refactor app

api/core/app/validators/dataset_retrieval.py

@@ -0,0 +1,140 @@
+import uuid
+
+from core.entities.agent_entities import PlanningStrategy
+from models.model import AppMode
+from services.dataset_service import DatasetService
+
+
+class DatasetValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, tenant_id: str, app_mode: AppMode, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for dataset feature
+
+        :param tenant_id: tenant ID
+        :param app_mode: app mode
+        :param config: app model config args
+        """
+        # Extract dataset config for legacy compatibility
+        config = cls.extract_dataset_config_for_legacy_compatibility(tenant_id, app_mode, config)
+
+        # dataset_configs
+        if not config.get("dataset_configs"):
+            config["dataset_configs"] = {'retrieval_model': 'single'}
+
+        if not config["dataset_configs"].get("datasets"):
+            config["dataset_configs"]["datasets"] = {
+                "strategy": "router",
+                "datasets": []
+            }
+
+        if not isinstance(config["dataset_configs"], dict):
+            raise ValueError("dataset_configs must be of object type")
+
+        if config["dataset_configs"]['retrieval_model'] == 'multiple':
+            if not config["dataset_configs"]['reranking_model']:
+                raise ValueError("reranking_model has not been set")
+            if not isinstance(config["dataset_configs"]['reranking_model'], dict):
+                raise ValueError("reranking_model must be of object type")
+
+        if not isinstance(config["dataset_configs"], dict):
+            raise ValueError("dataset_configs must be of object type")
+
+        need_manual_query_datasets = config.get("dataset_configs") and config["dataset_configs"].get("datasets")
+
+        if need_manual_query_datasets and app_mode == AppMode.COMPLETION:
+            # Only check when mode is completion
+            dataset_query_variable = config.get("dataset_query_variable")
+
+            if not dataset_query_variable:
+                raise ValueError("Dataset query variable is required when dataset is exist")
+
+        return config, ["agent_mode", "dataset_configs", "dataset_query_variable"]
+
+    @classmethod
+    def extract_dataset_config_for_legacy_compatibility(cls, tenant_id: str, app_mode: AppMode, config: dict) -> dict:
+        """
+        Extract dataset config for legacy compatibility
+
+        :param tenant_id: tenant ID
+        :param app_mode: app mode
+        :param config: app model config args
+        """
+        # Extract dataset config for legacy compatibility
+        if not config.get("agent_mode"):
+            config["agent_mode"] = {
+                "enabled": False,
+                "tools": []
+            }
+
+        if not isinstance(config["agent_mode"], dict):
+            raise ValueError("agent_mode must be of object type")
+
+        # enabled
+        if "enabled" not in config["agent_mode"] or not config["agent_mode"]["enabled"]:
+            config["agent_mode"]["enabled"] = False
+
+        if not isinstance(config["agent_mode"]["enabled"], bool):
+            raise ValueError("enabled in agent_mode must be of boolean type")
+
+        # tools
+        if not config["agent_mode"].get("tools"):
+            config["agent_mode"]["tools"] = []
+
+        if not isinstance(config["agent_mode"]["tools"], list):
+            raise ValueError("tools in agent_mode must be a list of objects")
+
+        # strategy
+        if not config["agent_mode"].get("strategy"):
+            config["agent_mode"]["strategy"] = PlanningStrategy.ROUTER.value
+
+        has_datasets = False
+        if config["agent_mode"]["strategy"] in [PlanningStrategy.ROUTER.value, PlanningStrategy.REACT_ROUTER.value]:
+            for tool in config["agent_mode"]["tools"]:
+                key = list(tool.keys())[0]
+                if key == "dataset":
+                    # old style, use tool name as key
+                    tool_item = tool[key]
+
+                    if "enabled" not in tool_item or not tool_item["enabled"]:
+                        tool_item["enabled"] = False
+
+                    if not isinstance(tool_item["enabled"], bool):
+                        raise ValueError("enabled in agent_mode.tools must be of boolean type")
+
+                    if 'id' not in tool_item:
+                        raise ValueError("id is required in dataset")
+
+                    try:
+                        uuid.UUID(tool_item["id"])
+                    except ValueError:
+                        raise ValueError("id in dataset must be of UUID type")
+
+                    if not cls.is_dataset_exists(tenant_id, tool_item["id"]):
+                        raise ValueError("Dataset ID does not exist, please check your permission.")
+
+                    has_datasets = True
+
+        need_manual_query_datasets = has_datasets and config["agent_mode"]["enabled"]
+
+        if need_manual_query_datasets and app_mode == AppMode.COMPLETION:
+            # Only check when mode is completion
+            dataset_query_variable = config.get("dataset_query_variable")
+
+            if not dataset_query_variable:
+                raise ValueError("Dataset query variable is required when dataset is exist")
+
+        return config
+
+    @classmethod
+    def is_dataset_exists(cls, tenant_id: str, dataset_id: str) -> bool:
+        # verify if the dataset ID exists
+        dataset = DatasetService.get_dataset(dataset_id)
+
+        if not dataset:
+            return False
+
+        if dataset.tenant_id != tenant_id:
+            return False
+
+        return True

api/core/app/validators/external_data_fetch.py

@@ -0,0 +1,39 @@
+
+from core.external_data_tool.factory import ExternalDataToolFactory
+
+
+class ExternalDataFetchValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, tenant_id: str, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for external data fetch feature
+
+        :param tenant_id: workspace id
+        :param config: app model config args
+        """
+        if not config.get("external_data_tools"):
+            config["external_data_tools"] = []
+
+        if not isinstance(config["external_data_tools"], list):
+            raise ValueError("external_data_tools must be of list type")
+
+        for tool in config["external_data_tools"]:
+            if "enabled" not in tool or not tool["enabled"]:
+                tool["enabled"] = False
+
+            if not tool["enabled"]:
+                continue
+
+            if "type" not in tool or not tool["type"]:
+                raise ValueError("external_data_tools[].type is required")
+
+            typ = tool["type"]
+            config = tool["config"]
+
+            ExternalDataToolFactory.validate_config(
+                name=typ,
+                tenant_id=tenant_id,
+                config=config
+            )
+
+        return config, ["external_data_tools"]

api/core/app/validators/file_upload.py

@@ -0,0 +1,37 @@
+
+
+class FileUploadValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for file upload feature
+
+        :param config: app model config args
+        """
+        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]")
+
+            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']")
+
+        return config, ["file_upload"]

api/core/app/validators/model_validator.py

@@ -0,0 +1,82 @@
+
+from core.model_runtime.entities.model_entities import ModelPropertyKey, ModelType
+from core.model_runtime.model_providers import model_provider_factory
+from core.provider_manager import ProviderManager
+
+
+class ModelValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, tenant_id: str, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for model config
+
+        :param tenant_id: tenant id
+        :param config: app model config args
+        """
+        if 'model' not in config:
+            raise ValueError("model is required")
+
+        if not isinstance(config["model"], dict):
+            raise ValueError("model must be of object type")
+
+        # model.provider
+        provider_entities = model_provider_factory.get_providers()
+        model_provider_names = [provider.provider for provider in provider_entities]
+        if 'provider' not in config["model"] or config["model"]["provider"] not in model_provider_names:
+            raise ValueError(f"model.provider is required and must be in {str(model_provider_names)}")
+
+        # model.name
+        if 'name' not in config["model"]:
+            raise ValueError("model.name is required")
+
+        provider_manager = ProviderManager()
+        models = provider_manager.get_configurations(tenant_id).get_models(
+            provider=config["model"]["provider"],
+            model_type=ModelType.LLM
+        )
+
+        if not models:
+            raise ValueError("model.name must be in the specified model list")
+
+        model_ids = [m.model for m in models]
+        if config["model"]["name"] not in model_ids:
+            raise ValueError("model.name must be in the specified model list")
+
+        model_mode = None
+        for model in models:
+            if model.model == config["model"]["name"]:
+                model_mode = model.model_properties.get(ModelPropertyKey.MODE)
+                break
+
+        # model.mode
+        if model_mode:
+            config['model']["mode"] = model_mode
+        else:
+            config['model']["mode"] = "completion"
+
+        # model.completion_params
+        if 'completion_params' not in config["model"]:
+            raise ValueError("model.completion_params is required")
+
+        config["model"]["completion_params"] = cls.validate_model_completion_params(
+            config["model"]["completion_params"]
+        )
+
+        return config, ["model"]
+
+    @classmethod
+    def validate_model_completion_params(cls, cp: dict) -> dict:
+        # model.completion_params
+        if not isinstance(cp, dict):
+            raise ValueError("model.completion_params must be of object type")
+
+        # stop
+        if 'stop' not in cp:
+            cp["stop"] = []
+        elif not isinstance(cp["stop"], list):
+            raise ValueError("stop in model.completion_params must be of list type")
+
+        if len(cp["stop"]) > 4:
+            raise ValueError("stop sequences must be less than 4")
+
+        return cp

api/core/app/validators/moderation.py

@@ -0,0 +1,35 @@
+import logging
+
+from core.moderation.factory import ModerationFactory
+
+logger = logging.getLogger(__name__)
+
+
+class ModerationValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, tenant_id, config: dict) -> tuple[dict, list[str]]:
+        if not config.get("sensitive_word_avoidance"):
+            config["sensitive_word_avoidance"] = {
+                "enabled": False
+            }
+
+        if not isinstance(config["sensitive_word_avoidance"], dict):
+            raise ValueError("sensitive_word_avoidance must be of dict type")
+
+        if "enabled" not in config["sensitive_word_avoidance"] or not config["sensitive_word_avoidance"]["enabled"]:
+            config["sensitive_word_avoidance"]["enabled"] = False
+
+        if config["sensitive_word_avoidance"]["enabled"]:
+            if not config["sensitive_word_avoidance"].get("type"):
+                raise ValueError("sensitive_word_avoidance.type is required")
+
+            typ = config["sensitive_word_avoidance"]["type"]
+            config = config["sensitive_word_avoidance"]["config"]
+
+            ModerationFactory.validate_config(
+                name=typ,
+                tenant_id=tenant_id,
+                config=config
+            )
+
+        return config, ["sensitive_word_avoidance"]

api/core/app/validators/more_like_this.py

@@ -0,0 +1,25 @@
+
+
+class MoreLikeThisValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for more like this feature
+
+        :param config: app model config args
+        """
+        if not config.get("more_like_this"):
+            config["more_like_this"] = {
+                "enabled": False
+            }
+
+        if not isinstance(config["more_like_this"], dict):
+            raise ValueError("more_like_this must be of dict type")
+
+        if "enabled" not in config["more_like_this"] or not config["more_like_this"]["enabled"]:
+            config["more_like_this"]["enabled"] = False
+
+        if not isinstance(config["more_like_this"]["enabled"], bool):
+            raise ValueError("enabled in more_like_this must be of boolean type")
+
+        return config, ["more_like_this"]

api/core/app/validators/opening_statement.py

@@ -0,0 +1,28 @@
+
+
+class OpeningStatementValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for opening statement feature
+
+        :param config: app model config args
+        """
+        if not config.get("opening_statement"):
+            config["opening_statement"] = ""
+
+        if not isinstance(config["opening_statement"], str):
+            raise ValueError("opening_statement must be of string type")
+
+        # suggested_questions
+        if not config.get("suggested_questions"):
+            config["suggested_questions"] = []
+
+        if not isinstance(config["suggested_questions"], list):
+            raise ValueError("suggested_questions must be of list type")
+
+        for question in config["suggested_questions"]:
+            if not isinstance(question, str):
+                raise ValueError("Elements in suggested_questions list must be of string type")
+
+        return config, ["opening_statement", "suggested_questions"]

api/core/app/validators/prompt.py

@@ -0,0 +1,86 @@
+
+from core.entities.application_entities import PromptTemplateEntity
+from core.prompt.simple_prompt_transform import ModelMode
+from models.model import AppMode
+
+
+class PromptValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, app_mode: AppMode, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate pre_prompt and set defaults for prompt feature
+        depending on the config['model']
+
+        :param app_mode: app mode
+        :param config: app model config args
+        """
+        if not config.get("prompt_type"):
+            config["prompt_type"] = PromptTemplateEntity.PromptType.SIMPLE.value
+
+        prompt_type_vals = [typ.value for typ in PromptTemplateEntity.PromptType]
+        if config['prompt_type'] not in prompt_type_vals:
+            raise ValueError(f"prompt_type must be in {prompt_type_vals}")
+
+        # chat_prompt_config
+        if not config.get("chat_prompt_config"):
+            config["chat_prompt_config"] = {}
+
+        if not isinstance(config["chat_prompt_config"], dict):
+            raise ValueError("chat_prompt_config must be of object type")
+
+        # completion_prompt_config
+        if not config.get("completion_prompt_config"):
+            config["completion_prompt_config"] = {}
+
+        if not isinstance(config["completion_prompt_config"], dict):
+            raise ValueError("completion_prompt_config must be of object type")
+
+        if config['prompt_type'] == PromptTemplateEntity.PromptType.ADVANCED.value:
+            if not config['chat_prompt_config'] and not config['completion_prompt_config']:
+                raise ValueError("chat_prompt_config or completion_prompt_config is required "
+                                 "when prompt_type is advanced")
+
+            model_mode_vals = [mode.value for mode in ModelMode]
+            if config['model']["mode"] not in model_mode_vals:
+                raise ValueError(f"model.mode must be in {model_mode_vals} when prompt_type is advanced")
+
+            if app_mode == AppMode.CHAT and config['model']["mode"] == ModelMode.COMPLETION.value:
+                user_prefix = config['completion_prompt_config']['conversation_histories_role']['user_prefix']
+                assistant_prefix = config['completion_prompt_config']['conversation_histories_role']['assistant_prefix']
+
+                if not user_prefix:
+                    config['completion_prompt_config']['conversation_histories_role']['user_prefix'] = 'Human'
+
+                if not assistant_prefix:
+                    config['completion_prompt_config']['conversation_histories_role']['assistant_prefix'] = 'Assistant'
+
+            if config['model']["mode"] == ModelMode.CHAT.value:
+                prompt_list = config['chat_prompt_config']['prompt']
+
+                if len(prompt_list) > 10:
+                    raise ValueError("prompt messages must be less than 10")
+        else:
+            # pre_prompt, for simple mode
+            if not config.get("pre_prompt"):
+                config["pre_prompt"] = ""
+
+            if not isinstance(config["pre_prompt"], str):
+                raise ValueError("pre_prompt must be of string type")
+
+        return config, ["prompt_type", "pre_prompt", "chat_prompt_config", "completion_prompt_config"]
+
+    @classmethod
+    def validate_post_prompt_and_set_defaults(cls, config: dict) -> dict:
+        """
+        Validate post_prompt and set defaults for prompt feature
+
+        :param config: app model config args
+        """
+        # post_prompt
+        if not config.get("post_prompt"):
+            config["post_prompt"] = ""
+
+        if not isinstance(config["post_prompt"], str):
+            raise ValueError("post_prompt must be of string type")
+
+        return config

api/core/app/validators/retriever_resource.py

@@ -0,0 +1,25 @@
+
+
+class RetrieverResourceValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for retriever resource feature
+
+        :param config: app model config args
+        """
+        if not config.get("retriever_resource"):
+            config["retriever_resource"] = {
+                "enabled": False
+            }
+
+        if not isinstance(config["retriever_resource"], dict):
+            raise ValueError("retriever_resource must be of dict type")
+
+        if "enabled" not in config["retriever_resource"] or not config["retriever_resource"]["enabled"]:
+            config["retriever_resource"]["enabled"] = False
+
+        if not isinstance(config["retriever_resource"]["enabled"], bool):
+            raise ValueError("enabled in retriever_resource must be of boolean type")
+
+        return config, ["retriever_resource"]

api/core/app/validators/speech_to_text.py

@@ -0,0 +1,25 @@
+
+
+class SpeechToTextValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for speech to text feature
+
+        :param config: app model config args
+        """
+        if not config.get("speech_to_text"):
+            config["speech_to_text"] = {
+                "enabled": False
+            }
+
+        if not isinstance(config["speech_to_text"], dict):
+            raise ValueError("speech_to_text must be of dict type")
+
+        if "enabled" not in config["speech_to_text"] or not config["speech_to_text"]["enabled"]:
+            config["speech_to_text"]["enabled"] = False
+
+        if not isinstance(config["speech_to_text"]["enabled"], bool):
+            raise ValueError("enabled in speech_to_text must be of boolean type")
+
+        return config, ["speech_to_text"]

api/core/app/validators/suggested_questions.py

@@ -0,0 +1,25 @@
+
+
+class SuggestedQuestionsValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for suggested questions feature
+
+        :param config: app model config args
+        """
+        if not config.get("suggested_questions_after_answer"):
+            config["suggested_questions_after_answer"] = {
+                "enabled": False
+            }
+
+        if not isinstance(config["suggested_questions_after_answer"], dict):
+            raise ValueError("suggested_questions_after_answer must be of dict type")
+
+        if "enabled" not in config["suggested_questions_after_answer"] or not config["suggested_questions_after_answer"]["enabled"]:
+            config["suggested_questions_after_answer"]["enabled"] = False
+
+        if not isinstance(config["suggested_questions_after_answer"]["enabled"], bool):
+            raise ValueError("enabled in suggested_questions_after_answer must be of boolean type")
+
+        return config, ["suggested_questions_after_answer"]

api/core/app/validators/text_to_speech.py

@@ -0,0 +1,29 @@
+
+
+class TextToSpeechValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for text to speech feature
+
+        :param config: app model config args
+        """
+        if not config.get("text_to_speech"):
+            config["text_to_speech"] = {
+                "enabled": False,
+                "voice": "",
+                "language": ""
+            }
+
+        if not isinstance(config["text_to_speech"], dict):
+            raise ValueError("text_to_speech must be of dict type")
+
+        if "enabled" not in config["text_to_speech"] or not config["text_to_speech"]["enabled"]:
+            config["text_to_speech"]["enabled"] = False
+            config["text_to_speech"]["voice"] = ""
+            config["text_to_speech"]["language"] = ""
+
+        if not isinstance(config["text_to_speech"]["enabled"], bool):
+            raise ValueError("enabled in text_to_speech must be of boolean type")
+
+        return config, ["text_to_speech"]

api/core/app/validators/user_input_form.py

@@ -0,0 +1,61 @@
+import re
+
+
+class UserInputFormValidator:
+    @classmethod
+    def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:
+        """
+        Validate and set defaults for user input form
+
+        :param config: app model config args
+        """
+        if not config.get("user_input_form"):
+            config["user_input_form"] = []
+
+        if not isinstance(config["user_input_form"], list):
+            raise ValueError("user_input_form must be a list of objects")
+
+        variables = []
+        for item in config["user_input_form"]:
+            key = list(item.keys())[0]
+            if key not in ["text-input", "select", "paragraph", "number", "external_data_tool"]:
+                raise ValueError("Keys in user_input_form list can only be 'text-input', 'paragraph'  or 'select'")
+
+            form_item = item[key]
+            if 'label' not in form_item:
+                raise ValueError("label is required in user_input_form")
+
+            if not isinstance(form_item["label"], str):
+                raise ValueError("label in user_input_form must be of string type")
+
+            if 'variable' not in form_item:
+                raise ValueError("variable is required in user_input_form")
+
+            if not isinstance(form_item["variable"], str):
+                raise ValueError("variable in user_input_form must be of string type")
+
+            pattern = re.compile(r"^(?!\d)[\u4e00-\u9fa5A-Za-z0-9_\U0001F300-\U0001F64F\U0001F680-\U0001F6FF]{1,100}$")
+            if pattern.match(form_item["variable"]) is None:
+                raise ValueError("variable in user_input_form must be a string, "
+                                 "and cannot start with a number")
+
+            variables.append(form_item["variable"])
+
+            if 'required' not in form_item or not form_item["required"]:
+                form_item["required"] = False
+
+            if not isinstance(form_item["required"], bool):
+                raise ValueError("required in user_input_form must be of boolean type")
+
+            if key == "select":
+                if 'options' not in form_item or not form_item["options"]:
+                    form_item["options"] = []
+
+                if not isinstance(form_item["options"], list):
+                    raise ValueError("options in user_input_form must be a list of strings")
+
+                if "default" in form_item and form_item['default'] \
+                        and form_item["default"] not in form_item["options"]:
+                    raise ValueError("default value in user_input_form must be in the options list")
+
+        return config, ["user_input_form"]