(
model=model,
label=I18nObject(en_US=model, zh_Hans=model),

2
api/core/model_runtime/model_providers/stepfun/llm/llm.py
View File

@ -50,7 +50,7 @@ class StepfunLargeLanguageModel(OAIAPICompatLargeLanguageModel):
self._add_custom_parameters(credentials)
super().validate_credentials(model, credentials)
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
return AIModelEntity(
model=model,
label=I18nObject(en_US=model, zh_Hans=model),

2
api/core/model_runtime/model_providers/tongyi/llm/llm.py
View File

@ -535,7 +535,7 @@ class TongyiLargeLanguageModel(LargeLanguageModel):
],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
Architecture for defining customizable models

1
api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
View File

@ -76,3 +76,4 @@ pricing:
output: '0.12'
unit: '0.001'
currency: RMB
deprecated: true

4
api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml
View File

@ -10,7 +10,7 @@ features:
- stream-tool-call
model_properties:
mode: chat
context_size: 8000
context_size: 32000
parameter_rules:
- name: temperature
use_template: temperature
@ -26,7 +26,7 @@ parameter_rules:
type: int
default: 2000
min: 1
max: 2000
max: 8192
help:
zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.

2
api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml
View File

@ -10,7 +10,7 @@ features:
- stream-tool-call
model_properties:
mode: chat
context_size: 131072
context_size: 128000
parameter_rules:
- name: temperature
use_template: temperature

4
api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml
View File

@ -10,7 +10,7 @@ features:
- stream-tool-call
model_properties:
mode: chat
context_size: 8000
context_size: 128000
parameter_rules:
- name: temperature
use_template: temperature
@ -26,7 +26,7 @@ parameter_rules:
type: int
default: 2000
min: 1
max: 2000
max: 8192
help:
zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.

3
api/core/model_runtime/model_providers/triton_inference_server/llm/llm.py
View File

@ -1,4 +1,5 @@
from collections.abc import Generator
from typing import Optional
from httpx import Response, post
from yarl import URL
@ -109,7 +110,7 @@ class TritonInferenceAILargeLanguageModel(LargeLanguageModel):
raise NotImplementedError(f"PromptMessage type {type(item)} is not supported")
return text
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

3
api/core/model_runtime/model_providers/volcengine_maas/llm/llm.py
View File

@ -1,5 +1,6 @@
import logging
from collections.abc import Generator
from typing import Optional
from volcenginesdkarkruntime.types.chat import ChatCompletion, ChatCompletionChunk
@ -298,7 +299,7 @@ class VolcengineMaaSLargeLanguageModel(LargeLanguageModel):
chunks = client.stream_chat(prompt_messages, **req_params)
return _handle_stream_chat_response(chunks)
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

41
api/core/model_runtime/model_providers/wenxin/rerank/rerank.py
View File

@ -2,20 +2,15 @@ from typing import Optional
import httpx
from core.model_runtime.entities.common_entities import I18nObject
from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType
from core.model_runtime.entities.rerank_entities import RerankDocument, RerankResult
from core.model_runtime.errors.invoke import (
InvokeAuthorizationError,
InvokeBadRequestError,
InvokeConnectionError,
InvokeError,
InvokeRateLimitError,
InvokeServerUnavailableError,
)
from core.model_runtime.errors.invoke import InvokeError
from core.model_runtime.errors.validate import CredentialsValidateFailedError
from core.model_runtime.model_providers.__base.rerank_model import RerankModel
from core.model_runtime.model_providers.wenxin._common import _CommonWenxin
from core.model_runtime.model_providers.wenxin.wenxin_errors import (
InternalServerError,
invoke_error_mapping,
)
class WenxinRerank(_CommonWenxin):
@ -32,7 +27,7 @@ class WenxinRerank(_CommonWenxin):
response.raise_for_status()
return response.json()
except httpx.HTTPStatusError as e:
raise InvokeServerUnavailableError(str(e))
raise InternalServerError(str(e))
class WenxinRerankModel(RerankModel):
@ -93,7 +88,7 @@ class WenxinRerankModel(RerankModel):
return RerankResult(model=model, docs=rerank_documents)
except httpx.HTTPStatusError as e:
raise InvokeServerUnavailableError(str(e))
raise InternalServerError(str(e))
def validate_credentials(self, model: str, credentials: dict) -> None:
"""
@ -124,24 +119,4 @@ class WenxinRerankModel(RerankModel):
"""
Map model invoke error to unified error
"""
return {
InvokeConnectionError: [httpx.ConnectError],
InvokeServerUnavailableError: [httpx.RemoteProtocolError],
InvokeRateLimitError: [],
InvokeAuthorizationError: [httpx.HTTPStatusError],
InvokeBadRequestError: [httpx.RequestError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity:
"""
generate custom model entities from credentials
"""
entity = AIModelEntity(
model=model,
label=I18nObject(en_US=model),
model_type=ModelType.RERANK,
fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
model_properties={ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size"))},
)
return entity
return invoke_error_mapping()

4
api/core/model_runtime/model_providers/xinference/llm/llm.py
View File

@ -1,5 +1,5 @@
from collections.abc import Generator, Iterator
from typing import cast
from typing import Optional, cast
from openai import (
APIConnectionError,
@ -321,7 +321,7 @@ class XinferenceAILargeLanguageModel(LargeLanguageModel):
return message_dict
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/xinference/rerank/rerank.py
View File

@ -142,7 +142,7 @@ class XinferenceRerankModel(RerankModel):
InvokeBadRequestError: [InvokeBadRequestError, KeyError, ValueError],
}
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/xinference/speech2text/speech2text.py
View File

@ -129,7 +129,7 @@ class XinferenceSpeech2TextModel(Speech2TextModel):
return response["text"]
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py
View File

@ -184,7 +184,7 @@ class XinferenceTextEmbeddingModel(TextEmbeddingModel):
return usage
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

2
api/core/model_runtime/model_providers/xinference/tts/tts.py
View File

@ -116,7 +116,7 @@ class XinferenceText2SpeechModel(TTSModel):
"""
return self._tts_invoke_streaming(model, credentials, content_text, voice)
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
"""
used to define customizable model schema
"""

67
api/core/model_runtime/model_providers/yi/llm/llm.py
View File

@ -4,12 +4,22 @@ from urllib.parse import urlparse
import tiktoken
from core.model_runtime.entities.llm_entities import LLMResult
from core.model_runtime.entities.common_entities import I18nObject
from core.model_runtime.entities.llm_entities import LLMMode, LLMResult
from core.model_runtime.entities.message_entities import (
PromptMessage,
PromptMessageTool,
SystemPromptMessage,
)
from core.model_runtime.entities.model_entities import (
AIModelEntity,
FetchFrom,
ModelFeature,
ModelPropertyKey,
ModelType,
ParameterRule,
ParameterType,
)
from core.model_runtime.model_providers.openai.llm.llm import OpenAILargeLanguageModel
@ -125,3 +135,58 @@ class YiLargeLanguageModel(OpenAILargeLanguageModel):
else:
parsed_url = urlparse(credentials["endpoint_url"])
credentials["openai_api_base"] = f"{parsed_url.scheme}://{parsed_url.netloc}"
def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
return AIModelEntity(
model=model,
label=I18nObject(en_US=model, zh_Hans=model),
model_type=ModelType.LLM,
features=[ModelFeature.TOOL_CALL, ModelFeature.MULTI_TOOL_CALL, ModelFeature.STREAM_TOOL_CALL]
if credentials.get("function_calling_type") == "tool_call"
else [],
fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
model_properties={
ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size", 8000)),
ModelPropertyKey.MODE: LLMMode.CHAT.value,
},
parameter_rules=[
ParameterRule(
name="temperature",
use_template="temperature",
label=I18nObject(en_US="Temperature", zh_Hans="温度"),
type=ParameterType.FLOAT,
),
ParameterRule(
name="max_tokens",
use_template="max_tokens",
default=512,
min=1,
max=int(credentials.get("max_tokens", 8192)),
label=I18nObject(
en_US="Max Tokens", zh_Hans="指定生成结果长度的上限。如果生成结果截断，可以调大该参数"
),
type=ParameterType.INT,
),
ParameterRule(
name="top_p",
use_template="top_p",
label=I18nObject(
en_US="Top P",
zh_Hans="控制生成结果的随机性。数值越小，随机性越弱；数值越大，随机性越强。",
),
type=ParameterType.FLOAT,
),
ParameterRule(
name="top_k",
use_template="top_k",
label=I18nObject(en_US="Top K", zh_Hans="取样数量"),
type=ParameterType.FLOAT,
),
ParameterRule(
name="frequency_penalty",
use_template="frequency_penalty",
label=I18nObject(en_US="Frequency Penalty", zh_Hans="重复惩罚"),
type=ParameterType.FLOAT,
),
],
)

55
api/core/model_runtime/model_providers/yi/yi.yaml
View File

@ -20,6 +20,7 @@ supported_model_types:
- llm
configurate_methods:
- predefined-model
- customizable-model
provider_credential_schema:
credential_form_schemas:
- variable: api_key
@ -39,3 +40,57 @@ provider_credential_schema:
placeholder:
zh_Hans: Base URL, e.g. https://api.lingyiwanwu.com/v1
en_US: Base URL, e.g. https://api.lingyiwanwu.com/v1
model_credential_schema:
model:
label:
en_US: Model Name
zh_Hans: 模型名称
placeholder:
en_US: Enter your model name
zh_Hans: 输入模型名称
credential_form_schemas:
- variable: api_key
label:
en_US: API Key
type: secret-input
required: true
placeholder:
zh_Hans: 在此输入您的 API Key
en_US: Enter your API Key
- variable: context_size
label:
zh_Hans: 模型上下文长度
en_US: Model context size
required: true
type: text-input
default: '4096'
placeholder:
zh_Hans: 在此输入您的模型上下文长度
en_US: Enter your Model context size
- variable: max_tokens
label:
zh_Hans: 最大 token 上限
en_US: Upper bound for max tokens
default: '4096'
type: text-input
show_on:
- variable: __model_type
value: llm
- variable: function_calling_type
label:
en_US: Function calling
type: select
required: false
default: no_call
options:
- value: no_call
label:
en_US: Not Support
zh_Hans: 不支持
- value: function_call
label:
en_US: Support
zh_Hans: 支持
show_on:
- variable: __model_type
value: llm

4
api/core/ops/ops_trace_manager.py
View File

@ -358,8 +358,8 @@ class TraceTask:
workflow_run_id = workflow_run.id
workflow_run_elapsed_time = workflow_run.elapsed_time
workflow_run_status = workflow_run.status
workflow_run_inputs = json.loads(workflow_run.inputs) if workflow_run.inputs else {}
workflow_run_outputs = json.loads(workflow_run.outputs) if workflow_run.outputs else {}
workflow_run_inputs = workflow_run.inputs_dict
workflow_run_outputs = workflow_run.outputs_dict
workflow_run_version = workflow_run.version
error = workflow_run.error or ""

121
api/core/prompt/advanced_prompt_transform.py
View File

@ -1,12 +1,15 @@
from typing import Optional, Union
from collections.abc import Sequence
from typing import Optional
from core.app.entities.app_invoke_entities import ModelConfigWithCredentialsEntity
from core.file.file_obj import FileVar
from core.file import file_manager
from core.file.models import File
from core.helper.code_executor.jinja2.jinja2_formatter import Jinja2Formatter
from core.memory.token_buffer_memory import TokenBufferMemory
from core.model_runtime.entities.message_entities import (
from core.model_runtime.entities import (
AssistantPromptMessage,
PromptMessage,
PromptMessageContent,
PromptMessageRole,
SystemPromptMessage,
TextPromptMessageContent,
@ -14,8 +17,8 @@ from core.model_runtime.entities.message_entities import (
)
from core.prompt.entities.advanced_prompt_entities import ChatModelMessage, CompletionModelPromptTemplate, MemoryConfig
from core.prompt.prompt_transform import PromptTransform
from core.prompt.simple_prompt_transform import ModelMode
from core.prompt.utils.prompt_template_parser import PromptTemplateParser
from core.workflow.entities.variable_pool import VariablePool
class AdvancedPromptTransform(PromptTransform):
@ -28,22 +31,19 @@ class AdvancedPromptTransform(PromptTransform):
def get_prompt(
self,
prompt_template: Union[list[ChatModelMessage], CompletionModelPromptTemplate],
inputs: dict,
*,
prompt_template: Sequence[ChatModelMessage] | CompletionModelPromptTemplate,
inputs: dict[str, str],
query: str,
files: list[FileVar],
files: Sequence[File],
context: Optional[str],
memory_config: Optional[MemoryConfig],
memory: Optional[TokenBufferMemory],
model_config: ModelConfigWithCredentialsEntity,
query_prompt_template: Optional[str] = None,
) -> list[PromptMessage]:
inputs = {key: str(value) for key, value in inputs.items()}
prompt_messages = []
model_mode = ModelMode.value_of(model_config.mode)
if model_mode == ModelMode.COMPLETION:
if isinstance(prompt_template, CompletionModelPromptTemplate):
prompt_messages = self._get_completion_model_prompt_messages(
prompt_template=prompt_template,
inputs=inputs,
@ -54,12 +54,11 @@ class AdvancedPromptTransform(PromptTransform):
memory=memory,
model_config=model_config,
)
elif model_mode == ModelMode.CHAT:
elif isinstance(prompt_template, list) and all(isinstance(item, ChatModelMessage) for item in prompt_template):
prompt_messages = self._get_chat_model_prompt_messages(
prompt_template=prompt_template,
inputs=inputs,
query=query,
query_prompt_template=query_prompt_template,
files=files,
context=context,
memory_config=memory_config,
@ -74,7 +73,7 @@ class AdvancedPromptTransform(PromptTransform):
prompt_template: CompletionModelPromptTemplate,
inputs: dict,
query: Optional[str],
files: list[FileVar],
files: Sequence[File],
context: Optional[str],
memory_config: Optional[MemoryConfig],
memory: Optional[TokenBufferMemory],
@ -88,10 +87,10 @@ class AdvancedPromptTransform(PromptTransform):
prompt_messages = []
if prompt_template.edition_type == "basic" or not prompt_template.edition_type:
prompt_template = PromptTemplateParser(template=raw_prompt, with_variable_tmpl=self.with_variable_tmpl)
prompt_inputs = {k: inputs[k] for k in prompt_template.variable_keys if k in inputs}
parser = PromptTemplateParser(template=raw_prompt, with_variable_tmpl=self.with_variable_tmpl)
prompt_inputs = {k: inputs[k] for k in parser.variable_keys if k in inputs}
prompt_inputs = self._set_context_variable(context, prompt_template, prompt_inputs)
prompt_inputs = self._set_context_variable(context, parser, prompt_inputs)
if memory and memory_config:
role_prefix = memory_config.role_prefix
@ -100,15 +99,15 @@ class AdvancedPromptTransform(PromptTransform):
memory_config=memory_config,
raw_prompt=raw_prompt,
role_prefix=role_prefix,
prompt_template=prompt_template,
parser=parser,
prompt_inputs=prompt_inputs,
model_config=model_config,
)
if query:
prompt_inputs = self._set_query_variable(query, prompt_template, prompt_inputs)
prompt_inputs = self._set_query_variable(query, parser, prompt_inputs)
prompt = prompt_template.format(prompt_inputs)
prompt = parser.format(prompt_inputs)
else:
prompt = raw_prompt
prompt_inputs = inputs
@ -116,9 +115,10 @@ class AdvancedPromptTransform(PromptTransform):
prompt = Jinja2Formatter.format(prompt, prompt_inputs)
if files:
prompt_message_contents = [TextPromptMessageContent(data=prompt)]
prompt_message_contents: list[PromptMessageContent] = []
prompt_message_contents.append(TextPromptMessageContent(data=prompt))
for file in files:
prompt_message_contents.append(file.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file))
prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
else:
@ -131,35 +131,38 @@ class AdvancedPromptTransform(PromptTransform):
prompt_template: list[ChatModelMessage],
inputs: dict,
query: Optional[str],
files: list[FileVar],
files: Sequence[File],
context: Optional[str],
memory_config: Optional[MemoryConfig],
memory: Optional[TokenBufferMemory],
model_config: ModelConfigWithCredentialsEntity,
query_prompt_template: Optional[str] = None,
) -> list[PromptMessage]:
"""
Get chat model prompt messages.
"""
raw_prompt_list = prompt_template
prompt_messages = []
for prompt_item in raw_prompt_list:
for prompt_item in prompt_template:
raw_prompt = prompt_item.text
if prompt_item.edition_type == "basic" or not prompt_item.edition_type:
prompt_template = PromptTemplateParser(template=raw_prompt, with_variable_tmpl=self.with_variable_tmpl)
prompt_inputs = {k: inputs[k] for k in prompt_template.variable_keys if k in inputs}
prompt_inputs = self._set_context_variable(context, prompt_template, prompt_inputs)
prompt = prompt_template.format(prompt_inputs)
if self.with_variable_tmpl:
vp = VariablePool()
for k, v in inputs.items():
if k.startswith("#"):
vp.add(k[1:-1].split("."), v)
raw_prompt = raw_prompt.replace("{{#context#}}", context or "")
prompt = vp.convert_template(raw_prompt).text
else:
parser = PromptTemplateParser(template=raw_prompt, with_variable_tmpl=self.with_variable_tmpl)
prompt_inputs = {k: inputs[k] for k in parser.variable_keys if k in inputs}
prompt_inputs = self._set_context_variable(
context=context, parser=parser, prompt_inputs=prompt_inputs
)
prompt = parser.format(prompt_inputs)
elif prompt_item.edition_type == "jinja2":
prompt = raw_prompt
prompt_inputs = inputs
prompt = Jinja2Formatter.format(prompt, prompt_inputs)
prompt = Jinja2Formatter.format(template=prompt, inputs=prompt_inputs)
else:
raise ValueError(f"Invalid edition type: {prompt_item.edition_type}")
@ -170,25 +173,25 @@ class AdvancedPromptTransform(PromptTransform):
elif prompt_item.role == PromptMessageRole.ASSISTANT:
prompt_messages.append(AssistantPromptMessage(content=prompt))
if query and query_prompt_template:
prompt_template = PromptTemplateParser(
template=query_prompt_template, with_variable_tmpl=self.with_variable_tmpl
if query and memory_config and memory_config.query_prompt_template:
parser = PromptTemplateParser(
template=memory_config.query_prompt_template, with_variable_tmpl=self.with_variable_tmpl
)
prompt_inputs = {k: inputs[k] for k in prompt_template.variable_keys if k in inputs}
prompt_inputs = {k: inputs[k] for k in parser.variable_keys if k in inputs}
prompt_inputs["#sys.query#"] = query
prompt_inputs = self._set_context_variable(context, prompt_template, prompt_inputs)
prompt_inputs = self._set_context_variable(context, parser, prompt_inputs)
query = prompt_template.format(prompt_inputs)
query = parser.format(prompt_inputs)
if memory and memory_config:
prompt_messages = self._append_chat_histories(memory, memory_config, prompt_messages, model_config)
if files:
prompt_message_contents = [TextPromptMessageContent(data=query)]
if files and query is not None:
prompt_message_contents: list[PromptMessageContent] = []
prompt_message_contents.append(TextPromptMessageContent(data=query))
for file in files:
prompt_message_contents.append(file.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file))
prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
else:
prompt_messages.append(UserPromptMessage(content=query))
@ -200,19 +203,19 @@ class AdvancedPromptTransform(PromptTransform):
# get last user message content and add files
prompt_message_contents = [TextPromptMessageContent(data=last_message.content)]
for file in files:
prompt_message_contents.append(file.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file))
last_message.content = prompt_message_contents
else:
prompt_message_contents = [TextPromptMessageContent(data="")] # not for query
for file in files:
prompt_message_contents.append(file.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file))
prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
else:
prompt_message_contents = [TextPromptMessageContent(data=query)]
for file in files:
prompt_message_contents.append(file.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file))
prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
elif query:
@ -220,8 +223,8 @@ class AdvancedPromptTransform(PromptTransform):
return prompt_messages
def _set_context_variable(self, context: str, prompt_template: PromptTemplateParser, prompt_inputs: dict) -> dict:
if "#context#" in prompt_template.variable_keys:
def _set_context_variable(self, context: str | None, parser: PromptTemplateParser, prompt_inputs: dict) -> dict:
if "#context#" in parser.variable_keys:
if context:
prompt_inputs["#context#"] = context
else:
@ -229,8 +232,8 @@ class AdvancedPromptTransform(PromptTransform):
return prompt_inputs
def _set_query_variable(self, query: str, prompt_template: PromptTemplateParser, prompt_inputs: dict) -> dict:
if "#query#" in prompt_template.variable_keys:
def _set_query_variable(self, query: str, parser: PromptTemplateParser, prompt_inputs: dict) -> dict:
if "#query#" in parser.variable_keys:
if query:
prompt_inputs["#query#"] = query
else:
@ -244,16 +247,16 @@ class AdvancedPromptTransform(PromptTransform):
memory_config: MemoryConfig,
raw_prompt: str,
role_prefix: MemoryConfig.RolePrefix,
prompt_template: PromptTemplateParser,
parser: PromptTemplateParser,
prompt_inputs: dict,
model_config: ModelConfigWithCredentialsEntity,
) -> dict:
if "#histories#" in prompt_template.variable_keys:
if "#histories#" in parser.variable_keys:
if memory:
inputs = {"#histories#": "", **prompt_inputs}
prompt_template = PromptTemplateParser(template=raw_prompt, with_variable_tmpl=self.with_variable_tmpl)
prompt_inputs = {k: inputs[k] for k in prompt_template.variable_keys if k in inputs}
tmp_human_message = UserPromptMessage(content=prompt_template.format(prompt_inputs))
parser = PromptTemplateParser(template=raw_prompt, with_variable_tmpl=self.with_variable_tmpl)
prompt_inputs = {k: inputs[k] for k in parser.variable_keys if k in inputs}
tmp_human_message = UserPromptMessage(content=parser.format(prompt_inputs))
rest_tokens = self._calculate_rest_token([tmp_human_message], model_config)

19
api/core/prompt/simple_prompt_transform.py
View File

@ -5,9 +5,11 @@ from typing import TYPE_CHECKING, Optional
from core.app.app_config.entities import PromptTemplateEntity
from core.app.entities.app_invoke_entities import ModelConfigWithCredentialsEntity
from core.file import file_manager
from core.memory.token_buffer_memory import TokenBufferMemory
from core.model_runtime.entities.message_entities import (
PromptMessage,
PromptMessageContent,
SystemPromptMessage,
TextPromptMessageContent,
UserPromptMessage,
@ -18,10 +20,10 @@ from core.prompt.utils.prompt_template_parser import PromptTemplateParser
from models.model import AppMode
if TYPE_CHECKING:
from core.file.file_obj import FileVar
from core.file.models import File
class ModelMode(enum.Enum):
class ModelMode(str, enum.Enum):
COMPLETION = "completion"
CHAT = "chat"
@ -53,7 +55,7 @@ class SimplePromptTransform(PromptTransform):
prompt_template_entity: PromptTemplateEntity,
inputs: dict,
query: str,
files: list["FileVar"],
files: list["File"],
context: Optional[str],
memory: Optional[TokenBufferMemory],
model_config: ModelConfigWithCredentialsEntity,
@ -169,7 +171,7 @@ class SimplePromptTransform(PromptTransform):
inputs: dict,
query: str,
context: Optional[str],
files: list["FileVar"],
files: list["File"],
memory: Optional[TokenBufferMemory],
model_config: ModelConfigWithCredentialsEntity,
) -> tuple[list[PromptMessage], Optional[list[str]]]:
@ -214,7 +216,7 @@ class SimplePromptTransform(PromptTransform):
inputs: dict,
query: str,
context: Optional[str],
files: list["FileVar"],
files: list["File"],
memory: Optional[TokenBufferMemory],
model_config: ModelConfigWithCredentialsEntity,
) -> tuple[list[PromptMessage], Optional[list[str]]]:
@ -261,11 +263,12 @@ class SimplePromptTransform(PromptTransform):
return [self.get_last_user_message(prompt, files)], stops
def get_last_user_message(self, prompt: str, files: list["FileVar"]) -> UserPromptMessage:
def get_last_user_message(self, prompt: str, files: list["File"]) -> UserPromptMessage:
if files:
prompt_message_contents = [TextPromptMessageContent(data=prompt)]
prompt_message_contents: list[PromptMessageContent] = []
prompt_message_contents.append(TextPromptMessageContent(data=prompt))
for file in files:
prompt_message_contents.append(file.prompt_message_content)
prompt_message_contents.append(file_manager.to_prompt_message_content(file))
prompt_message = UserPromptMessage(content=prompt_message_contents)
else:

4
api/core/prompt/utils/extract_thread_messages.py
View File

@ -1,7 +1,9 @@
from typing import Any
from constants import UUID_NIL
def extract_thread_messages(messages: list[dict]) -> list[dict]:
def extract_thread_messages(messages: list[Any]):
thread_messages = []
next_message = None

19
api/core/prompt/utils/prompt_message_util.py
View File

@ -1,7 +1,8 @@
from typing import cast
from core.model_runtime.entities.message_entities import (
from core.model_runtime.entities import (
AssistantPromptMessage,
AudioPromptMessageContent,
ImagePromptMessageContent,
PromptMessage,
PromptMessageContentType,
@ -21,7 +22,7 @@ class PromptMessageUtil:
:return:
"""
prompts = []
if model_mode == ModelMode.CHAT.value:
if model_mode == ModelMode.CHAT:
tool_calls = []
for prompt_message in prompt_messages:
if prompt_message.role == PromptMessageRole.USER:
@ -51,11 +52,9 @@ class PromptMessageUtil:
files = []
if isinstance(prompt_message.content, list):
for content in prompt_message.content:
if content.type == PromptMessageContentType.TEXT:
content = cast(TextPromptMessageContent, content)
if isinstance(content, TextPromptMessageContent):
text += content.data
else:
content = cast(ImagePromptMessageContent, content)
elif isinstance(content, ImagePromptMessageContent):
files.append(
{
"type": "image",
@ -63,6 +62,14 @@ class PromptMessageUtil:
"detail": content.detail.value,
}
)
elif isinstance(content, AudioPromptMessageContent):
files.append(
{
"type": "audio",
"data": content.data[:10] + "...[TRUNCATED]..." + content.data[-10:],
"format": content.format,
}
)
else:
text = prompt_message.content

2
api/core/prompt/utils/prompt_template_parser.py
View File

@ -33,7 +33,7 @@ class PromptTemplateParser:
key = match.group(1)
value = inputs.get(key, match.group(0)) # return original matched string if key not found
if remove_template_variables:
if remove_template_variables and isinstance(value, str):
return PromptTemplateParser.remove_template_variables(value, self.with_variable_tmpl)
return value

Some files were not shown because too many files have changed in this diff Show More