chore: bump version to 1.1.0 (#16128)

Signed-off-by: -LAN- <laipz8200@outlook.com>

.github/ISSUE_TEMPLATE/tracker.yml

@@ -0,0 +1,13 @@
+name: "👾 Tracker"
+description: For inner usages, please donot use this template.
+title: "[Tracker] "
+labels:
+  - tracker
+body:
+  - type: textarea
+    id: content
+    attributes:
+      label: Blockers
+      placeholder: "- [ ] ..."
+    validations:
+      required: true

.github/workflows/build-push.yml

@@ -5,6 +5,7 @@ on:
     branches:
       - "main"
       - "deploy/dev"
+      - "deploy/enterprise"
   release:
     types: [published]
 

.github/workflows/deploy-enterprise.yml

@@ -0,0 +1,29 @@
+name: Deploy Enterprise
+
+permissions:
+  contents: read
+
+on:
+  workflow_run:
+    workflows: ["Build and Push API & Web"]
+    branches:
+      - "deploy/enterprise"
+    types:
+      - completed
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+    if: |
+      github.event.workflow_run.conclusion == 'success' &&
+      github.event.workflow_run.head_branch == 'deploy/enterprise'
+
+    steps:
+      - name: Deploy to server
+        uses: appleboy/ssh-action@v0.1.8
+        with:
+          host: ${{ secrets.ENTERPRISE_SSH_HOST }}
+          username: ${{ secrets.ENTERPRISE_SSH_USER }}
+          password: ${{ secrets.ENTERPRISE_SSH_PASSWORD }}
+          script: |
+            ${{ vars.ENTERPRISE_SSH_SCRIPT || secrets.ENTERPRISE_SSH_SCRIPT }}

.github/workflows/expose_service_ports.sh

@@ -10,5 +10,6 @@ yq eval '.services["elasticsearch"].ports += ["9200:9200"]' -i docker/docker-com
 yq eval '.services.couchbase-server.ports += ["8091-8096:8091-8096"]' -i docker/docker-compose.yaml
 yq eval '.services.couchbase-server.ports += ["11210:11210"]' -i docker/docker-compose.yaml
 yq eval '.services.tidb.ports += ["4000:4000"]' -i docker/tidb/docker-compose.yaml
+yq eval '.services.opengauss.ports += ["6600:6600"]' -i docker/docker-compose.yaml
 
-echo "Ports exposed for sandbox, weaviate, tidb, qdrant, chroma, milvus, pgvector, pgvecto-rs, elasticsearch, couchbase"
+echo "Ports exposed for sandbox, weaviate, tidb, qdrant, chroma, milvus, pgvector, pgvecto-rs, elasticsearch, couchbase, opengauss"

.github/workflows/vdb-tests.yml

@@ -76,6 +76,7 @@ jobs:
             milvus-standalone
             pgvecto-rs
             pgvector
+            opengauss
             chroma
             elasticsearch
 

.gitignore

@@ -202,3 +202,6 @@ api/.vscode
 
 # plugin migrate
 plugins.jsonl
+
+# mise
+mise.toml

CONTRIBUTING_CN.md

@@ -26,7 +26,7 @@
   | [@jyong](https://github.com/JohnJyong)                       | RAG 流水线设计                                  |
   | [@GarfieldDai](https://github.com/GarfieldDai)               | 构建 workflow 编排                     |
   | [@iamjoel](https://github.com/iamjoel) & [@zxhlyh](https://github.com/zxhlyh) | 让我们的前端更易用                  |
-  | [@guchenhe](https://github.com/guchenhe) & [@crazywoola](https://github.com/crazywoola) | 开发人员体验, 综合事项联系人 |
+  | [@guchenhe](https://github.com/guchenhe) & [@crazywoola](https://github.com/crazywoola) | 开发人员体验，综合事项联系人 |
   | [@takatost](https://github.com/takatost)                     | 产品整体方向和架构           |
 
   事项优先级：
@@ -47,7 +47,7 @@
   | ------------------------------------------------------------ | --------------- |
   | 核心功能的 Bugs（例如无法登录、应用无法工作、安全漏洞） | 紧急        |
   | 非紧急 bugs, 性能提升                        | 中等优先级 |
-  | 小幅修复(错别字, 能正常工作但存在误导的 UI)                | 低优先级    |
+  | 小幅修复 (错别字，能正常工作但存在误导的 UI)                | 低优先级    |
 
 ## 安装
 

README_CN.md

@@ -79,7 +79,7 @@ Dify 是一个开源的 LLM 应用开发平台。其直观的界面结合了 AI
   广泛的 RAG 功能，涵盖从文档摄入到检索的所有内容，支持从 PDF、PPT 和其他常见文档格式中提取文本的开箱即用的支持。
 
 **5. Agent 智能体**: 
-  您可以基于 LLM 函数调用或 ReAct 定义 Agent，并为 Agent 添加预构建或自定义工具。Dify 为 AI Agent 提供了50多种内置工具，如谷歌搜索、DALL·E、Stable Diffusion 和 WolframAlpha 等。
+  您可以基于 LLM 函数调用或 ReAct 定义 Agent，并为 Agent 添加预构建或自定义工具。Dify 为 AI Agent 提供了 50 多种内置工具，如谷歌搜索、DALL·E、Stable Diffusion 和 WolframAlpha 等。
 
 **6. LLMOps**: 
   随时间监视和分析应用程序日志和性能。您可以根据生产数据和标注持续改进提示、数据集和模型。
@@ -112,7 +112,7 @@ Dify 是一个开源的 LLM 应用开发平台。其直观的界面结合了 AI
     <td align="center">仅限 OpenAI</td>
   </tr>
   <tr>
-    <td align="center">RAG引擎</td>
+    <td align="center">RAG 引擎</td>
     <td align="center">✅</td>
     <td align="center">✅</td>
     <td align="center">✅</td>
@@ -234,7 +234,7 @@ docker compose up -d
 对于那些想要贡献代码的人，请参阅我们的[贡献指南](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md)。
 同时，请考虑通过社交媒体、活动和会议来支持 Dify 的分享。
 
-> 我们正在寻找贡献者来帮助将Dify翻译成除了中文和英文之外的其他语言。如果您有兴趣帮助，请参阅我们的[i18n README](https://github.com/langgenius/dify/blob/main/web/i18n/README.md)获取更多信息，并在我们的[Discord社区服务器](https://discord.gg/8Tpq4AcN9c)的`global-users`频道中留言。
+> 我们正在寻找贡献者来帮助将 Dify 翻译成除了中文和英文之外的其他语言。如果您有兴趣帮助，请参阅我们的[i18n README](https://github.com/langgenius/dify/blob/main/web/i18n/README.md)获取更多信息，并在我们的[Discord 社区服务器](https://discord.gg/8Tpq4AcN9c)的`global-users`频道中留言。
 
 **Contributors**
 

api/.env.example

@@ -137,7 +137,7 @@ WEB_API_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
 CONSOLE_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
 
 # Vector database configuration
-# support: weaviate, qdrant, milvus, myscale, relyt, pgvecto_rs, pgvector, pgvector, chroma, opensearch, tidb_vector, couchbase, vikingdb, upstash, lindorm, oceanbase
+# support: weaviate, qdrant, milvus, myscale, relyt, pgvecto_rs, pgvector, pgvector, chroma, opensearch, tidb_vector, couchbase, vikingdb, upstash, lindorm, oceanbase, opengauss
 VECTOR_STORE=weaviate
 
 # Weaviate configuration
@@ -298,6 +298,14 @@ OCEANBASE_VECTOR_PASSWORD=difyai123456
 OCEANBASE_VECTOR_DATABASE=test
 OCEANBASE_MEMORY_LIMIT=6G
 
+# openGauss configuration
+OPENGAUSS_HOST=127.0.0.1
+OPENGAUSS_PORT=6600
+OPENGAUSS_USER=postgres
+OPENGAUSS_PASSWORD=Dify@123
+OPENGAUSS_DATABASE=dify
+OPENGAUSS_MIN_CONNECTION=1
+OPENGAUSS_MAX_CONNECTION=5
 
 # Upload configuration
 UPLOAD_FILE_SIZE_LIMIT=15
@@ -378,6 +386,7 @@ HTTP_REQUEST_MAX_READ_TIMEOUT=600
 HTTP_REQUEST_MAX_WRITE_TIMEOUT=600
 HTTP_REQUEST_NODE_MAX_BINARY_SIZE=10485760
 HTTP_REQUEST_NODE_MAX_TEXT_SIZE=1048576
+HTTP_REQUEST_NODE_SSL_VERIFY=True
 
 # Respect X-* headers to redirect clients
 RESPECT_XFORWARD_HEADERS_ENABLED=false
@@ -444,4 +453,4 @@ CREATE_TIDB_SERVICE_JOB_ENABLED=false
 # Maximum number of submitted thread count in a ThreadPool for parallel node execution
 MAX_SUBMIT_COUNT=100
 # Lockout duration in seconds
-LOGIN_LOCKOUT_DURATION=86400
+LOGIN_LOCKOUT_DURATION=86400

api/Dockerfile

@@ -56,8 +56,6 @@ RUN \
         curl nodejs libgmp-dev libmpfr-dev libmpc-dev \
         # For Security
         expat libldap-2.5-0 perl libsqlite3-0 zlib1g \
-        # install a chinese font to support the use of tools like matplotlib
-        fonts-noto-cjk \
         # install a package to improve the accuracy of guessing mime type and file extension
         media-types \
         # install libmagic to support the use of python-magic guess MIMETYPE

api/commands.py

@@ -160,11 +160,17 @@ def migrate_annotation_vector_database():
     while True:
         try:
             # get apps info
+            per_page = 50
             apps = (
-                App.query.filter(App.status == "normal")
+                db.session.query(App)
+                .filter(App.status == "normal")
                 .order_by(App.created_at.desc())
-                .paginate(page=page, per_page=50)
+                .limit(per_page)
+                .offset((page - 1) * per_page)
+                .all()
             )
+            if not apps:
+                break
         except NotFound:
             break
 
@@ -267,6 +273,7 @@ def migrate_knowledge_vector_database():
         VectorType.WEAVIATE,
         VectorType.ORACLE,
         VectorType.ELASTICSEARCH,
+        VectorType.OPENGAUSS,
     }
     lower_collection_vector_types = {
         VectorType.ANALYTICDB,

api/configs/feature/__init__.py

@@ -332,6 +332,11 @@ class HttpConfig(BaseSettings):
         default=1 * 1024 * 1024,
     )
 
+    HTTP_REQUEST_NODE_SSL_VERIFY: bool = Field(
+        description="Enable or disable SSL verification for HTTP requests",
+        default=True,
+    )
+
     SSRF_DEFAULT_MAX_RETRIES: PositiveInt = Field(
         description="Maximum number of retries for network requests (SSRF)",
         default=3,

api/configs/middleware/__init__.py

@@ -26,6 +26,7 @@ from .vdb.lindorm_config import LindormConfig
 from .vdb.milvus_config import MilvusConfig
 from .vdb.myscale_config import MyScaleConfig
 from .vdb.oceanbase_config import OceanBaseVectorConfig
+from .vdb.opengauss_config import OpenGaussConfig
 from .vdb.opensearch_config import OpenSearchConfig
 from .vdb.oracle_config import OracleConfig
 from .vdb.pgvector_config import PGVectorConfig
@@ -281,5 +282,6 @@ class MiddlewareConfig(
     LindormConfig,
     OceanBaseVectorConfig,
     BaiduVectorDBConfig,
+    OpenGaussConfig,
 ):
     pass

api/configs/middleware/vdb/opengauss_config.py

@@ -0,0 +1,45 @@
+from typing import Optional
+
+from pydantic import Field, PositiveInt
+from pydantic_settings import BaseSettings
+
+
+class OpenGaussConfig(BaseSettings):
+    """
+    Configuration settings for OpenGauss
+    """
+
+    OPENGAUSS_HOST: Optional[str] = Field(
+        description="Hostname or IP address of the OpenGauss server(e.g., 'localhost')",
+        default=None,
+    )
+
+    OPENGAUSS_PORT: PositiveInt = Field(
+        description="Port number on which the OpenGauss server is listening (default is 6600)",
+        default=6600,
+    )
+
+    OPENGAUSS_USER: Optional[str] = Field(
+        description="Username for authenticating with the OpenGauss database",
+        default=None,
+    )
+
+    OPENGAUSS_PASSWORD: Optional[str] = Field(
+        description="Password for authenticating with the OpenGauss database",
+        default=None,
+    )
+
+    OPENGAUSS_DATABASE: Optional[str] = Field(
+        description="Name of the OpenGauss database to connect to",
+        default=None,
+    )
+
+    OPENGAUSS_MIN_CONNECTION: PositiveInt = Field(
+        description="Min connection of the OpenGauss database",
+        default=1,
+    )
+
+    OPENGAUSS_MAX_CONNECTION: PositiveInt = Field(
+        description="Max connection of the OpenGauss database",
+        default=5,
+    )

api/configs/middleware/vdb/pgvector_config.py

@@ -43,3 +43,8 @@ class PGVectorConfig(BaseSettings):
         description="Max connection of the PostgreSQL database",
         default=5,
     )
+
+    PGVECTOR_PG_BIGM: bool = Field(
+        description="Whether to use pg_bigm module for full text search",
+        default=False,
+    )

api/configs/packaging/__init__.py

@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="1.0.0",
+        default="1.1.0",
     )
 
     COMMIT_SHA: str = Field(

api/controllers/console/__init__.py

@@ -81,6 +81,7 @@ from .datasets import (
     datasets_segments,
     external,
     hit_testing,
+    metadata,
     website,
 )
 

api/controllers/console/app/app.py

@@ -316,7 +316,7 @@ class AppTraceApi(Resource):
     @account_initialization_required
     def post(self, app_id):
         # add app trace
-        if not current_user.is_admin_or_owner:
+        if not current_user.is_editor:
             raise Forbidden()
         parser = reqparse.RequestParser()
         parser.add_argument("enabled", type=bool, required=True, location="json")

api/controllers/console/app/generator.py

@@ -1,5 +1,4 @@
 import os
-from typing import cast
 
 from flask_login import current_user  # type: ignore
 from flask_restful import Resource, reqparse  # type: ignore
@@ -12,11 +11,8 @@ from controllers.console.app.error import (
     ProviderQuotaExceededError,
 )
 from controllers.console.wraps import account_initialization_required, setup_required
-from core.auto.workflow_generator.workflow_generator import WorkflowGenerator
 from core.errors.error import ModelCurrentlyNotSupportError, ProviderTokenNotInitError, QuotaExceededError
 from core.llm_generator.llm_generator import LLMGenerator
-from core.model_manager import ModelManager
-from core.model_runtime.entities.model_entities import ModelType
 from core.model_runtime.errors.invoke import InvokeError
 from libs.login import login_required
 
@@ -89,45 +85,5 @@ class RuleCodeGenerateApi(Resource):
         return code_result
 
 
-class AutoGenerateWorkflowApi(Resource):
-    @setup_required
-    @login_required
-    @account_initialization_required
-    def post(self):
-        """
-        Auto generate workflow
-        """
-
-        parser = reqparse.RequestParser()
-        parser.add_argument("instruction", type=str, required=True, location="json")
-        parser.add_argument("model_config", type=dict, required=True, location="json")
-        tenant_id = cast(str, current_user.current_tenant_id)
-        args = parser.parse_args()
-        instruction = args.get("instruction")
-        if not instruction:
-            raise ValueError("Instruction is required")
-        if not args.get("model_config"):
-            raise ValueError("Model config is required")
-        model_config = cast(dict, args.get("model_config"))
-        model_manager = ModelManager()
-        model_instance = model_manager.get_model_instance(
-            tenant_id=tenant_id,
-            model_type=ModelType.LLM,
-            provider=model_config.get("provider", ""),
-            model=model_config.get("name", ""),
-        )
-        workflow_generator = WorkflowGenerator(
-            model_instance=model_instance,
-        )
-        workflow_yaml = workflow_generator.generate_workflow(
-            user_requirement=instruction,
-        )
-        return workflow_yaml
-
-
 api.add_resource(RuleGenerateApi, "/rule-generate")
 api.add_resource(RuleCodeGenerateApi, "/rule-code-generate")
-api.add_resource(
-    AutoGenerateWorkflowApi,
-    "/auto-generate",
-)

api/controllers/console/datasets/data_source.py

@@ -122,7 +122,7 @@ class DataSourceNotionListApi(Resource):
                 if dataset.data_source_type != "notion_import":
                     raise ValueError("Dataset is not notion type.")
 
-                documents = session.execute(
+                documents = session.scalars(
                     select(Document).filter_by(
                         dataset_id=dataset_id,
                         tenant_id=current_user.current_tenant_id,

api/controllers/console/datasets/datasets.py

@@ -10,7 +10,12 @@ from controllers.console import api
 from controllers.console.apikey import api_key_fields, api_key_list
 from controllers.console.app.error import ProviderNotInitializeError
 from controllers.console.datasets.error import DatasetInUseError, DatasetNameDuplicateError, IndexingEstimateError
-from controllers.console.wraps import account_initialization_required, enterprise_license_required, setup_required
+from controllers.console.wraps import (
+    account_initialization_required,
+    cloud_edition_billing_rate_limit_check,
+    enterprise_license_required,
+    setup_required,
+)
 from core.errors.error import LLMBadRequestError, ProviderTokenNotInitError
 from core.indexing_runner import IndexingRunner
 from core.model_runtime.entities.model_entities import ModelType
@@ -96,6 +101,7 @@ class DatasetListApi(Resource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self):
         parser = reqparse.RequestParser()
         parser.add_argument(
@@ -178,6 +184,10 @@ class DatasetApi(Resource):
         except services.errors.account.NoPermissionError as e:
             raise Forbidden(str(e))
         data = marshal(dataset, dataset_detail_fields)
+        if dataset.indexing_technique == "high_quality":
+            if dataset.embedding_model_provider:
+                provider_id = ModelProviderID(dataset.embedding_model_provider)
+                data["embedding_model_provider"] = str(provider_id)
         if data.get("permission") == "partial_members":
             part_users_list = DatasetPermissionService.get_dataset_partial_member_list(dataset_id_str)
             data.update({"partial_member_list": part_users_list})
@@ -210,6 +220,7 @@ class DatasetApi(Resource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id):
         dataset_id_str = str(dataset_id)
         dataset = DatasetService.get_dataset(dataset_id_str)
@@ -276,7 +287,11 @@ class DatasetApi(Resource):
         data = request.get_json()
 
         # check embedding model setting
-        if data.get("indexing_technique") == "high_quality":
+        if (
+            data.get("indexing_technique") == "high_quality"
+            and data.get("embedding_model_provider") is not None
+            and data.get("embedding_model") is not None
+        ):
             DatasetService.check_embedding_model_setting(
                 dataset.tenant_id, data.get("embedding_model_provider"), data.get("embedding_model")
             )
@@ -313,6 +328,7 @@ class DatasetApi(Resource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def delete(self, dataset_id):
         dataset_id_str = str(dataset_id)
 
@@ -647,6 +663,7 @@ class DatasetRetrievalSettingApi(Resource):
                 | VectorType.LINDORM
                 | VectorType.COUCHBASE
                 | VectorType.MILVUS
+                | VectorType.OPENGAUSS
             ):
                 return {
                     "retrieval_method": [
@@ -690,6 +707,7 @@ class DatasetRetrievalSettingMockApi(Resource):
                 | VectorType.COUCHBASE
                 | VectorType.PGVECTOR
                 | VectorType.LINDORM
+                | VectorType.OPENGAUSS
             ):
                 return {
                     "retrieval_method": [

api/controllers/console/datasets/datasets_document.py

@@ -26,6 +26,7 @@ from controllers.console.datasets.error import (
 )
 from controllers.console.wraps import (
     account_initialization_required,
+    cloud_edition_billing_rate_limit_check,
     cloud_edition_billing_resource_check,
     setup_required,
 )
@@ -242,6 +243,7 @@ class DatasetDocumentListApi(Resource):
     @account_initialization_required
     @marshal_with(documents_and_batch_fields)
     @cloud_edition_billing_resource_check("vector_space")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self, dataset_id):
         dataset_id = str(dataset_id)
 
@@ -297,6 +299,7 @@ class DatasetDocumentListApi(Resource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def delete(self, dataset_id):
         dataset_id = str(dataset_id)
         dataset = DatasetService.get_dataset(dataset_id)
@@ -320,9 +323,10 @@ class DatasetInitApi(Resource):
     @account_initialization_required
     @marshal_with(dataset_and_document_fields)
     @cloud_edition_billing_resource_check("vector_space")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self):
-        # The role of the current user in the ta table must be admin, owner, or editor
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
 
         parser = reqparse.RequestParser()
@@ -617,7 +621,7 @@ class DocumentDetailApi(DocumentResource):
             raise InvalidMetadataError(f"Invalid metadata value: {metadata}")
 
         if metadata == "only":
-            response = {"id": document.id, "doc_type": document.doc_type, "doc_metadata": document.doc_metadata}
+            response = {"id": document.id, "doc_type": document.doc_type, "doc_metadata": document.doc_metadata_details}
         elif metadata == "without":
             dataset_process_rules = DatasetService.get_process_rules(dataset_id)
             document_process_rules = document.dataset_process_rule.to_dict()
@@ -678,7 +682,7 @@ class DocumentDetailApi(DocumentResource):
                 "disabled_by": document.disabled_by,
                 "archived": document.archived,
                 "doc_type": document.doc_type,
-                "doc_metadata": document.doc_metadata,
+                "doc_metadata": document.doc_metadata_details,
                 "segment_count": document.segment_count,
                 "average_segment_length": document.average_segment_length,
                 "hit_count": document.hit_count,
@@ -694,13 +698,14 @@ class DocumentProcessingApi(DocumentResource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, document_id, action):
         dataset_id = str(dataset_id)
         document_id = str(document_id)
         document = self.get_document(dataset_id, document_id)
 
-        # The role of the current user in the ta table must be admin, owner, or editor
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
 
         if action == "pause":
@@ -730,6 +735,7 @@ class DocumentDeleteApi(DocumentResource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def delete(self, dataset_id, document_id):
         dataset_id = str(dataset_id)
         document_id = str(document_id)
@@ -763,8 +769,8 @@ class DocumentMetadataApi(DocumentResource):
         doc_type = req_data.get("doc_type")
         doc_metadata = req_data.get("doc_metadata")
 
-        # The role of the current user in the ta table must be admin, owner, or editor
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
 
         if doc_type is None or doc_metadata is None:
@@ -798,6 +804,7 @@ class DocumentStatusApi(DocumentResource):
     @login_required
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, action):
         dataset_id = str(dataset_id)
         dataset = DatasetService.get_dataset(dataset_id)
@@ -893,6 +900,7 @@ class DocumentPauseApi(DocumentResource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, document_id):
         """pause document."""
         dataset_id = str(dataset_id)
@@ -925,6 +933,7 @@ class DocumentRecoverApi(DocumentResource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, document_id):
         """recover document."""
         dataset_id = str(dataset_id)
@@ -954,6 +963,7 @@ class DocumentRetryApi(DocumentResource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self, dataset_id):
         """retry document."""
 

api/controllers/console/datasets/datasets_segments.py

@@ -19,6 +19,7 @@ from controllers.console.datasets.error import (
 from controllers.console.wraps import (
     account_initialization_required,
     cloud_edition_billing_knowledge_limit_check,
+    cloud_edition_billing_rate_limit_check,
     cloud_edition_billing_resource_check,
     setup_required,
 )
@@ -106,6 +107,7 @@ class DatasetDocumentSegmentListApi(Resource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def delete(self, dataset_id, document_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -121,8 +123,8 @@ class DatasetDocumentSegmentListApi(Resource):
             raise NotFound("Document not found.")
         segment_ids = request.args.getlist("segment_id")
 
-        # The role of the current user in the ta table must be admin or owner
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         try:
             DatasetService.check_dataset_permission(dataset, current_user)
@@ -137,6 +139,7 @@ class DatasetDocumentSegmentApi(Resource):
     @login_required
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, document_id, action):
         dataset_id = str(dataset_id)
         dataset = DatasetService.get_dataset(dataset_id)
@@ -148,8 +151,8 @@ class DatasetDocumentSegmentApi(Resource):
             raise NotFound("Document not found.")
         # check user's model setting
         DatasetService.check_dataset_model_setting(dataset)
-        # The role of the current user in the ta table must be admin, owner, or editor
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
 
         try:
@@ -191,6 +194,7 @@ class DatasetDocumentSegmentAddApi(Resource):
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
     @cloud_edition_billing_knowledge_limit_check("add_segment")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self, dataset_id, document_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -202,7 +206,7 @@ class DatasetDocumentSegmentAddApi(Resource):
         document = DocumentService.get_document(dataset_id, document_id)
         if not document:
             raise NotFound("Document not found.")
-        if not current_user.is_editor:
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         # check embedding model setting
         if dataset.indexing_technique == "high_quality":
@@ -240,6 +244,7 @@ class DatasetDocumentSegmentUpdateApi(Resource):
     @login_required
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, document_id, segment_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -276,8 +281,8 @@ class DatasetDocumentSegmentUpdateApi(Resource):
         ).first()
         if not segment:
             raise NotFound("Segment not found.")
-        # The role of the current user in the ta table must be admin, owner, or editor
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         try:
             DatasetService.check_dataset_permission(dataset, current_user)
@@ -299,6 +304,7 @@ class DatasetDocumentSegmentUpdateApi(Resource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def delete(self, dataset_id, document_id, segment_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -319,8 +325,8 @@ class DatasetDocumentSegmentUpdateApi(Resource):
         ).first()
         if not segment:
             raise NotFound("Segment not found.")
-        # The role of the current user in the ta table must be admin or owner
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         try:
             DatasetService.check_dataset_permission(dataset, current_user)
@@ -336,6 +342,7 @@ class DatasetDocumentSegmentBatchImportApi(Resource):
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
     @cloud_edition_billing_knowledge_limit_check("add_segment")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self, dataset_id, document_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -402,6 +409,7 @@ class ChildChunkAddApi(Resource):
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
     @cloud_edition_billing_knowledge_limit_check("add_segment")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self, dataset_id, document_id, segment_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -420,7 +428,7 @@ class ChildChunkAddApi(Resource):
         ).first()
         if not segment:
             raise NotFound("Segment not found.")
-        if not current_user.is_editor:
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         # check embedding model setting
         if dataset.indexing_technique == "high_quality":
@@ -499,6 +507,7 @@ class ChildChunkAddApi(Resource):
     @login_required
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, document_id, segment_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -519,8 +528,8 @@ class ChildChunkAddApi(Resource):
         ).first()
         if not segment:
             raise NotFound("Segment not found.")
-        # The role of the current user in the ta table must be admin, owner, or editor
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         try:
             DatasetService.check_dataset_permission(dataset, current_user)
@@ -542,6 +551,7 @@ class ChildChunkUpdateApi(Resource):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def delete(self, dataset_id, document_id, segment_id, child_chunk_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -569,8 +579,8 @@ class ChildChunkUpdateApi(Resource):
         ).first()
         if not child_chunk:
             raise NotFound("Child chunk not found.")
-        # The role of the current user in the ta table must be admin or owner
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         try:
             DatasetService.check_dataset_permission(dataset, current_user)
@@ -586,6 +596,7 @@ class ChildChunkUpdateApi(Resource):
     @login_required
     @account_initialization_required
     @cloud_edition_billing_resource_check("vector_space")
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def patch(self, dataset_id, document_id, segment_id, child_chunk_id):
         # check dataset
         dataset_id = str(dataset_id)
@@ -613,8 +624,8 @@ class ChildChunkUpdateApi(Resource):
         ).first()
         if not child_chunk:
             raise NotFound("Child chunk not found.")
-        # The role of the current user in the ta table must be admin or owner
-        if not current_user.is_editor:
+        # The role of the current user in the ta table must be admin, owner, dataset_operator, or editor
+        if not current_user.is_dataset_editor:
             raise Forbidden()
         try:
             DatasetService.check_dataset_permission(dataset, current_user)

api/controllers/console/datasets/hit_testing.py

@@ -2,7 +2,11 @@ from flask_restful import Resource  # type: ignore
 
 from controllers.console import api
 from controllers.console.datasets.hit_testing_base import DatasetsHitTestingBase
-from controllers.console.wraps import account_initialization_required, setup_required
+from controllers.console.wraps import (
+    account_initialization_required,
+    cloud_edition_billing_rate_limit_check,
+    setup_required,
+)
 from libs.login import login_required
 
 
@@ -10,6 +14,7 @@ class HitTestingApi(Resource, DatasetsHitTestingBase):
     @setup_required
     @login_required
     @account_initialization_required
+    @cloud_edition_billing_rate_limit_check("knowledge")
     def post(self, dataset_id):
         dataset_id_str = str(dataset_id)
 

api/controllers/console/datasets/metadata.py

@@ -0,0 +1,155 @@
+from flask_login import current_user  # type: ignore  # type: ignore
+from flask_restful import Resource, marshal_with, reqparse  # type: ignore
+from werkzeug.exceptions import NotFound
+
+from controllers.console import api
+from controllers.console.wraps import account_initialization_required, enterprise_license_required, setup_required
+from fields.dataset_fields import dataset_metadata_fields
+from libs.login import login_required
+from services.dataset_service import DatasetService
+from services.entities.knowledge_entities.knowledge_entities import (
+    MetadataArgs,
+    MetadataOperationData,
+)
+from services.metadata_service import MetadataService
+
+
+def _validate_name(name):
+    if not name or len(name) < 1 or len(name) > 40:
+        raise ValueError("Name must be between 1 to 40 characters.")
+    return name
+
+
+def _validate_description_length(description):
+    if len(description) > 400:
+        raise ValueError("Description cannot exceed 400 characters.")
+    return description
+
+
+class DatasetMetadataCreateApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    @enterprise_license_required
+    @marshal_with(dataset_metadata_fields)
+    def post(self, dataset_id):
+        parser = reqparse.RequestParser()
+        parser.add_argument("type", type=str, required=True, nullable=True, location="json")
+        parser.add_argument("name", type=str, required=True, nullable=True, location="json")
+        args = parser.parse_args()
+        metadata_args = MetadataArgs(**args)
+
+        dataset_id_str = str(dataset_id)
+        dataset = DatasetService.get_dataset(dataset_id_str)
+        if dataset is None:
+            raise NotFound("Dataset not found.")
+        DatasetService.check_dataset_permission(dataset, current_user)
+
+        metadata = MetadataService.create_metadata(dataset_id_str, metadata_args)
+        return metadata, 201
+
+    @setup_required
+    @login_required
+    @account_initialization_required
+    @enterprise_license_required
+    def get(self, dataset_id):
+        dataset_id_str = str(dataset_id)
+        dataset = DatasetService.get_dataset(dataset_id_str)
+        if dataset is None:
+            raise NotFound("Dataset not found.")
+        return MetadataService.get_dataset_metadatas(dataset), 200
+
+
+class DatasetMetadataApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    @enterprise_license_required
+    @marshal_with(dataset_metadata_fields)
+    def patch(self, dataset_id, metadata_id):
+        parser = reqparse.RequestParser()
+        parser.add_argument("name", type=str, required=True, nullable=True, location="json")
+        args = parser.parse_args()
+
+        dataset_id_str = str(dataset_id)
+        metadata_id_str = str(metadata_id)
+        dataset = DatasetService.get_dataset(dataset_id_str)
+        if dataset is None:
+            raise NotFound("Dataset not found.")
+        DatasetService.check_dataset_permission(dataset, current_user)
+
+        metadata = MetadataService.update_metadata_name(dataset_id_str, metadata_id_str, args.get("name"))
+        return metadata, 200
+
+    @setup_required
+    @login_required
+    @account_initialization_required
+    @enterprise_license_required
+    def delete(self, dataset_id, metadata_id):
+        dataset_id_str = str(dataset_id)
+        metadata_id_str = str(metadata_id)
+        dataset = DatasetService.get_dataset(dataset_id_str)
+        if dataset is None:
+            raise NotFound("Dataset not found.")
+        DatasetService.check_dataset_permission(dataset, current_user)
+
+        MetadataService.delete_metadata(dataset_id_str, metadata_id_str)
+        return 200
+
+
+class DatasetMetadataBuiltInFieldApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    @enterprise_license_required
+    def get(self):
+        built_in_fields = MetadataService.get_built_in_fields()
+        return {"fields": built_in_fields}, 200
+
+
+class DatasetMetadataBuiltInFieldActionApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    @enterprise_license_required
+    def post(self, dataset_id, action):
+        dataset_id_str = str(dataset_id)
+        dataset = DatasetService.get_dataset(dataset_id_str)
+        if dataset is None:
+            raise NotFound("Dataset not found.")
+        DatasetService.check_dataset_permission(dataset, current_user)
+
+        if action == "enable":
+            MetadataService.enable_built_in_field(dataset)
+        elif action == "disable":
+            MetadataService.disable_built_in_field(dataset)
+        return 200
+
+
+class DocumentMetadataEditApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    @enterprise_license_required
+    def post(self, dataset_id):
+        dataset_id_str = str(dataset_id)
+        dataset = DatasetService.get_dataset(dataset_id_str)
+        if dataset is None:
+            raise NotFound("Dataset not found.")
+        DatasetService.check_dataset_permission(dataset, current_user)
+
+        parser = reqparse.RequestParser()
+        parser.add_argument("operation_data", type=list, required=True, nullable=True, location="json")
+        args = parser.parse_args()
+        metadata_args = MetadataOperationData(**args)
+
+        MetadataService.update_documents_metadata(dataset, metadata_args)
+
+        return 200
+
+
+api.add_resource(DatasetMetadataCreateApi, "/datasets/<uuid:dataset_id>/metadata")
+api.add_resource(DatasetMetadataApi, "/datasets/<uuid:dataset_id>/metadata/<uuid:metadata_id>")
+api.add_resource(DatasetMetadataBuiltInFieldApi, "/datasets/metadata/built-in")
+api.add_resource(DatasetMetadataBuiltInFieldActionApi, "/datasets/<uuid:dataset_id>/metadata/built-in/<string:action>")
+api.add_resource(DocumentMetadataEditApi, "/datasets/<uuid:dataset_id>/documents/metadata")

api/controllers/console/workspace/workspace.py

@@ -26,6 +26,7 @@ from libs.helper import TimestampField
 from libs.login import login_required
 from models.account import Tenant, TenantStatus
 from services.account_service import TenantService
+from services.feature_service import FeatureService
 from services.file_service import FileService
 from services.workspace_service import WorkspaceService
 
@@ -68,6 +69,11 @@ class TenantListApi(Resource):
         tenants = TenantService.get_join_tenants(current_user)
 
         for tenant in tenants:
+            features = FeatureService.get_features(tenant.id)
+            if features.billing.enabled:
+                tenant.plan = features.billing.subscription.plan
+            else:
+                tenant.plan = "sandbox"
             if tenant.id == current_user.current_tenant_id:
                 tenant.current = True  # Set current=True for current tenant
         return {"workspaces": marshal(tenants, tenants_fields)}, 200
@@ -82,28 +88,20 @@ class WorkspaceListApi(Resource):
         parser.add_argument("limit", type=inputs.int_range(1, 100), required=False, default=20, location="args")
         args = parser.parse_args()
 
-        tenants = Tenant.query.order_by(Tenant.created_at.desc()).paginate(page=args["page"], per_page=args["limit"])
-
+        tenants = Tenant.query.order_by(Tenant.created_at.desc()).paginate(
+            page=args["page"], per_page=args["limit"], error_out=False
+        )
         has_more = False
-        if len(tenants.items) == args["limit"]:
-            current_page_first_tenant = tenants[-1]
-            rest_count = (
-                db.session.query(Tenant)
-                .filter(
-                    Tenant.created_at < current_page_first_tenant.created_at, Tenant.id != current_page_first_tenant.id
-                )
-                .count()
-            )
 
-            if rest_count > 0:
-                has_more = True
-        total = db.session.query(Tenant).count()
+        if tenants.has_next:
+            has_more = True
+
         return {
             "data": marshal(tenants.items, workspace_fields),
             "has_more": has_more,
             "limit": args["limit"],
             "page": args["page"],
-            "total": total,
+            "total": tenants.total,
         }, 200
 
 

api/controllers/console/wraps.py

@@ -1,5 +1,6 @@
 import json
 import os
+import time
 from functools import wraps
 
 from flask import abort, request
@@ -8,6 +9,8 @@ from flask_login import current_user  # type: ignore
 from configs import dify_config
 from controllers.console.workspace.error import AccountNotInitializedError
 from extensions.ext_database import db
+from extensions.ext_redis import redis_client
+from models.dataset import RateLimitLog
 from models.model import DifySetup
 from services.feature_service import FeatureService, LicenseStatus
 from services.operation_service import OperationService
@@ -67,7 +70,9 @@ def cloud_edition_billing_resource_check(resource: str):
                 elif resource == "apps" and 0 < apps.limit <= apps.size:
                     abort(403, "The number of apps has reached the limit of your subscription.")
                 elif resource == "vector_space" and 0 < vector_space.limit <= vector_space.size:
-                    abort(403, "The capacity of the vector space has reached the limit of your subscription.")
+                    abort(
+                        403, "The capacity of the knowledge storage space has reached the limit of your subscription."
+                    )
                 elif resource == "documents" and 0 < documents_upload_quota.limit <= documents_upload_quota.size:
                     # The api of file upload is used in the multiple places,
                     # so we need to check the source of the request from datasets
@@ -112,6 +117,41 @@ def cloud_edition_billing_knowledge_limit_check(resource: str):
     return interceptor
 
 
+def cloud_edition_billing_rate_limit_check(resource: str):
+    def interceptor(view):
+        @wraps(view)
+        def decorated(*args, **kwargs):
+            if resource == "knowledge":
+                knowledge_rate_limit = FeatureService.get_knowledge_rate_limit(current_user.current_tenant_id)
+                if knowledge_rate_limit.enabled:
+                    current_time = int(time.time() * 1000)
+                    key = f"rate_limit_{current_user.current_tenant_id}"
+
+                    redis_client.zadd(key, {current_time: current_time})
+
+                    redis_client.zremrangebyscore(key, 0, current_time - 60000)
+
+                    request_count = redis_client.zcard(key)
+
+                    if request_count > knowledge_rate_limit.limit:
+                        # add ratelimit record
+                        rate_limit_log = RateLimitLog(
+                            tenant_id=current_user.current_tenant_id,
+                            subscription_plan=knowledge_rate_limit.subscription_plan,
+                            operation="knowledge",
+                        )
+                        db.session.add(rate_limit_log)
+                        db.session.commit()
+                        abort(
+                            403, "Sorry, you have reached the knowledge base request rate limit of your subscription."
+                        )
+            return view(*args, **kwargs)
+
+        return decorated
+
+    return interceptor
+
+
 def cloud_utm_record(view):
     @wraps(view)
     def decorated(*args, **kwargs):

api/controllers/service_api/app/message.py

@@ -10,7 +10,7 @@ from controllers.service_api.app.error import NotChatAppError
 from controllers.service_api.wraps import FetchUserArg, WhereisUserArg, validate_app_token
 from core.app.entities.app_invoke_entities import InvokeFrom
 from fields.conversation_fields import message_file_fields
-from fields.message_fields import feedback_fields, retriever_resource_fields
+from fields.message_fields import agent_thought_fields, feedback_fields, retriever_resource_fields
 from fields.raws import FilesContainedField
 from libs.helper import TimestampField, uuid_value
 from models.model import App, AppMode, EndUser
@@ -19,20 +19,6 @@ from services.message_service import MessageService
 
 
 class MessageListApi(Resource):
-    agent_thought_fields = {
-        "id": fields.String,
-        "chain_id": fields.String,
-        "message_id": fields.String,
-        "position": fields.Integer,
-        "thought": fields.String,
-        "tool": fields.String,
-        "tool_labels": fields.Raw,
-        "tool_input": fields.String,
-        "created_at": TimestampField,
-        "observation": fields.String,
-        "message_files": fields.List(fields.Nested(message_file_fields)),
-    }
-
     message_fields = {
         "id": fields.String,
         "conversation_id": fields.String,

api/controllers/service_api/wraps.py

@@ -1,3 +1,4 @@
+import time
 from collections.abc import Callable
 from datetime import UTC, datetime, timedelta
 from enum import Enum
@@ -13,8 +14,10 @@ from sqlalchemy.orm import Session
 from werkzeug.exceptions import Forbidden, Unauthorized
 
 from extensions.ext_database import db
+from extensions.ext_redis import redis_client
 from libs.login import _get_user
 from models.account import Account, Tenant, TenantAccountJoin, TenantStatus
+from models.dataset import RateLimitLog
 from models.model import ApiToken, App, EndUser
 from services.feature_service import FeatureService
 
@@ -139,6 +142,43 @@ def cloud_edition_billing_knowledge_limit_check(resource: str, api_token_type: s
     return interceptor
 
 
+def cloud_edition_billing_rate_limit_check(resource: str, api_token_type: str):
+    def interceptor(view):
+        @wraps(view)
+        def decorated(*args, **kwargs):
+            api_token = validate_and_get_api_token(api_token_type)
+
+            if resource == "knowledge":
+                knowledge_rate_limit = FeatureService.get_knowledge_rate_limit(api_token.tenant_id)
+                if knowledge_rate_limit.enabled:
+                    current_time = int(time.time() * 1000)
+                    key = f"rate_limit_{api_token.tenant_id}"
+
+                    redis_client.zadd(key, {current_time: current_time})
+
+                    redis_client.zremrangebyscore(key, 0, current_time - 60000)
+
+                    request_count = redis_client.zcard(key)
+
+                    if request_count > knowledge_rate_limit.limit:
+                        # add ratelimit record
+                        rate_limit_log = RateLimitLog(
+                            tenant_id=api_token.tenant_id,
+                            subscription_plan=knowledge_rate_limit.subscription_plan,
+                            operation="knowledge",
+                        )
+                        db.session.add(rate_limit_log)
+                        db.session.commit()
+                        raise Forbidden(
+                            "Sorry, you have reached the knowledge base request rate limit of your subscription."
+                        )
+            return view(*args, **kwargs)
+
+        return decorated
+
+    return interceptor
+
+
 def validate_dataset_token(view=None):
     def decorator(view):
         @wraps(view)

api/core/app/app_config/easy_ui_based_app/dataset/manager.py

@@ -1,7 +1,12 @@
 import uuid
 from typing import Optional
 
-from core.app.app_config.entities import DatasetEntity, DatasetRetrieveConfigEntity
+from core.app.app_config.entities import (
+    DatasetEntity,
+    DatasetRetrieveConfigEntity,
+    MetadataFilteringCondition,
+    ModelConfig,
+)
 from core.entities.agent_entities import PlanningStrategy
 from models.model import AppMode
 from services.dataset_service import DatasetService
@@ -78,6 +83,15 @@ class DatasetConfigManager:
                     retrieve_strategy=DatasetRetrieveConfigEntity.RetrieveStrategy.value_of(
                         dataset_configs["retrieval_model"]
                     ),
+                    metadata_filtering_mode=dataset_configs.get("metadata_filtering_mode", "disabled"),
+                    metadata_model_config=ModelConfig(**dataset_configs.get("metadata_model_config"))
+                    if dataset_configs.get("metadata_model_config")
+                    else None,
+                    metadata_filtering_conditions=MetadataFilteringCondition(
+                        **dataset_configs.get("metadata_filtering_conditions", {})
+                    )
+                    if dataset_configs.get("metadata_filtering_conditions")
+                    else None,
                 ),
             )
         else:
@@ -89,11 +103,22 @@ class DatasetConfigManager:
                         dataset_configs["retrieval_model"]
                     ),
                     top_k=dataset_configs.get("top_k", 4),
-                    score_threshold=dataset_configs.get("score_threshold"),
+                    score_threshold=dataset_configs.get("score_threshold")
+                    if dataset_configs.get("score_threshold_enabled", False)
+                    else None,
                     reranking_model=dataset_configs.get("reranking_model"),
                     weights=dataset_configs.get("weights"),
                     reranking_enabled=dataset_configs.get("reranking_enabled", True),
                     rerank_mode=dataset_configs.get("reranking_mode", "reranking_model"),
+                    metadata_filtering_mode=dataset_configs.get("metadata_filtering_mode", "disabled"),
+                    metadata_model_config=ModelConfig(**dataset_configs.get("metadata_model_config"))
+                    if dataset_configs.get("metadata_model_config")
+                    else None,
+                    metadata_filtering_conditions=MetadataFilteringCondition(
+                        **dataset_configs.get("metadata_filtering_conditions", {})
+                    )
+                    if dataset_configs.get("metadata_filtering_conditions")
+                    else None,
                 ),
             )
 

api/core/app/app_config/entities.py

@@ -1,10 +1,11 @@
 from collections.abc import Sequence
 from enum import Enum, StrEnum
-from typing import Any, Optional
+from typing import Any, Literal, Optional
 
 from pydantic import BaseModel, Field, field_validator
 
 from core.file import FileTransferMethod, FileType, FileUploadConfig
+from core.model_runtime.entities.llm_entities import LLMMode
 from core.model_runtime.entities.message_entities import PromptMessageRole
 from models.model import AppMode
 
@@ -135,6 +136,55 @@ class ExternalDataVariableEntity(BaseModel):
     config: dict[str, Any] = Field(default_factory=dict)
 
 
+SupportedComparisonOperator = Literal[
+    # for string or array
+    "contains",
+    "not contains",
+    "start with",
+    "end with",
+    "is",
+    "is not",
+    "empty",
+    "not empty",
+    # for number
+    "=",
+    "≠",
+    ">",
+    "<",
+    "≥",
+    "≤",
+    # for time
+    "before",
+    "after",
+]
+
+
+class ModelConfig(BaseModel):
+    provider: str
+    name: str
+    mode: LLMMode
+    completion_params: dict[str, Any] = {}
+
+
+class Condition(BaseModel):
+    """
+    Conditon detail
+    """
+
+    name: str
+    comparison_operator: SupportedComparisonOperator
+    value: str | Sequence[str] | None | int | float = None
+
+
+class MetadataFilteringCondition(BaseModel):
+    """
+    Metadata Filtering Condition.
+    """
+
+    logical_operator: Optional[Literal["and", "or"]] = "and"
+    conditions: Optional[list[Condition]] = Field(default=None, deprecated=True)
+
+
 class DatasetRetrieveConfigEntity(BaseModel):
     """
     Dataset Retrieve Config Entity.
@@ -171,6 +221,9 @@ class DatasetRetrieveConfigEntity(BaseModel):
     reranking_model: Optional[dict] = None
     weights: Optional[dict] = None
     reranking_enabled: Optional[bool] = True
+    metadata_filtering_mode: Optional[Literal["disabled", "automatic", "manual"]] = "disabled"
+    metadata_model_config: Optional[ModelConfig] = None
+    metadata_filtering_conditions: Optional[MetadataFilteringCondition] = None
 
 
 class DatasetEntity(BaseModel):

api/core/app/app_config/features/file_upload/manager.py

@@ -17,17 +17,15 @@ class FileUploadConfigManager:
         if file_upload_dict:
             if file_upload_dict.get("enabled"):
                 transform_methods = file_upload_dict.get("allowed_file_upload_methods", [])
-                data = {
-                    "image_config": {
-                        "number_limits": file_upload_dict["number_limits"],
-                        "transfer_methods": transform_methods,
-                    }
+                file_upload_dict["image_config"] = {
+                    "number_limits": file_upload_dict.get("number_limits", 1),
+                    "transfer_methods": transform_methods,
                 }
 
                 if is_vision:
-                    data["image_config"]["detail"] = file_upload_dict.get("image", {}).get("detail", "low")
+                    file_upload_dict["image_config"]["detail"] = file_upload_dict.get("image", {}).get("detail", "high")
 
-                return FileUploadConfig.model_validate(data)
+                return FileUploadConfig.model_validate(file_upload_dict)
 
     @classmethod
     def validate_and_set_defaults(cls, config: dict) -> tuple[dict, list[str]]:

api/core/app/apps/base_app_generator.py

@@ -151,7 +151,7 @@ class BaseAppGenerator:
 
             def gen():
                 for message in generator:
-                    if isinstance(message, (Mapping, dict)):
+                    if isinstance(message, Mapping | dict):
                         yield f"data: {json.dumps(message)}\n\n"
                     else:
                         yield f"event: {message}\n\n"

api/core/app/apps/base_app_runner.py

@@ -17,7 +17,11 @@ from core.external_data_tool.external_data_fetch import ExternalDataFetch
 from core.memory.token_buffer_memory import TokenBufferMemory
 from core.model_manager import ModelInstance
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta, LLMUsage
-from core.model_runtime.entities.message_entities import AssistantPromptMessage, PromptMessage
+from core.model_runtime.entities.message_entities import (
+    AssistantPromptMessage,
+    ImagePromptMessageContent,
+    PromptMessage,
+)
 from core.model_runtime.entities.model_entities import ModelPropertyKey
 from core.model_runtime.errors.invoke import InvokeBadRequestError
 from core.moderation.input_moderation import InputModeration
@@ -141,6 +145,7 @@ class AppRunner:
         query: Optional[str] = None,
         context: Optional[str] = None,
         memory: Optional[TokenBufferMemory] = None,
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
     ) -> tuple[list[PromptMessage], Optional[list[str]]]:
         """
         Organize prompt messages
@@ -167,6 +172,7 @@ class AppRunner:
                 context=context,
                 memory=memory,
                 model_config=model_config,
+                image_detail_config=image_detail_config,
             )
         else:
             memory_config = MemoryConfig(window=MemoryConfig.WindowConfig(enabled=False))
@@ -201,6 +207,7 @@ class AppRunner:
                 memory_config=memory_config,
                 memory=memory,
                 model_config=model_config,
+                image_detail_config=image_detail_config,
             )
             stop = model_config.stop
 

api/core/app/apps/chat/app_runner.py

@@ -11,6 +11,7 @@ from core.app.entities.queue_entities import QueueAnnotationReplyEvent
 from core.callback_handler.index_tool_callback_handler import DatasetIndexToolCallbackHandler
 from core.memory.token_buffer_memory import TokenBufferMemory
 from core.model_manager import ModelInstance
+from core.model_runtime.entities.message_entities import ImagePromptMessageContent
 from core.moderation.base import ModerationError
 from core.rag.retrieval.dataset_retrieval import DatasetRetrieval
 from extensions.ext_database import db
@@ -50,6 +51,16 @@ class ChatAppRunner(AppRunner):
         query = application_generate_entity.query
         files = application_generate_entity.files
 
+        image_detail_config = (
+            application_generate_entity.file_upload_config.image_config.detail
+            if (
+                application_generate_entity.file_upload_config
+                and application_generate_entity.file_upload_config.image_config
+            )
+            else None
+        )
+        image_detail_config = image_detail_config or ImagePromptMessageContent.DETAIL.LOW
+
         # Pre-calculate the number of tokens of the prompt messages,
         # and return the rest number of tokens by model context token size limit and max token size limit.
         # If the rest number of tokens is not enough, raise exception.
@@ -85,6 +96,7 @@ class ChatAppRunner(AppRunner):
             files=files,
             query=query,
             memory=memory,
+            image_detail_config=image_detail_config,
         )
 
         # moderation
@@ -168,6 +180,7 @@ class ChatAppRunner(AppRunner):
                 hit_callback=hit_callback,
                 memory=memory,
                 message_id=message.id,
+                inputs=inputs,
             )
 
         # reorganize all inputs and template to prompt messages
@@ -182,6 +195,7 @@ class ChatAppRunner(AppRunner):
             query=query,
             context=context,
             memory=memory,
+            image_detail_config=image_detail_config,
         )
 
         # check hosting moderation

api/core/app/apps/completion/app_runner.py

@@ -9,6 +9,7 @@ from core.app.entities.app_invoke_entities import (
 )
 from core.callback_handler.index_tool_callback_handler import DatasetIndexToolCallbackHandler
 from core.model_manager import ModelInstance
+from core.model_runtime.entities.message_entities import ImagePromptMessageContent
 from core.moderation.base import ModerationError
 from core.rag.retrieval.dataset_retrieval import DatasetRetrieval
 from extensions.ext_database import db
@@ -43,6 +44,16 @@ class CompletionAppRunner(AppRunner):
         query = application_generate_entity.query
         files = application_generate_entity.files
 
+        image_detail_config = (
+            application_generate_entity.file_upload_config.image_config.detail
+            if (
+                application_generate_entity.file_upload_config
+                and application_generate_entity.file_upload_config.image_config
+            )
+            else None
+        )
+        image_detail_config = image_detail_config or ImagePromptMessageContent.DETAIL.LOW
+
         # Pre-calculate the number of tokens of the prompt messages,
         # and return the rest number of tokens by model context token size limit and max token size limit.
         # If the rest number of tokens is not enough, raise exception.
@@ -66,6 +77,7 @@ class CompletionAppRunner(AppRunner):
             inputs=inputs,
             files=files,
             query=query,
+            image_detail_config=image_detail_config,
         )
 
         # moderation
@@ -127,6 +139,7 @@ class CompletionAppRunner(AppRunner):
                 show_retrieve_source=app_config.additional_features.show_retrieve_source,
                 hit_callback=hit_callback,
                 message_id=message.id,
+                inputs=inputs,
             )
 
         # reorganize all inputs and template to prompt messages
@@ -140,6 +153,7 @@ class CompletionAppRunner(AppRunner):
             files=files,
             query=query,
             context=context,
+            image_detail_config=image_detail_config,
         )
 
         # check hosting moderation

api/core/auto/config/custom.yaml

@@ -1,27 +0,0 @@
-# 自定义配置文件
-workflow_generator:
-  # 用于生成工作流的模型配置
-  models:
-    default: my-gpt-4o-mini  # 默认使用的模型
-    available:  # 可用的模型列表
-      my-gpt-4o-mini:
-        model_name: gpt-4o-mini
-        base_url: https://api.pandalla.ai/v1
-        key_path: ./openai_key
-        max_tokens: 4096
-      my-gpt-4o:
-        model_name: gpt-4o
-        base_url: https://api.pandalla.ai/v1
-        key_path: ./openai_key
-        max_tokens: 4096
-
-# 调试配置
-debug:
-  enabled: false  # 默认不启用调试模式，可通过命令行参数 --debug 启用
-  dir: debug/     # 调试信息保存目录
-  save_options:   # 调试信息保存选项
-    prompt: true  # 保存提示词
-    response: true  # 保存大模型响应
-    json: true  # 保存JSON解析过程
-    workflow: true  # 保存工作流生成过程
-  case_id_format: "%Y%m%d_%H%M%S_%f"  # 运行ID格式，使用datetime.strftime格式

api/core/auto/config/default.yaml

@@ -1,33 +0,0 @@
-# 默认配置文件
-
-# 工作流生成器配置
-workflow_generator:
-  # 用于生成工作流的模型配置
-  models:
-    default: gpt-4  # 默认使用的模型
-    available:  # 可用的模型列表
-      gpt-4:
-        model_name: gpt-4
-        base_url: https://api.openai.com/v1
-        key_path: ./openai_key
-        max_tokens: 8192
-      gpt-4-turbo:
-        model_name: gpt-4-1106-preview
-        base_url: https://api.openai.com/v1
-        key_path: ./openai_key
-        max_tokens: 4096
-
-# 工作流节点配置
-workflow_nodes:
-  # LLM节点默认配置（使用 Dify 平台配置的模型）
-  llm:
-    provider: zhipuai
-    model: glm-4-flash
-    max_tokens: 16384
-    temperature: 0.7
-    mode: chat
-
-# 输出配置
-output:
-  dir: output/
-  filename: generated_workflow.yml 

api/core/auto/node_types/__init__.py

@@ -1,78 +0,0 @@
-from .agent import AgentNodeType
-from .answer import AnswerNodeType
-from .assigner import AssignerNodeType
-from .code import CodeLanguage, CodeNodeType, OutputVar
-from .common import (
-    BlockEnum,
-    CommonEdgeType,
-    CommonNodeType,
-    CompleteEdge,
-    CompleteNode,
-    Context,
-    InputVar,
-    InputVarType,
-    Memory,
-    ModelConfig,
-    PromptItem,
-    PromptRole,
-    ValueSelector,
-    Variable,
-    VarType,
-    VisionSetting,
-)
-from .end import EndNodeType
-from .http import HttpNodeType
-from .if_else import IfElseNodeType
-from .iteration import IterationNodeType
-from .iteration_start import IterationStartNodeType
-from .knowledge_retrieval import KnowledgeRetrievalNodeType
-from .list_operator import ListFilterNodeType
-from .llm import LLMNodeType, VisionConfig
-from .note_node import NoteNodeType
-from .parameter_extractor import ParameterExtractorNodeType
-from .question_classifier import QuestionClassifierNodeType
-from .start import StartNodeType
-from .template_transform import TemplateTransformNodeType
-from .tool import ToolNodeType
-from .variable_assigner import VariableAssignerNodeType
-
-__all__ = [
-    "AgentNodeType",
-    "AnswerNodeType",
-    "AssignerNodeType",
-    "BlockEnum",
-    "CodeLanguage",
-    "CodeNodeType",
-    "CommonEdgeType",
-    "CommonNodeType",
-    "CompleteEdge",
-    "CompleteNode",
-    "Context",
-    "EndNodeType",
-    "HttpNodeType",
-    "IfElseNodeType",
-    "InputVar",
-    "InputVarType",
-    "IterationNodeType",
-    "IterationStartNodeType",
-    "KnowledgeRetrievalNodeType",
-    "LLMNodeType",
-    "ListFilterNodeType",
-    "Memory",
-    "ModelConfig",
-    "NoteNodeType",
-    "OutputVar",
-    "ParameterExtractorNodeType",
-    "PromptItem",
-    "PromptRole",
-    "QuestionClassifierNodeType",
-    "StartNodeType",
-    "TemplateTransformNodeType",
-    "ToolNodeType",
-    "ValueSelector",
-    "VarType",
-    "Variable",
-    "VariableAssignerNodeType",
-    "VisionConfig",
-    "VisionSetting",
-]

api/core/auto/node_types/agent.py

@@ -1,34 +0,0 @@
-from typing import Any, Optional
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType
-
-# Introduce previously defined CommonNodeType and ToolVarInputs
-# Assume they are defined in the same module
-
-
-class ToolVarInputs(BaseModel):
-    variable_name: Optional[str] = None
-    default_value: Optional[Any] = None
-
-
-class AgentNodeType(CommonNodeType):
-    agent_strategy_provider_name: Optional[str] = None
-    agent_strategy_name: Optional[str] = None
-    agent_strategy_label: Optional[str] = None
-    agent_parameters: Optional[ToolVarInputs] = None
-    output_schema: dict[str, Any]
-    plugin_unique_identifier: Optional[str] = None
-
-
-# 示例用法
-if __name__ == "__main__":
-    example_node = AgentNodeType(
-        title="Example Agent",
-        desc="An agent node example",
-        type=BlockEnum.agent,
-        output_schema={"key": "value"},
-        agent_parameters=ToolVarInputs(variable_name="example_var", default_value="default"),
-    )
-    print(example_node)

api/core/auto/node_types/answer.py

@@ -1,21 +0,0 @@
-from .common import BlockEnum, CommonNodeType, Variable
-
-
-class AnswerNodeType(CommonNodeType):
-    variables: list[Variable]
-    answer: str
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = AnswerNodeType(
-        title="Example Answer Node",
-        desc="An answer node example",
-        type=BlockEnum.answer,
-        answer="This is the answer",
-        variables=[
-            Variable(variable="var1", value_selector=["node1", "key1"]),
-            Variable(variable="var2", value_selector=["node2", "key2"]),
-        ],
-    )
-    print(example_node)

api/core/auto/node_types/assigner.py

@@ -1,62 +0,0 @@
-from enum import Enum
-from typing import Any, Optional
-
-from pydantic import BaseModel, Field
-
-from .common import BlockEnum, CommonNodeType
-
-# Import previously defined CommonNodeType and ValueSelector
-# Assume they are defined in the same module
-
-
-class WriteMode(str, Enum):
-    overwrite = "over-write"
-    clear = "clear"
-    append = "append"
-    extend = "extend"
-    set = "set"
-    increment = "+="
-    decrement = "-="
-    multiply = "*="
-    divide = "/="
-
-
-class AssignerNodeInputType(str, Enum):
-    variable = "variable"
-    constant = "constant"
-
-
-class AssignerNodeOperation(BaseModel):
-    variable_selector: Any  # Placeholder for ValueSelector type
-    input_type: AssignerNodeInputType
-    operation: WriteMode
-    value: Any
-
-
-class AssignerNodeType(CommonNodeType):
-    version: Optional[str] = Field(None, pattern="^[12]$")  # Version is '1' or '2'
-    items: list[AssignerNodeOperation]
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = AssignerNodeType(
-        title="Example Assigner Node",
-        desc="An assigner node example",
-        type=BlockEnum.variable_assigner,
-        items=[
-            AssignerNodeOperation(
-                variable_selector={"nodeId": "node1", "key": "value"},  # Example ValueSelector
-                input_type=AssignerNodeInputType.variable,
-                operation=WriteMode.set,
-                value="newValue",
-            ),
-            AssignerNodeOperation(
-                variable_selector={"nodeId": "node2", "key": "value"},
-                input_type=AssignerNodeInputType.constant,
-                operation=WriteMode.increment,
-                value=1,
-            ),
-        ],
-    )
-    print(example_node)

api/core/auto/node_types/code.py

@@ -1,56 +0,0 @@
-from enum import Enum
-from typing import Optional
-
-from pydantic import BaseModel
-
-from core.auto.node_types.common import BlockEnum, CommonNodeType, Variable, VarType
-
-# 引入之前定义的 CommonNodeType、VarType 和 Variable
-# 假设它们在同一模块中定义
-
-
-class CodeLanguage(str, Enum):
-    python3 = "python3"
-    javascript = "javascript"
-    json = "json"
-
-
-class OutputVar(BaseModel):
-    type: VarType
-    children: Optional[None] = None  # 未来支持嵌套
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，确保正确序列化"""
-        result = {"type": self.type.value if isinstance(self.type, Enum) else self.type}
-
-        if self.children is not None:
-            result["children"] = self.children
-
-        return result
-
-
-class CodeNodeType(CommonNodeType):
-    variables: list[Variable]
-    code_language: CodeLanguage
-    code: str
-    outputs: dict[str, OutputVar]
-
-
-# 示例用法
-if __name__ == "__main__":
-    # 创建示例节点
-    example_node = CodeNodeType(
-        title="Example Code Node",
-        desc="A code node example",
-        type=BlockEnum.code,
-        code_language=CodeLanguage.python3,
-        code="print('Hello, World!')",
-        outputs={
-            "output1": OutputVar(type=VarType.string),
-            "output2": OutputVar(type=VarType.number),
-        },
-        variables=[
-            Variable(variable="var1", value_selector=["node1", "key1"]),
-        ],
-    )
-    print(example_node.get_all_required_fields())

api/core/auto/node_types/common.py

@@ -1,690 +0,0 @@
-from enum import Enum
-from typing import Any, Optional, Union
-
-import yaml
-from pydantic import BaseModel, Field
-
-
-# BlockEnum 枚举
-class BlockEnum(str, Enum):
-    start = "start"
-    end = "end"
-    answer = "answer"
-    llm = "llm"
-    knowledge_retrieval = "knowledge-retrieval"
-    question_classifier = "question-classifier"
-    if_else = "if-else"
-    code = "code"
-    template_transform = "template-transform"
-    http_request = "http-request"
-    variable_assigner = "variable-assigner"
-    variable_aggregator = "variable-aggregator"
-    tool = "tool"
-    parameter_extractor = "parameter-extractor"
-    iteration = "iteration"
-    document_extractor = "document-extractor"
-    list_operator = "list-operator"
-    iteration_start = "iteration-start"
-    assigner = "assigner"  # is now named as VariableAssigner
-    agent = "agent"
-
-
-# Error枚举
-class ErrorHandleMode(str, Enum):
-    terminated = "terminated"
-    continue_on_error = "continue-on-error"
-    remove_abnormal_output = "remove-abnormal-output"
-
-
-class ErrorHandleTypeEnum(str, Enum):
-    none = ("none",)
-    failBranch = ("fail-branch",)
-    defaultValue = ("default-value",)
-
-
-# Branch 类型
-class Branch(BaseModel):
-    id: str
-    name: str
-
-
-# NodeRunningStatus 枚举
-class NodeRunningStatus(str, Enum):
-    not_start = "not-start"
-    waiting = "waiting"
-    running = "running"
-    succeeded = "succeeded"
-    failed = "failed"
-    exception = "exception"
-    retry = "retry"
-
-
-# 创建一个基类来统一CommonNodeType和CommonEdgeType的序列化逻辑
-class BaseType(BaseModel):
-    """基类，用于统一CommonNodeType和CommonEdgeType的序列化逻辑"""
-
-    def to_json(self) -> dict[str, Any]:
-        """
-        将对象转换为JSON格式的字典,通过循环模型字段来构建JSON数据
-        """
-        json_data = {}
-
-        # 获取模型的所有字段
-        for field_name, field_value in self.__dict__.items():
-            if field_value is not None:
-                # 特殊处理Branch类型的列表
-                if field_name == "_targetBranches" and field_value is not None:
-                    json_data[field_name] = [branch.dict(exclude_none=True) for branch in field_value]
-                # 处理枚举类型
-                elif isinstance(field_value, Enum):
-                    json_data[field_name] = field_value.value
-                # 处理嵌套的Pydantic模型
-                elif hasattr(field_value, "dict") and callable(field_value.dict):
-                    json_data[field_name] = field_value.dict(exclude_none=True)
-                # 处理列表中的Pydantic模型
-                elif isinstance(field_value, list):
-                    processed_list = []
-                    for item in field_value:
-                        if hasattr(item, "dict") and callable(item.dict):
-                            processed_list.append(item.dict(exclude_none=True))
-                        else:
-                            processed_list.append(item)
-                    json_data[field_name] = processed_list
-                # 处理字典中的Pydantic模型
-                elif isinstance(field_value, dict):
-                    processed_dict = {}
-                    for key, value in field_value.items():
-                        if hasattr(value, "dict") and callable(value.dict):
-                            processed_dict[key] = value.dict(exclude_none=True)
-                        else:
-                            processed_dict[key] = value
-                    json_data[field_name] = processed_dict
-                # 其他字段直接添加
-                else:
-                    json_data[field_name] = field_value
-
-        return json_data
-
-
-# CommonNodeType 类型
-class CommonNodeType(BaseType):
-    _connectedSourceHandleIds: Optional[list[str]] = None
-    _connectedTargetHandleIds: Optional[list[str]] = None
-    _targetBranches: Optional[list[Branch]] = None
-    _isSingleRun: Optional[bool] = None
-    _runningStatus: Optional[NodeRunningStatus] = None
-    _singleRunningStatus: Optional[NodeRunningStatus] = None
-    _isCandidate: Optional[bool] = None
-    _isBundled: Optional[bool] = None
-    _children: Optional[list[str]] = None
-    _isEntering: Optional[bool] = None
-    _showAddVariablePopup: Optional[bool] = None
-    _holdAddVariablePopup: Optional[bool] = None
-    _iterationLength: Optional[int] = None
-    _iterationIndex: Optional[int] = None
-    _inParallelHovering: Optional[bool] = None
-    isInIteration: Optional[bool] = None
-    iteration_id: Optional[str] = None
-    selected: Optional[bool] = None
-    title: str
-    desc: str
-    type: BlockEnum
-    width: Optional[float] = None
-    height: Optional[float] = None
-
-    @classmethod
-    def get_all_required_fields(cls) -> dict[str, str]:
-        """
-        获取所有必选字段，包括从父类继承的字段
-        这是一个类方法，可以通过类直接调用
-        """
-        all_required_fields = {}
-
-        # 获取所有父类（除了 object 和 BaseModel）
-        mro = [c for c in cls.__mro__ if c not in (object, BaseModel, BaseType)]
-
-        # 从父类到子类的顺序处理，这样子类的字段会覆盖父类的同名字段
-        for class_type in reversed(mro):
-            if hasattr(class_type, "__annotations__"):
-                for field_name, field_info in class_type.__annotations__.items():
-                    # 检查字段是否有默认值
-                    has_default = hasattr(class_type, field_name)
-                    # 检查字段是否为可选类型
-                    is_optional = "Optional" in str(field_info)
-
-                    # 如果字段没有默认值且不是Optional类型，则为必选字段
-                    if not has_default and not is_optional:
-                        all_required_fields[field_name] = str(field_info)
-
-        return all_required_fields
-
-
-# CommonEdgeType 类型
-class CommonEdgeType(BaseType):
-    _hovering: Optional[bool] = None
-    _connectedNodeIsHovering: Optional[bool] = None
-    _connectedNodeIsSelected: Optional[bool] = None
-    _run: Optional[bool] = None
-    _isBundled: Optional[bool] = None
-    isInIteration: Optional[bool] = None
-    iteration_id: Optional[str] = None
-    sourceType: BlockEnum
-    targetType: BlockEnum
-
-
-class ValueSelector(BaseModel):
-    """Value selector for selecting values from other nodes."""
-
-    value: list[str] = Field(default_factory=list)
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，直接返回 value 列表"""
-        return self.value
-
-
-# Add Context class for LLM node
-class Context(BaseModel):
-    """Context configuration for LLM node."""
-
-    enabled: bool = False
-    variable_selector: Optional[ValueSelector] = None
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，确保 variable_selector 字段正确序列化"""
-        result = {"enabled": self.enabled}
-
-        if self.variable_selector:
-            result["variable_selector"] = self.variable_selector.dict()
-        else:
-            result["variable_selector"] = []
-
-        return result
-
-
-# Variable 类型
-class Variable(BaseModel):
-    """
-    变量类型，用于定义节点的输入/输出变量
-    与Dify中的Variable类型保持一致
-    """
-
-    variable: str  # 变量名
-    label: Optional[Union[str, dict[str, str]]] = None  # 变量标签，可以是字符串或对象
-    value_selector: list[str]  # 变量值选择器，格式为[nodeId, key]
-    variable_type: Optional[str] = None  # 变量类型，对应Dify中的VarType枚举
-    value: Optional[str] = None  # 变量值（常量值）
-    options: Optional[list[str]] = None  # 选项列表（用于select类型）
-    required: Optional[bool] = None  # 是否必填
-    isParagraph: Optional[bool] = None  # 是否为段落
-    max_length: Optional[int] = None  # 最大长度
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，确保正确序列化"""
-        result = {"variable": self.variable}
-
-        if self.label is not None:
-            result["label"] = self.label
-
-        if self.value_selector:
-            result["value_selector"] = self.value_selector
-
-        if self.variable_type is not None:
-            result["type"] = self.variable_type  # 使用type而不是variable_type，与Dify保持一致
-
-        if self.value is not None:
-            result["value"] = self.value
-
-        if self.options is not None:
-            result["options"] = self.options
-
-        if self.required is not None:
-            result["required"] = self.required
-
-        if self.isParagraph is not None:
-            result["isParagraph"] = self.isParagraph
-
-        if self.max_length is not None:
-            result["max_length"] = self.max_length
-
-        return result
-
-
-# EnvironmentVariable 类型
-class EnvironmentVariable(BaseModel):
-    id: str
-    name: str
-    value: Any
-    value_type: str  # Expecting to be either 'string', 'number', or 'secret'
-
-
-# ConversationVariable 类型
-class ConversationVariable(BaseModel):
-    id: str
-    name: str
-    value_type: str
-    value: Any
-    description: str
-
-
-# GlobalVariable 类型
-class GlobalVariable(BaseModel):
-    name: str
-    value_type: str  # Expecting to be either 'string' or 'number'
-    description: str
-
-
-# VariableWithValue 类型
-class VariableWithValue(BaseModel):
-    key: str
-    value: str
-
-
-# InputVarType 枚举
-class InputVarType(str, Enum):
-    text_input = "text-input"
-    paragraph = "paragraph"
-    select = "select"
-    number = "number"
-    url = "url"
-    files = "files"
-    json = "json"
-    contexts = "contexts"
-    iterator = "iterator"
-    file = "file"
-    file_list = "file-list"
-
-
-# InputVar 类型
-class InputVar(BaseModel):
-    type: InputVarType
-    label: Union[str, dict[str, Any]]  # 可以是字符串或对象
-    variable: str
-    max_length: Optional[int] = None
-    default: Optional[str] = None
-    required: bool
-    hint: Optional[str] = None
-    options: Optional[list[str]] = None
-    value_selector: Optional[list[str]] = None
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，确保正确序列化"""
-        result = {
-            "type": self.type.value if isinstance(self.type, Enum) else self.type,
-            "label": self.label,
-            "variable": self.variable,
-            "required": self.required,
-        }
-
-        if self.max_length is not None:
-            result["max_length"] = self.max_length
-
-        if self.default is not None:
-            result["default"] = self.default
-
-        if self.hint is not None:
-            result["hint"] = self.hint
-
-        if self.options is not None:
-            result["options"] = self.options
-
-        if self.value_selector is not None:
-            result["value_selector"] = self.value_selector
-
-        return result
-
-
-# ModelConfig 类型
-class ModelConfig(BaseModel):
-    provider: str
-    name: str
-    mode: str
-    completion_params: dict[str, Any]
-
-
-# PromptRole 枚举
-class PromptRole(str, Enum):
-    system = "system"
-    user = "user"
-    assistant = "assistant"
-
-
-# EditionType 枚举
-class EditionType(str, Enum):
-    basic = "basic"
-    jinja2 = "jinja2"
-
-
-# PromptItem 类型
-class PromptItem(BaseModel):
-    id: Optional[str] = None
-    role: Optional[PromptRole] = None
-    text: str
-    edition_type: Optional[EditionType] = None
-    jinja2_text: Optional[str] = None
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，确保 role 字段正确序列化"""
-        result = {"id": self.id, "text": self.text}
-
-        if self.role:
-            result["role"] = self.role.value
-
-        if self.edition_type:
-            result["edition_type"] = self.edition_type.value
-
-        if self.jinja2_text:
-            result["jinja2_text"] = self.jinja2_text
-
-        return result
-
-
-# MemoryRole 枚举
-class MemoryRole(str, Enum):
-    user = "user"
-    assistant = "assistant"
-
-
-# RolePrefix 类型
-class RolePrefix(BaseModel):
-    user: str
-    assistant: str
-
-
-# Memory 类型
-class Memory(BaseModel):
-    role_prefix: Optional[RolePrefix] = None
-    window: dict[str, Any]  # Expecting to have 'enabled' and 'size'
-    query_prompt_template: str
-
-
-# VarType 枚举
-class VarType(str, Enum):
-    string = "string"
-    number = "number"
-    secret = "secret"
-    boolean = "boolean"
-    object = "object"
-    file = "file"
-    array = "array"
-    arrayString = "array[string]"
-    arrayNumber = "array[number]"
-    arrayObject = "array[object]"
-    arrayFile = "array[file]"
-    any = "any"
-
-
-# Var 类型
-class Var(BaseModel):
-    variable: str
-    type: VarType
-    children: Optional[list["Var"]] = None  # Self-reference
-    isParagraph: Optional[bool] = None
-    isSelect: Optional[bool] = None
-    options: Optional[list[str]] = None
-    required: Optional[bool] = None
-    des: Optional[str] = None
-    isException: Optional[bool] = None
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，确保type字段正确序列化"""
-        result = {"variable": self.variable, "type": self.type.value if isinstance(self.type, Enum) else self.type}
-
-        if self.children is not None:
-            result["children"] = [child.dict() for child in self.children]
-
-        if self.isParagraph is not None:
-            result["isParagraph"] = self.isParagraph
-
-        if self.isSelect is not None:
-            result["isSelect"] = self.isSelect
-
-        if self.options is not None:
-            result["options"] = self.options
-
-        if self.required is not None:
-            result["required"] = self.required
-
-        if self.des is not None:
-            result["des"] = self.des
-
-        if self.isException is not None:
-            result["isException"] = self.isException
-
-        return result
-
-
-# NodeOutPutVar 类型
-class NodeOutPutVar(BaseModel):
-    nodeId: str
-    title: str
-    vars: list[Var]
-    isStartNode: Optional[bool] = None
-
-
-# Block 类型
-class Block(BaseModel):
-    classification: Optional[str] = None
-    type: BlockEnum
-    title: str
-    description: Optional[str] = None
-
-
-# NodeDefault 类型
-class NodeDefault(BaseModel):
-    defaultValue: dict[str, Any]
-    getAvailablePrevNodes: Any  # Placeholder for function reference
-    getAvailableNextNodes: Any  # Placeholder for function reference
-    checkValid: Any  # Placeholder for function reference
-
-
-# OnSelectBlock 类型
-class OnSelectBlock(BaseModel):
-    nodeType: BlockEnum
-    additional_data: Optional[dict[str, Any]] = None
-
-
-# WorkflowRunningStatus 枚举
-class WorkflowRunningStatus(str, Enum):
-    waiting = "waiting"
-    running = "running"
-    succeeded = "succeeded"
-    failed = "failed"
-    stopped = "stopped"
-
-
-# WorkflowVersion 枚举
-class WorkflowVersion(str, Enum):
-    draft = "draft"
-    latest = "latest"
-
-
-# OnNodeAdd 类型
-class OnNodeAdd(BaseModel):
-    nodeType: BlockEnum
-    sourceHandle: Optional[str] = None
-    targetHandle: Optional[str] = None
-    toolDefaultValue: Optional[dict[str, Any]] = None
-
-
-# CheckValidRes 类型
-class CheckValidRes(BaseModel):
-    isValid: bool
-    errorMessage: Optional[str] = None
-
-
-# RunFile 类型
-class RunFile(BaseModel):
-    type: str
-    transfer_method: list[str]
-    url: Optional[str] = None
-    upload_file_id: Optional[str] = None
-
-
-# WorkflowRunningData 类型
-class WorkflowRunningData(BaseModel):
-    task_id: Optional[str] = None
-    message_id: Optional[str] = None
-    conversation_id: Optional[str] = None
-    result: dict[str, Any]  # Expecting a structured object
-    tracing: Optional[list[dict[str, Any]]] = None  # Placeholder for NodeTracing
-
-
-# HistoryWorkflowData 类型
-class HistoryWorkflowData(BaseModel):
-    id: str
-    sequence_number: int
-    status: str
-    conversation_id: Optional[str] = None
-
-
-# ChangeType 枚举
-class ChangeType(str, Enum):
-    changeVarName = "changeVarName"
-    remove = "remove"
-
-
-# MoreInfo 类型
-class MoreInfo(BaseModel):
-    type: ChangeType
-    payload: Optional[dict[str, Any]] = None
-
-
-# ToolWithProvider 类型
-class ToolWithProvider(BaseModel):
-    tools: list[dict[str, Any]]  # Placeholder for Tool type
-
-
-# SupportUploadFileTypes 枚举
-class SupportUploadFileTypes(str, Enum):
-    image = "image"
-    document = "document"
-    audio = "audio"
-    video = "video"
-    custom = "custom"
-
-
-# UploadFileSetting 类型
-class UploadFileSetting(BaseModel):
-    allowed_file_upload_methods: list[str]
-    allowed_file_types: list[SupportUploadFileTypes]
-    allowed_file_extensions: Optional[list[str]] = None
-    max_length: int
-    number_limits: Optional[int] = None
-
-
-# VisionSetting 类型
-class VisionSetting(BaseModel):
-    variable_selector: list[str]
-    detail: dict[str, Any]  # Placeholder for Resolution type
-
-
-# 创建一个基类来统一序列化逻辑
-class CompleteBase(BaseModel):
-    """基类，用于统一CompleteNode和CompleteEdge的序列化逻辑"""
-
-    def to_json(self):
-        """将对象转换为JSON格式的字典"""
-        json_data = {}
-
-        # 获取模型的所有字段
-        for field_name, field_value in self.__dict__.items():
-            if field_value is not None:
-                # 处理嵌套的数据对象
-                if field_name == "data" and hasattr(field_value, "to_json"):
-                    json_data[field_name] = field_value.to_json()
-                # 处理枚举类型
-                elif isinstance(field_value, Enum):
-                    json_data[field_name] = field_value.value
-                # 处理嵌套的Pydantic模型
-                elif hasattr(field_value, "dict") and callable(field_value.dict):
-                    json_data[field_name] = field_value.dict(exclude_none=True)
-                # 处理列表中的Pydantic模型
-                elif isinstance(field_value, list):
-                    processed_list = []
-                    for item in field_value:
-                        if hasattr(item, "dict") and callable(item.dict):
-                            processed_list.append(item.dict(exclude_none=True))
-                        else:
-                            processed_list.append(item)
-                    json_data[field_name] = processed_list
-                # 处理字典中的Pydantic模型
-                elif isinstance(field_value, dict):
-                    processed_dict = {}
-                    for key, value in field_value.items():
-                        if hasattr(value, "dict") and callable(value.dict):
-                            processed_dict[key] = value.dict(exclude_none=True)
-                        else:
-                            processed_dict[key] = value
-                    json_data[field_name] = processed_dict
-                # 其他字段直接添加
-                else:
-                    json_data[field_name] = field_value
-
-        return json_data
-
-    def to_yaml(self):
-        """将对象转换为YAML格式的字符串"""
-        return yaml.dump(self.to_json(), allow_unicode=True)
-
-
-class CompleteNode(CompleteBase):
-    id: str
-    position: dict
-    height: int
-    width: float
-    positionAbsolute: dict
-    selected: bool
-    sourcePosition: Union[dict, str]
-    targetPosition: Union[dict, str]
-    type: str
-    data: Optional[Union[CommonNodeType, None]] = None  # Flexible field to store CommonNodeType or None
-
-    def add_data(self, data: Union[CommonNodeType, None]):
-        self.data = data
-
-    def to_json(self):
-        json_data = super().to_json()
-
-        # 特殊处理sourcePosition和targetPosition
-        json_data["sourcePosition"] = "right"  # 直接输出为字符串"right"
-        json_data["targetPosition"] = "left"  # 直接输出为字符串"left"
-
-        # 确保 width 是整数而不是浮点数
-        if isinstance(json_data["width"], float):
-            json_data["width"] = int(json_data["width"])
-
-        return json_data
-
-
-class CompleteEdge(CompleteBase):
-    id: str
-    source: str
-    sourceHandle: str
-    target: str
-    targetHandle: str
-    type: str
-    zIndex: int
-    data: Optional[Union[CommonEdgeType, None]] = None  # Flexible field to store CommonEdgeType or None
-
-    def add_data(self, data: Union[CommonEdgeType, None]):
-        self.data = data
-
-
-# 示例用法
-if __name__ == "__main__":
-    # 这里可以添加示例数据进行验证
-    common_node = CompleteNode(
-        id="1740019130520",
-        position={"x": 80, "y": 282},
-        height=100,
-        width=100,
-        positionAbsolute={"x": 80, "y": 282},
-        selected=True,
-        sourcePosition={"x": 80, "y": 282},
-        targetPosition={"x": 80, "y": 282},
-        type="custom",
-    )
-    common_data = CommonNodeType(title="示例节点", desc="这是一个示例节点", type="")
-    print(CommonNodeType.get_all_required_fields())
-    common_node.add_data(common_data)
-    # print(common_node)

api/core/auto/node_types/end.py

@@ -1,22 +0,0 @@
-from .common import BlockEnum, CommonNodeType, Variable
-
-# Import previously defined CommonNodeType and Variable
-# Assume they are defined in the same module
-
-
-class EndNodeType(CommonNodeType):
-    outputs: list[Variable]
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = EndNodeType(
-        title="Example End Node",
-        desc="An end node example",
-        type=BlockEnum.end,
-        outputs=[
-            Variable(variable="outputVar1", value_selector=["node1", "key1"]),
-            Variable(variable="outputVar2", value_selector=["node2", "key2"]),
-        ],
-    )
-    print(example_node)

api/core/auto/node_types/http.py

@@ -1,127 +0,0 @@
-from enum import Enum
-from typing import Optional, Union
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, ValueSelector, Variable
-
-# Import previously defined CommonNodeType, ValueSelector, and Variable
-# Assume they are defined in the same module
-
-
-class Method(str, Enum):
-    """HTTP request methods."""
-
-    get = "get"
-    post = "post"
-    head = "head"
-    patch = "patch"
-    put = "put"
-    delete = "delete"
-
-
-class BodyType(str, Enum):
-    """HTTP request body types."""
-
-    none = "none"
-    formData = "form-data"
-    xWwwFormUrlencoded = "x-www-form-urlencoded"
-    rawText = "raw-text"
-    json = "json"
-    binary = "binary"
-
-
-class BodyPayloadValueType(str, Enum):
-    """Types of values in body payload."""
-
-    text = "text"
-    file = "file"
-
-
-class BodyPayload(BaseModel):
-    """Body payload item for HTTP requests."""
-
-    id: Optional[str] = None
-    key: Optional[str] = None
-    type: BodyPayloadValueType
-    file: Optional[ValueSelector] = None  # Used when type is file
-    value: Optional[str] = None  # Used when type is text
-
-
-class Body(BaseModel):
-    """HTTP request body configuration."""
-
-    type: BodyType
-    data: Union[str, list[BodyPayload]]  # string is deprecated, will convert to BodyPayload
-
-
-class AuthorizationType(str, Enum):
-    """HTTP authorization types."""
-
-    none = "no-auth"
-    apiKey = "api-key"
-
-
-class APIType(str, Enum):
-    """API key types."""
-
-    basic = "basic"
-    bearer = "bearer"
-    custom = "custom"
-
-
-class AuthConfig(BaseModel):
-    """Authorization configuration."""
-
-    type: APIType
-    api_key: str
-    header: Optional[str] = None
-
-
-class Authorization(BaseModel):
-    """HTTP authorization settings."""
-
-    type: AuthorizationType
-    config: Optional[AuthConfig] = None
-
-
-class Timeout(BaseModel):
-    """HTTP request timeout settings."""
-
-    connect: Optional[int] = None
-    read: Optional[int] = None
-    write: Optional[int] = None
-    max_connect_timeout: Optional[int] = None
-    max_read_timeout: Optional[int] = None
-    max_write_timeout: Optional[int] = None
-
-
-class HttpNodeType(CommonNodeType):
-    """HTTP request node type implementation."""
-
-    variables: list[Variable]
-    method: Method
-    url: str
-    headers: str
-    params: str
-    body: Body
-    authorization: Authorization
-    timeout: Timeout
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = HttpNodeType(
-        title="Example HTTP Node",
-        desc="An HTTP request node example",
-        type=BlockEnum.http_request,
-        variables=[Variable(variable="var1", value_selector=["node1", "key1"])],
-        method=Method.get,
-        url="https://api.example.com/data",
-        headers="{}",
-        params="{}",
-        body=Body(type=BodyType.none, data=[]),
-        authorization=Authorization(type=AuthorizationType.none),
-        timeout=Timeout(connect=30, read=30, write=30),
-    )
-    print(example_node)

api/core/auto/node_types/if_else.py

@@ -1,99 +0,0 @@
-from enum import Enum
-from typing import Optional, Union
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, ValueSelector, VarType
-from .tool import VarType as NumberVarType
-
-# Import previously defined CommonNodeType, ValueSelector, Var, and VarType
-# Assume they are defined in the same module
-
-
-class LogicalOperator(str, Enum):
-    and_ = "and"
-    or_ = "or"
-
-
-class ComparisonOperator(str, Enum):
-    contains = "contains"
-    notContains = "not contains"
-    startWith = "start with"
-    endWith = "end with"
-    is_ = "is"
-    isNot = "is not"
-    empty = "empty"
-    notEmpty = "not empty"
-    equal = "="
-    notEqual = "≠"
-    largerThan = ">"
-    lessThan = "<"
-    largerThanOrEqual = "≥"
-    lessThanOrEqual = "≤"
-    isNull = "is null"
-    isNotNull = "is not null"
-    in_ = "in"
-    notIn = "not in"
-    allOf = "all of"
-    exists = "exists"
-    notExists = "not exists"
-    equals = "="  # Alias for equal for compatibility
-
-
-class Condition(BaseModel):
-    id: str
-    varType: VarType
-    variable_selector: Optional[ValueSelector]
-    key: Optional[str] = None  # Sub variable key
-    comparison_operator: Optional[ComparisonOperator] = None
-    value: Union[str, list[str]]
-    numberVarType: Optional[NumberVarType]
-    sub_variable_condition: Optional["CaseItem"] = None  # Recursive reference
-
-
-class CaseItem(BaseModel):
-    case_id: str
-    logical_operator: LogicalOperator
-    conditions: list[Condition]
-
-
-class IfElseNodeType(CommonNodeType):
-    logical_operator: Optional[LogicalOperator] = None
-    conditions: Optional[list[Condition]] = None
-    cases: list[CaseItem]
-    isInIteration: bool
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = IfElseNodeType(
-        title="Example IfElse Node",
-        desc="An if-else node example",
-        type=BlockEnum.if_else,
-        logical_operator=LogicalOperator.and_,
-        conditions=[
-            Condition(
-                id="condition1",
-                varType=VarType.string,
-                variable_selector={"nodeId": "varNode", "key": "value"},
-                comparison_operator=ComparisonOperator.is_,
-                value="exampleValue",
-            )
-        ],
-        cases=[
-            CaseItem(
-                case_id="case1",
-                logical_operator=LogicalOperator.or_,
-                conditions=[
-                    Condition(
-                        id="condition2",
-                        varType=VarType.number,
-                        value="10",
-                        comparison_operator=ComparisonOperator.largerThan,
-                    )
-                ],
-            )
-        ],
-        isInIteration=True,
-    )
-    print(example_node)

api/core/auto/node_types/iteration.py

@@ -1,45 +0,0 @@
-from enum import Enum
-from typing import Optional
-
-from .common import BlockEnum, CommonNodeType, ValueSelector, VarType
-
-
-class ErrorHandleMode(str, Enum):
-    """Error handling modes for iteration."""
-
-    terminated = "terminated"
-    continue_on_error = "continue-on-error"
-    remove_abnormal_output = "remove-abnormal-output"
-
-
-class IterationNodeType(CommonNodeType):
-    """Iteration node type implementation."""
-
-    startNodeType: Optional[BlockEnum] = None
-    start_node_id: str  # Start node ID in the iteration
-    iteration_id: Optional[str] = None
-    iterator_selector: ValueSelector
-    output_selector: ValueSelector
-    output_type: VarType  # Output type
-    is_parallel: bool  # Open the parallel mode or not
-    parallel_nums: int  # The numbers of parallel
-    error_handle_mode: ErrorHandleMode  # How to handle error in the iteration
-    _isShowTips: bool  # Show tips when answer node in parallel mode iteration
-
-
-# 示例用法
-if __name__ == "__main__":
-    example_node = IterationNodeType(
-        title="Example Iteration Node",
-        desc="An iteration node example",
-        type=BlockEnum.iteration,
-        start_node_id="startNode1",
-        iterator_selector=ValueSelector(value=["iteratorNode", "value"]),
-        output_selector=ValueSelector(value=["outputNode", "value"]),
-        output_type=VarType.string,
-        is_parallel=True,
-        parallel_nums=5,
-        error_handle_mode=ErrorHandleMode.continue_on_error,
-        _isShowTips=True,
-    )
-    print(example_node)

api/core/auto/node_types/iteration_start.py

@@ -1,25 +0,0 @@
-from .common import BlockEnum, CommonNodeType
-
-# 引入之前定义的 CommonNodeType
-# 假设它们在同一模块中定义
-
-
-class IterationStartNodeType(CommonNodeType):
-    """
-    Iteration Start node type implementation.
-
-    This node type is used as the starting point within an iteration block.
-    It inherits all properties from CommonNodeType without adding any additional fields.
-    """
-
-    pass  # 仅仅继承 CommonNodeType，无其他字段
-
-
-# 示例用法
-if __name__ == "__main__":
-    example_node = IterationStartNodeType(
-        title="Example Iteration Start Node",
-        desc="An iteration start node example",
-        type=BlockEnum.iteration_start,
-    )
-    print(example_node)

api/core/auto/node_types/knowledge_retrieval.py

@@ -1,115 +0,0 @@
-from enum import Enum
-from typing import Optional
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, ModelConfig, ValueSelector
-
-
-class RetrieveType(str, Enum):
-    """Retrieval mode types."""
-
-    single = "single"
-    multiple = "multiple"
-
-
-class RerankingModeEnum(str, Enum):
-    """Reranking mode types."""
-
-    simple = "simple"
-    advanced = "advanced"
-
-
-class VectorSetting(BaseModel):
-    """Vector weight settings."""
-
-    vector_weight: float
-    embedding_provider_name: str
-    embedding_model_name: str
-
-
-class KeywordSetting(BaseModel):
-    """Keyword weight settings."""
-
-    keyword_weight: float
-
-
-class Weights(BaseModel):
-    """Weight configuration for retrieval."""
-
-    vector_setting: VectorSetting
-    keyword_setting: KeywordSetting
-
-
-class RerankingModel(BaseModel):
-    """Reranking model configuration."""
-
-    provider: str
-    model: str
-
-
-class MultipleRetrievalConfig(BaseModel):
-    """Configuration for multiple retrieval mode."""
-
-    top_k: int
-    score_threshold: Optional[float] = None
-    reranking_model: Optional[RerankingModel] = None
-    reranking_mode: Optional[RerankingModeEnum] = None
-    weights: Optional[Weights] = None
-    reranking_enable: Optional[bool] = None
-
-
-class SingleRetrievalConfig(BaseModel):
-    """Configuration for single retrieval mode."""
-
-    model: ModelConfig
-
-
-class DataSet(BaseModel):
-    """Dataset information."""
-
-    id: str
-    name: str
-    description: Optional[str] = None
-
-
-class KnowledgeRetrievalNodeType(CommonNodeType):
-    """Knowledge retrieval node type implementation."""
-
-    query_variable_selector: ValueSelector
-    dataset_ids: list[str]
-    retrieval_mode: RetrieveType
-    multiple_retrieval_config: Optional[MultipleRetrievalConfig] = None
-    single_retrieval_config: Optional[SingleRetrievalConfig] = None
-    _datasets: Optional[list[DataSet]] = None
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = KnowledgeRetrievalNodeType(
-        title="Example Knowledge Retrieval Node",
-        desc="A knowledge retrieval node example",
-        type=BlockEnum.knowledge_retrieval,
-        query_variable_selector=ValueSelector(value=["queryNode", "query"]),
-        dataset_ids=["dataset1", "dataset2"],
-        retrieval_mode=RetrieveType.multiple,
-        multiple_retrieval_config=MultipleRetrievalConfig(
-            top_k=10,
-            score_threshold=0.5,
-            reranking_model=RerankingModel(provider="example_provider", model="example_model"),
-            reranking_mode=RerankingModeEnum.simple,
-            weights=Weights(
-                vector_setting=VectorSetting(
-                    vector_weight=0.7, embedding_provider_name="provider1", embedding_model_name="model1"
-                ),
-                keyword_setting=KeywordSetting(keyword_weight=0.3),
-            ),
-            reranking_enable=True,
-        ),
-        single_retrieval_config=SingleRetrievalConfig(
-            model=ModelConfig(
-                provider="example_provider", name="example_model", mode="chat", completion_params={"temperature": 0.7}
-            )
-        ),
-    )
-    print(example_node)

api/core/auto/node_types/list_operator.py

@@ -1,73 +0,0 @@
-from enum import Enum
-from typing import Optional, Union
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, ValueSelector, VarType
-
-# Import ComparisonOperator from if_else.py
-from .if_else import ComparisonOperator
-
-
-class OrderBy(str, Enum):
-    ASC = "asc"
-    DESC = "desc"
-
-
-class Limit(BaseModel):
-    enabled: bool
-    size: Optional[int] = None
-
-
-class Condition(BaseModel):
-    key: str
-    comparison_operator: ComparisonOperator
-    value: Union[str, int, list[str]]
-
-
-class FilterBy(BaseModel):
-    enabled: bool
-    conditions: list[Condition]
-
-
-class ExtractBy(BaseModel):
-    enabled: bool
-    serial: Optional[str] = None
-
-
-class OrderByConfig(BaseModel):
-    enabled: bool
-    key: Union[ValueSelector, str]
-    value: OrderBy
-
-
-class ListFilterNodeType(CommonNodeType):
-    """List filter/operator node type implementation."""
-
-    variable: ValueSelector
-    var_type: VarType
-    item_var_type: VarType
-    filter_by: FilterBy
-    extract_by: ExtractBy
-    order_by: OrderByConfig
-    limit: Limit
-
-
-# 示例用法
-if __name__ == "__main__":
-    example_node = ListFilterNodeType(
-        title="Example List Filter Node",
-        desc="A list filter node example",
-        type=BlockEnum.list_operator,  # Fixed: use list_operator instead of list_filter
-        variable=ValueSelector(value=["varNode", "value"]),
-        var_type=VarType.string,
-        item_var_type=VarType.number,
-        filter_by=FilterBy(
-            enabled=True,
-            conditions=[Condition(key="status", comparison_operator=ComparisonOperator.equals, value="active")],
-        ),
-        extract_by=ExtractBy(enabled=True, serial="serial_1"),
-        order_by=OrderByConfig(enabled=True, key="created_at", value=OrderBy.DESC),
-        limit=Limit(enabled=True, size=100),
-    )
-    print(example_node)

api/core/auto/node_types/llm.py

@@ -1,66 +0,0 @@
-from typing import Optional, Union
-
-from pydantic import BaseModel
-
-from .common import (
-    BlockEnum,
-    CommonNodeType,
-    Context,
-    Memory,
-    ModelConfig,
-    PromptItem,
-    Variable,
-    VisionSetting,
-)
-
-
-class PromptConfig(BaseModel):
-    """Configuration for prompt template variables."""
-
-    jinja2_variables: Optional[list[Variable]] = None
-
-
-class VisionConfig(BaseModel):
-    """Configuration for vision settings."""
-
-    enabled: bool = False
-    configs: Optional[VisionSetting] = None
-
-    def dict(self, *args, **kwargs):
-        """自定义序列化方法，确保正确序列化"""
-        result = {"enabled": self.enabled}
-
-        if self.configs:
-            result["configs"] = self.configs.dict()
-
-        return result
-
-
-class LLMNodeType(CommonNodeType):
-    """LLM node type implementation."""
-
-    model: ModelConfig
-    prompt_template: Union[list[PromptItem], PromptItem]
-    prompt_config: Optional[PromptConfig] = None
-    memory: Optional[Memory] = None
-    context: Optional[Context] = Context(enabled=False, variable_selector=None)
-    vision: Optional[VisionConfig] = VisionConfig(enabled=False)
-
-
-# 示例用法
-if __name__ == "__main__":
-    example_node = LLMNodeType(
-        title="Example LLM Node",
-        desc="A LLM node example",
-        type=BlockEnum.llm,
-        model=ModelConfig(provider="zhipuai", name="glm-4-flash", mode="chat", completion_params={"temperature": 0.7}),
-        prompt_template=[
-            PromptItem(
-                id="system-id", role="system", text="你是一个代码工程师，你会根据用户的需求给出用户所需要的函数"
-            ),
-            PromptItem(id="user-id", role="user", text="给出两数相加的python 函数代码，函数名 func 不要添加其他内容"),
-        ],
-        context=Context(enabled=False, variable_selector=None),
-        vision=VisionConfig(enabled=False),
-    )
-    print(example_node)

api/core/auto/node_types/note_node.py

@@ -1,38 +0,0 @@
-from enum import Enum
-
-from .common import BlockEnum, CommonNodeType
-
-# Import previously defined CommonNodeType
-# Assume it is defined in the same module
-
-
-class NoteTheme(str, Enum):
-    blue = "blue"
-    cyan = "cyan"
-    green = "green"
-    yellow = "yellow"
-    pink = "pink"
-    violet = "violet"
-
-
-class NoteNodeType(CommonNodeType):
-    """Custom note node type implementation."""
-
-    text: str
-    theme: NoteTheme
-    author: str
-    showAuthor: bool
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = NoteNodeType(
-        title="Example Note Node",
-        desc="A note node example",
-        type=BlockEnum.custom_note,
-        text="This is a note.",
-        theme=NoteTheme.green,
-        author="John Doe",
-        showAuthor=True,
-    )
-    print(example_node)

api/core/auto/node_types/parameter_extractor.py

@@ -1,85 +0,0 @@
-from enum import Enum
-from typing import Optional
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, Memory, ModelConfig, ValueSelector, VisionSetting
-
-# Import previously defined CommonNodeType, Memory, ModelConfig, ValueSelector, and VisionSetting
-# Assume they are defined in the same module
-
-
-class ParamType(str, Enum):
-    """Parameter types for extraction."""
-
-    string = "string"
-    number = "number"
-    bool = "bool"
-    select = "select"
-    arrayString = "array[string]"
-    arrayNumber = "array[number]"
-    arrayObject = "array[object]"
-
-
-class Param(BaseModel):
-    """Parameter definition for extraction."""
-
-    name: str
-    type: ParamType
-    options: Optional[list[str]] = None
-    description: str
-    required: Optional[bool] = None
-
-
-class ReasoningModeType(str, Enum):
-    """Reasoning mode types for parameter extraction."""
-
-    prompt = "prompt"
-    functionCall = "function_call"
-
-
-class VisionConfig(BaseModel):
-    """Vision configuration."""
-
-    enabled: bool
-    configs: Optional[VisionSetting] = None
-
-
-class ParameterExtractorNodeType(CommonNodeType):
-    """Parameter extractor node type implementation."""
-
-    model: ModelConfig
-    query: ValueSelector
-    reasoning_mode: ReasoningModeType
-    parameters: List[Param]
-    instruction: str
-    memory: Optional[Memory] = None
-    vision: VisionConfig
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = ParameterExtractorNodeType(
-        title="Example Parameter Extractor Node",
-        desc="A parameter extractor node example",
-        type=BlockEnum.parameter_extractor,
-        model=ModelConfig(
-            provider="example_provider", name="example_model", mode="chat", completion_params={"temperature": 0.7}
-        ),
-        query=ValueSelector(value=["queryNode", "value"]),
-        reasoning_mode=ReasoningModeType.prompt,
-        parameters=[
-            Param(name="param1", type=ParamType.string, description="This is a string parameter", required=True),
-            Param(
-                name="param2",
-                type=ParamType.number,
-                options=["1", "2", "3"],
-                description="This is a number parameter",
-                required=False,
-            ),
-        ],
-        instruction="Please extract the parameters from the input.",
-        memory=Memory(window={"enabled": True, "size": 10}, query_prompt_template="Extract parameters from: {{query}}"),
-        vision=VisionConfig(enabled=True, configs={"setting": "example_setting"}),
-    )
-    print(example_node)

api/core/auto/node_types/question_classifier.py

@@ -1,51 +0,0 @@
-from typing import Optional
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, Memory, ModelConfig, ValueSelector, VisionSetting
-
-# Import previously defined CommonNodeType, Memory, ModelConfig, ValueSelector, and VisionSetting
-# Assume they are defined in the same module
-
-
-class Topic(BaseModel):
-    """Topic for classification."""
-
-    id: str
-    name: str
-
-
-class VisionConfig(BaseModel):
-    """Vision configuration."""
-
-    enabled: bool
-    configs: Optional[VisionSetting] = None
-
-
-class QuestionClassifierNodeType(CommonNodeType):
-    """Question classifier node type implementation."""
-
-    query_variable_selector: ValueSelector
-    model: ModelConfig
-    classes: list[Topic]
-    instruction: str
-    memory: Optional[Memory] = None
-    vision: VisionConfig
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = QuestionClassifierNodeType(
-        title="Example Question Classifier Node",
-        desc="A question classifier node example",
-        type=BlockEnum.question_classifier,
-        query_variable_selector=ValueSelector(value=["queryNode", "value"]),
-        model=ModelConfig(
-            provider="example_provider", name="example_model", mode="chat", completion_params={"temperature": 0.7}
-        ),
-        classes=[Topic(id="1", name="Science"), Topic(id="2", name="Mathematics"), Topic(id="3", name="Literature")],
-        instruction="Classify the given question into the appropriate topic.",
-        memory=Memory(window={"enabled": True, "size": 10}, query_prompt_template="Classify this question: {{query}}"),
-        vision=VisionConfig(enabled=True, configs={"setting": "example_setting"}),
-    )
-    print(example_node)

api/core/auto/node_types/start.py

@@ -1,22 +0,0 @@
-from .common import BlockEnum, CommonNodeType, InputVar
-
-# Import previously defined CommonNodeType and InputVar
-# Assume they are defined in the same module
-
-
-class StartNodeType(CommonNodeType):
-    variables: list[InputVar]
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = StartNodeType(
-        title="Example Start Node",
-        desc="A start node example",
-        type=BlockEnum.start,
-        variables=[
-            InputVar(type="text-input", label="Input 1", variable="input1", required=True),
-            InputVar(type="number", label="Input 2", variable="input2", required=True),
-        ],
-    )
-    print(example_node)

api/core/auto/node_types/template_transform.py

@@ -1,26 +0,0 @@
-from .common import BlockEnum, CommonNodeType, Variable
-
-# 引入之前定义的 CommonNodeType 和 Variable
-# 假设它们在同一模块中定义
-
-
-class TemplateTransformNodeType(CommonNodeType):
-    """Template transform node type implementation."""
-
-    variables: list[Variable]
-    template: str
-
-
-# 示例用法
-if __name__ == "__main__":
-    example_node = TemplateTransformNodeType(
-        title="Example Template Transform Node",
-        desc="A template transform node example",
-        type=BlockEnum.template_transform,
-        variables=[
-            Variable(variable="var1", value_selector=["node1", "key1"]),
-            Variable(variable="var2", value_selector=["node2", "key2"]),
-        ],
-        template="Hello, {{ var1 }}! You have {{ var2 }} new messages.",
-    )
-    print(example_node)

api/core/auto/node_types/tool.py

@@ -1,54 +0,0 @@
-from enum import Enum
-from typing import Any, Optional, Union
-
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, ValueSelector
-
-# Import previously defined CommonNodeType and ValueSelector
-# Assume they are defined in the same module
-
-
-class VarType(str, Enum):
-    variable = "variable"
-    constant = "constant"
-    mixed = "mixed"
-
-
-class ToolVarInputs(BaseModel):
-    type: VarType
-    value: Optional[Union[str, ValueSelector, Any]] = None
-
-
-class ToolNodeType(CommonNodeType):
-    """Tool node type implementation."""
-
-    provider_id: str
-    provider_type: Any  # Placeholder for CollectionType
-    provider_name: str
-    tool_name: str
-    tool_label: str
-    tool_parameters: dict[str, ToolVarInputs]
-    tool_configurations: dict[str, Any]
-    output_schema: dict[str, Any]
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = ToolNodeType(
-        title="Example Tool Node",
-        desc="A tool node example",
-        type=BlockEnum.tool,
-        provider_id="12345",
-        provider_type="some_collection_type",  # Placeholder for CollectionType
-        provider_name="Example Provider",
-        tool_name="Example Tool",
-        tool_label="Example Tool Label",
-        tool_parameters={
-            "input1": ToolVarInputs(type=VarType.variable, value="some_value"),
-            "input2": ToolVarInputs(type=VarType.constant, value="constant_value"),
-        },
-        tool_configurations={"config1": "value1", "config2": {"nested": "value2"}},
-        output_schema={"output1": "string", "output2": "number"},
-    )
-    print(example_node.json(indent=2))  # Print as JSON format for viewing

api/core/auto/node_types/variable_assigner.py

@@ -1,56 +0,0 @@
-from pydantic import BaseModel
-
-from .common import BlockEnum, CommonNodeType, ValueSelector, VarType
-
-
-class VarGroupItem(BaseModel):
-    """Variable group item configuration."""
-
-    output_type: VarType
-    variables: list[ValueSelector]
-
-
-class GroupConfig(VarGroupItem):
-    """Group configuration for advanced settings."""
-
-    group_name: str
-    groupId: str
-
-
-class AdvancedSettings(BaseModel):
-    """Advanced settings for variable assigner."""
-
-    group_enabled: bool
-    groups: list[GroupConfig]
-
-
-class VariableAssignerNodeType(CommonNodeType, VarGroupItem):
-    """Variable assigner node type implementation."""
-
-    advanced_settings: AdvancedSettings
-
-    class Config:
-        arbitrary_types_allowed = True
-
-
-# Example usage
-if __name__ == "__main__":
-    example_node = VariableAssignerNodeType(
-        title="Example Variable Assigner Node",
-        desc="A variable assigner node example",
-        type=BlockEnum.variable_assigner,
-        output_type=VarType.string,
-        variables=[ValueSelector(value=["varNode1", "value1"]), ValueSelector(value=["varNode2", "value2"])],
-        advanced_settings=AdvancedSettings(
-            group_enabled=True,
-            groups=[
-                GroupConfig(
-                    group_name="Group 1",
-                    groupId="group1",
-                    output_type=VarType.number,
-                    variables=[ValueSelector(value=["varNode3", "value3"])],
-                )
-            ],
-        ),
-    )
-    print(example_node.json(indent=2))  # Print as JSON format for viewing

api/core/auto/output/emotion_analysis_workflow.yml

@@ -1,239 +0,0 @@
-app:
-  description: ''
-  icon: 🤖
-  icon_background: '#FFEAD5'
-  mode: workflow
-  name: 情绪分析工作流
-  use_icon_as_answer_icon: false
-kind: app
-version: 0.1.2
-workflow:
-  conversation_variables: []
-  environment_variables: []
-  features:
-    file_upload:
-      allowed_file_extensions:
-      - .JPG
-      - .JPEG
-      - .PNG
-      - .GIF
-      - .WEBP
-      - .SVG
-      allowed_file_types:
-      - image
-      allowed_file_upload_methods:
-      - local_file
-      - remote_url
-      enabled: false
-      fileUploadConfig:
-        audio_file_size_limit: 50
-        batch_count_limit: 5
-        file_size_limit: 15
-        image_file_size_limit: 10
-        video_file_size_limit: 100
-      image:
-        enabled: false
-        number_limits: 3
-        transfer_methods:
-        - local_file
-        - remote_url
-      number_limits: 3
-    opening_statement: ''
-    retriever_resource:
-      enabled: true
-    sensitive_word_avoidance:
-      enabled: false
-    speech_to_text:
-      enabled: false
-    suggested_questions: []
-    suggested_questions_after_answer:
-      enabled: false
-    text_to_speech:
-      enabled: false
-      language: ''
-      voice: ''
-  graph:
-    edges:
-    - id: 1740019130520-source-1740019130521-target
-      source: '1740019130520'
-      sourceHandle: source
-      target: '1740019130521'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: start
-        targetType: llm
-    - id: 1740019130521-source-1740019130522-target
-      source: '1740019130521'
-      sourceHandle: source
-      target: '1740019130522'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: llm
-        targetType: code
-    - id: 1740019130522-source-1740019130523-target
-      source: '1740019130522'
-      sourceHandle: source
-      target: '1740019130523'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: code
-        targetType: template-transform
-    - id: 1740019130523-source-1740019130524-target
-      source: '1740019130523'
-      sourceHandle: source
-      target: '1740019130524'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: template-transform
-        targetType: end
-    nodes:
-    - id: '1740019130520'
-      position:
-        x: 80
-        y: 282
-      height: 116
-      width: 244
-      positionAbsolute:
-        x: 80
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 开始节点
-        desc: 开始节点，接收用户输入的文本。
-        type: start
-        variables:
-        - type: text-input
-          label: input_text
-          variable: input_text
-          required: true
-          max_length: 48
-          options: []
-    - id: '1740019130521'
-      position:
-        x: 380
-        y: 282
-      height: 98
-      width: 244
-      positionAbsolute:
-        x: 380
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: LLM节点
-        desc: LLM节点分析文本情绪，识别出积极、消极或中性情绪。
-        type: llm
-        model:
-          provider: zhipuai
-          name: glm-4-flash
-          mode: chat
-          completion_params:
-            temperature: 0.7
-        prompt_template:
-        - id: 1740019130521-system
-          text: 请分析以下文本的情绪，并返回情绪类型（积极、消极或中性）。
-          role: system
-        - id: 1740019130521-user
-          text: 分析此文本的情绪：{{input_text}}
-          role: user
-        context:
-          enabled: false
-          variable_selector: []
-        vision:
-          enabled: false
-    - id: '1740019130522'
-      position:
-        x: 680
-        y: 282
-      height: 54
-      width: 244
-      positionAbsolute:
-        x: 680
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 代码节点
-        desc: 代码节点将根据LLM分析的结果处理情绪类型。
-        type: code
-        variables:
-        - variable: emotion
-          value_selector:
-          - '1740019130521'
-          - emotion
-        code_language: python3
-        code: "def analyze_sentiment(emotion):\n    if emotion == 'positive':\n  \
-          \      return '积极'\n    elif emotion == 'negative':\n        return '消极'\n\
-          \    else:\n        return '中性'\n\nemotion = '{{emotion}}'\nresult = analyze_sentiment(emotion)\n\
-          return {'result': result}"
-        outputs:
-          sentiment_result:
-            type: string
-    - id: '1740019130523'
-      position:
-        x: 980
-        y: 282
-      height: 54
-      width: 244
-      positionAbsolute:
-        x: 980
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 模板节点
-        desc: 模板节点将情绪分析结果格式化输出。
-        type: template-transform
-        variables:
-        - variable: sentiment_result
-          value_selector:
-          - '1740019130522'
-          - sentiment_result
-        template: 文本的情绪分析结果为：{{sentiment_result}}
-    - id: '1740019130524'
-      position:
-        x: 1280
-        y: 282
-      height: 90
-      width: 244
-      positionAbsolute:
-        x: 1280
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 结束节点
-        desc: 结束节点，返回格式化后的情绪分析结果。
-        type: end
-        outputs:
-        - variable: output
-          value_selector:
-          - '1740019130523'
-          - output
-    viewport:
-      x: 92.96659905656679
-      y: 79.13437154762897
-      zoom: 0.9002006986311041

api/core/auto/output/test_workflow.yml

@@ -1,247 +0,0 @@
-app:
-  description: ''
-  icon: 🤖
-  icon_background: '#FFEAD5'
-  mode: workflow
-  name: 计算两个数字之和
-  use_icon_as_answer_icon: false
-kind: app
-version: 0.1.2
-workflow:
-  conversation_variables: []
-  environment_variables: []
-  features:
-    file_upload:
-      allowed_file_extensions:
-      - .JPG
-      - .JPEG
-      - .PNG
-      - .GIF
-      - .WEBP
-      - .SVG
-      allowed_file_types:
-      - image
-      allowed_file_upload_methods:
-      - local_file
-      - remote_url
-      enabled: false
-      fileUploadConfig:
-        audio_file_size_limit: 50
-        batch_count_limit: 5
-        file_size_limit: 15
-        image_file_size_limit: 10
-        video_file_size_limit: 100
-      image:
-        enabled: false
-        number_limits: 3
-        transfer_methods:
-        - local_file
-        - remote_url
-      number_limits: 3
-    opening_statement: ''
-    retriever_resource:
-      enabled: true
-    sensitive_word_avoidance:
-      enabled: false
-    speech_to_text:
-      enabled: false
-    suggested_questions: []
-    suggested_questions_after_answer:
-      enabled: false
-    text_to_speech:
-      enabled: false
-      language: ''
-      voice: ''
-  graph:
-    edges:
-    - id: 1740019130520-source-1740019130521-target
-      source: '1740019130520'
-      sourceHandle: source
-      target: '1740019130521'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: start
-        targetType: llm
-    - id: 1740019130521-source-1740019130522-target
-      source: '1740019130521'
-      sourceHandle: source
-      target: '1740019130522'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: llm
-        targetType: code
-    - id: 1740019130522-source-1740019130523-target
-      source: '1740019130522'
-      sourceHandle: source
-      target: '1740019130523'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: code
-        targetType: template-transform
-    - id: 1740019130523-source-1740019130524-target
-      source: '1740019130523'
-      sourceHandle: source
-      target: '1740019130524'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: template-transform
-        targetType: end
-    nodes:
-    - id: '1740019130520'
-      position:
-        x: 80
-        y: 282
-      height: 116
-      width: 244
-      positionAbsolute:
-        x: 80
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 开始节点
-        desc: 开始节点，接收两个数字输入参数。
-        type: start
-        variables:
-        - type: number
-          label: num1
-          variable: num1
-          required: true
-          max_length: 48
-          options: []
-        - type: number
-          label: num2
-          variable: num2
-          required: true
-          max_length: 48
-          options: []
-    - id: '1740019130521'
-      position:
-        x: 380
-        y: 282
-      height: 98
-      width: 244
-      positionAbsolute:
-        x: 380
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: LLM节点
-        desc: LLM节点，根据输入的两个数字生成计算它们之和的Python函数。
-        type: llm
-        model:
-          provider: openai
-          name: gpt-4
-          mode: chat
-          completion_params:
-            temperature: 0.7
-        prompt_template:
-        - id: 1740019130521-system
-          text: 你是一个Python开发助手，请根据以下输入生成一个计算两数之和的Python函数。
-          role: system
-        - id: 1740019130521-user
-          text: 请为两个数字{{num1}}和{{num2}}生成一个Python函数，计算它们的和。
-          role: user
-        context:
-          enabled: false
-          variable_selector: []
-        vision:
-          enabled: false
-    - id: '1740019130522'
-      position:
-        x: 680
-        y: 282
-      height: 54
-      width: 244
-      positionAbsolute:
-        x: 680
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 代码节点
-        desc: 代码节点，执行LLM生成的Python函数，并计算结果。
-        type: code
-        variables:
-        - variable: num1
-          value_selector:
-          - '1740019130520'
-          - num1
-        - variable: num2
-          value_selector:
-          - '1740019130520'
-          - num2
-        code_language: python3
-        code: "def calculate_sum(num1, num2):\n    return num1 + num2\n\nresult =\
-          \ calculate_sum({{num1}}, {{num2}})\nreturn result"
-        outputs:
-          result:
-            type: number
-    - id: '1740019130523'
-      position:
-        x: 980
-        y: 282
-      height: 54
-      width: 244
-      positionAbsolute:
-        x: 980
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 模板节点
-        desc: 模板节点，将计算结果格式化为输出字符串。
-        type: template-transform
-        variables:
-        - variable: result
-          value_selector:
-          - '1740019130522'
-          - result
-        template: '计算结果为: {{result}}'
-    - id: '1740019130524'
-      position:
-        x: 1280
-        y: 282
-      height: 90
-      width: 244
-      positionAbsolute:
-        x: 1280
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 结束节点
-        desc: 结束节点，输出格式化后的结果。
-        type: end
-        outputs:
-        - variable: output
-          value_selector:
-          - '1740019130523'
-          - output
-    viewport:
-      x: 92.96659905656679
-      y: 79.13437154762897
-      zoom: 0.9002006986311041

api/core/auto/output/text_analysis_workflow.yml

@@ -1,262 +0,0 @@
-app:
-  description: ''
-  icon: 🤖
-  icon_background: '#FFEAD5'
-  mode: workflow
-  name: 文本分析工作流
-  use_icon_as_answer_icon: false
-kind: app
-version: 0.1.2
-workflow:
-  conversation_variables: []
-  environment_variables: []
-  features:
-    file_upload:
-      allowed_file_extensions:
-      - .JPG
-      - .JPEG
-      - .PNG
-      - .GIF
-      - .WEBP
-      - .SVG
-      allowed_file_types:
-      - image
-      allowed_file_upload_methods:
-      - local_file
-      - remote_url
-      enabled: false
-      fileUploadConfig:
-        audio_file_size_limit: 50
-        batch_count_limit: 5
-        file_size_limit: 15
-        image_file_size_limit: 10
-        video_file_size_limit: 100
-      image:
-        enabled: false
-        number_limits: 3
-        transfer_methods:
-        - local_file
-        - remote_url
-      number_limits: 3
-    opening_statement: ''
-    retriever_resource:
-      enabled: true
-    sensitive_word_avoidance:
-      enabled: false
-    speech_to_text:
-      enabled: false
-    suggested_questions: []
-    suggested_questions_after_answer:
-      enabled: false
-    text_to_speech:
-      enabled: false
-      language: ''
-      voice: ''
-  graph:
-    edges:
-    - id: 1740019130520-source-1740019130521-target
-      source: '1740019130520'
-      sourceHandle: source
-      target: '1740019130521'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: start
-        targetType: llm
-    - id: 1740019130520-source-1740019130522-target
-      source: '1740019130520'
-      sourceHandle: source
-      target: '1740019130522'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: start
-        targetType: code
-    - id: 1740019130521-source-1740019130523-target
-      source: '1740019130521'
-      sourceHandle: source
-      target: '1740019130523'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: llm
-        targetType: template-transform
-    - id: 1740019130522-source-1740019130523-target
-      source: '1740019130522'
-      sourceHandle: source
-      target: '1740019130523'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: code
-        targetType: template-transform
-    - id: 1740019130523-source-1740019130524-target
-      source: '1740019130523'
-      sourceHandle: source
-      target: '1740019130524'
-      targetHandle: target
-      type: custom
-      zIndex: 0
-      data:
-        isInIteration: false
-        sourceType: template-transform
-        targetType: end
-    nodes:
-    - id: '1740019130520'
-      position:
-        x: 80
-        y: 282
-      height: 116
-      width: 244
-      positionAbsolute:
-        x: 80
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 开始节点
-        desc: 接收用户输入的文本参数
-        type: start
-        variables:
-        - type: text-input
-          label: user_text
-          variable: user_text
-          required: true
-          max_length: 48
-          options: []
-    - id: '1740019130521'
-      position:
-        x: 380
-        y: 282
-      height: 98
-      width: 244
-      positionAbsolute:
-        x: 380
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: LLM节点
-        desc: 使用大语言模型进行情感分析，返回文本的情感结果
-        type: llm
-        model:
-          provider: zhipuai
-          name: glm-4-flash
-          mode: chat
-          completion_params:
-            temperature: 0.7
-        prompt_template:
-        - id: 1740019130521-system
-          text: 请分析以下文本的情感，返回积极、消极或中性
-          role: system
-        - id: 1740019130521-user
-          text: '{{user_text}}'
-          role: user
-        context:
-          enabled: false
-          variable_selector: []
-        vision:
-          enabled: false
-    - id: '1740019130522'
-      position:
-        x: 680
-        y: 282
-      height: 54
-      width: 244
-      positionAbsolute:
-        x: 680
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 代码节点
-        desc: 计算文本的统计信息，包括字符数、单词数和句子数
-        type: code
-        variables:
-        - variable: text_for_analysis
-          value_selector:
-          - '1740019130520'
-          - user_text
-        code_language: python3
-        code: "import re\n\ndef main(text):\n    char_count = len(text)\n    word_count\
-          \ = len(text.split())\n    sentence_count = len(re.findall(r'[.!?]', text))\n\
-          \    return {'char_count': char_count, 'word_count': word_count, 'sentence_count':\
-          \ sentence_count}"
-        outputs:
-          text_statistics:
-            type: object
-    - id: '1740019130523'
-      position:
-        x: 980
-        y: 282
-      height: 54
-      width: 244
-      positionAbsolute:
-        x: 980
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 模板节点
-        desc: 将情感分析结果和统计信息组合成格式化报告
-        type: template-transform
-        variables:
-        - variable: sentiment_result
-          value_selector:
-          - '1740019130521'
-          - sentiment_result
-        - variable: text_statistics
-          value_selector:
-          - '1740019130522'
-          - text_statistics
-        template: '情感分析结果：{{sentiment_result}}
-
-          文本统计信息：
-
-          字符数：{{text_statistics.char_count}}
-
-          单词数：{{text_statistics.word_count}}
-
-          句子数：{{text_statistics.sentence_count}}'
-    - id: '1740019130524'
-      position:
-        x: 1280
-        y: 282
-      height: 90
-      width: 244
-      positionAbsolute:
-        x: 1280
-        y: 282
-      selected: false
-      sourcePosition: right
-      targetPosition: left
-      type: custom
-      data:
-        title: 结束节点
-        desc: 返回最终的分析报告
-        type: end
-        outputs:
-        - variable: final_report
-          value_selector:
-          - '1740019130523'
-          - output
-    viewport:
-      x: 92.96659905656679
-      y: 79.13437154762897
-      zoom: 0.9002006986311041

api/core/auto/workflow_generator/__init__.py

@@ -1,8 +0,0 @@
-"""
-工作流生成器包
-用于根据用户需求生成Dify工作流
-"""
-
-from .workflow_generator import WorkflowGenerator
-
-__all__ = ["WorkflowGenerator"]

api/core/auto/workflow_generator/generators/__init__.py

@@ -1,9 +0,0 @@
-"""
-节点和边生成器包
-"""
-
-from .edge_generator import EdgeGenerator
-from .layout_engine import LayoutEngine
-from .node_generator import NodeGenerator
-
-__all__ = ["EdgeGenerator", "LayoutEngine", "NodeGenerator"]

api/core/auto/workflow_generator/generators/edge_generator.py

@@ -1,101 +0,0 @@
-"""
-Edge Generator
-Used to generate edges in the workflow
-"""
-
-from core.auto.node_types.common import CommonEdgeType, CompleteEdge, CompleteNode
-from core.auto.workflow_generator.models.workflow_description import ConnectionDescription
-
-
-class EdgeGenerator:
-    """Edge generator for creating workflow edges"""
-
-    @staticmethod
-    def create_edges(nodes: list[CompleteNode], connections: list[ConnectionDescription]) -> list[CompleteEdge]:
-        """
-        Create edges based on nodes and connection information
-
-        Args:
-            nodes: list of nodes
-            connections: list of connection descriptions
-
-        Returns:
-            list of edges
-        """
-        edges = []
-
-        # If connection information is provided, create edges based on it
-        if connections:
-            for connection in connections:
-                source_id = connection.source
-                target_id = connection.target
-
-                if not source_id or not target_id:
-                    continue
-
-                # Find source and target nodes
-                source_node = next((node for node in nodes if node.id == source_id), None)
-                target_node = next((node for node in nodes if node.id == target_id), None)
-
-                if not source_node or not target_node:
-                    continue
-
-                # Get node types
-                source_type = source_node.data.type
-                target_type = target_node.data.type
-
-                # Create edge
-                edge_id = f"{source_id}-source-{target_id}-target"
-
-                # Create edge data
-                edge_data = CommonEdgeType(isInIteration=False, sourceType=source_type, targetType=target_type)
-
-                # Create complete edge
-                edge = CompleteEdge(
-                    id=edge_id,
-                    source=source_id,
-                    sourceHandle="source",
-                    target=target_id,
-                    targetHandle="target",
-                    type="custom",
-                    zIndex=0,
-                )
-
-                # Add edge data
-                edge.add_data(edge_data)
-
-                edges.append(edge)
-        # If no connection information is provided, automatically create edges
-        else:
-            # Create edges based on node order
-            for i in range(len(nodes) - 1):
-                source_node = nodes[i]
-                target_node = nodes[i + 1]
-
-                # Get node types
-                source_type = source_node.data.type
-                target_type = target_node.data.type
-
-                # Create edge
-                edge_id = f"{source_node.id}-source-{target_node.id}-target"
-
-                # Create edge data
-                edge_data = CommonEdgeType(isInIteration=False, sourceType=source_type, targetType=target_type)
-
-                # Create complete edge
-                edge = CompleteEdge(
-                    id=edge_id,
-                    source=source_node.id,
-                    sourceHandle="source",
-                    target=target_node.id,
-                    targetHandle="target",
-                    type="custom",
-                    zIndex=0,
-                )
-
-                # Add edge data
-                edge.add_data(edge_data)
-
-                edges.append(edge)
-
-        return edges

api/core/auto/workflow_generator/generators/layout_engine.py

@@ -1,77 +0,0 @@
-"""
-Layout Engine
-Used to arrange the positions of nodes and edges
-"""
-
-from core.auto.node_types.common import CompleteEdge, CompleteNode
-
-
-class LayoutEngine:
-    """Layout engine"""
-
-    @staticmethod
-    def apply_layout(nodes: list[CompleteNode]) -> None:
-        """
-        Apply layout, arranging nodes in a row
-
-        Args:
-            nodes: list of nodes
-        """
-        # Simple linear layout, arranging nodes from left to right
-        x_position = 80
-        y_position = 282
-
-        for node in nodes:
-            node.position = {"x": x_position, "y": y_position}
-            node.positionAbsolute = {"x": x_position, "y": y_position}
-
-            # Update position for the next node
-            x_position += 300  # Horizontal spacing between nodes
-
-    @staticmethod
-    def apply_topological_layout(nodes: list[CompleteNode], edges: list[CompleteEdge]) -> None:
-        """
-        Apply topological sort layout, arranging nodes based on their dependencies
-
-        Args:
-            nodes: list of nodes
-            edges: list of edges
-        """
-        # Create mapping from node ID to node
-        node_map = {node.id: node for node in nodes}
-
-        # Create adjacency list
-        adjacency_list = {node.id: [] for node in nodes}
-        for edge in edges:
-            adjacency_list[edge.source].append(edge.target)
-
-        # Create in-degree table
-        in_degree = {node.id: 0 for node in nodes}
-        for source, targets in adjacency_list.items():
-            for target in targets:
-                in_degree[target] += 1
-
-        # Topological sort
-        queue = [node_id for node_id, degree in in_degree.items() if degree == 0]
-        sorted_nodes = []
-
-        while queue:
-            current = queue.pop(0)
-            sorted_nodes.append(current)
-
-            for neighbor in adjacency_list[current]:
-                in_degree[neighbor] -= 1
-                if in_degree[neighbor] == 0:
-                    queue.append(neighbor)
-
-        # Apply layout
-        x_position = 80
-        y_position = 282
-
-        for node_id in sorted_nodes:
-            node = node_map[node_id]
-            node.position = {"x": x_position, "y": y_position}
-            node.positionAbsolute = {"x": x_position, "y": y_position}
-
-            # Update position for the next node
-            x_position += 300  # Horizontal spacing between nodes

api/core/auto/workflow_generator/generators/node_generator.py

@@ -1,446 +0,0 @@
-"""
-Node Generator
-Generate nodes based on workflow description
-"""
-
-from core.auto.node_types.code import CodeLanguage, CodeNodeType, OutputVar
-from core.auto.node_types.common import (
-    BlockEnum,
-    CompleteNode,
-    Context,
-    InputVar,
-    ModelConfig,
-    PromptItem,
-    PromptRole,
-    ValueSelector,
-    Variable,
-)
-from core.auto.node_types.end import EndNodeType
-from core.auto.node_types.llm import LLMNodeType, VisionConfig
-from core.auto.node_types.start import StartNodeType
-from core.auto.node_types.template_transform import TemplateTransformNodeType
-from core.auto.workflow_generator.models.workflow_description import NodeDescription
-from core.auto.workflow_generator.utils.prompts import DEFAULT_MODEL_CONFIG, DEFAULT_SYSTEM_PROMPT
-from core.auto.workflow_generator.utils.type_mapper import map_string_to_var_type, map_var_type_to_input_type
-
-
-class NodeGenerator:
-    """Node generator for creating workflow nodes"""
-
-    @staticmethod
-    def create_nodes(node_descriptions: list[NodeDescription]) -> list[CompleteNode]:
-        """
-        Create nodes based on node descriptions
-
-        Args:
-            node_descriptions: list of node descriptions
-
-        Returns:
-            list of nodes
-        """
-        nodes = []
-
-        for node_desc in node_descriptions:
-            node_type = node_desc.type
-
-            if node_type == "start":
-                node = NodeGenerator._create_start_node(node_desc)
-            elif node_type == "llm":
-                node = NodeGenerator._create_llm_node(node_desc)
-            elif node_type == "code":
-                node = NodeGenerator._create_code_node(node_desc)
-            elif node_type == "template":
-                node = NodeGenerator._create_template_node(node_desc)
-            elif node_type == "end":
-                node = NodeGenerator._create_end_node(node_desc)
-            else:
-                raise ValueError(f"Unsupported node type: {node_type}")
-
-            nodes.append(node)
-
-        return nodes
-
-    @staticmethod
-    def _create_start_node(node_desc: NodeDescription) -> CompleteNode:
-        """Create start node"""
-        variables = []
-
-        for var in node_desc.variables or []:
-            input_var = InputVar(
-                type=map_var_type_to_input_type(var.type),
-                label=var.name,
-                variable=var.name,
-                required=var.required,
-                max_length=48,
-                options=[],
-            )
-            variables.append(input_var)
-
-        start_node = StartNodeType(
-            title=node_desc.title, desc=node_desc.description or "", type=BlockEnum.start, variables=variables
-        )
-
-        return CompleteNode(
-            id=node_desc.id,
-            type="custom",
-            position={"x": 0, "y": 0},  # Temporary position, will be updated later
-            height=118,  # Increase height to match reference file
-            width=244,
-            positionAbsolute={"x": 0, "y": 0},
-            selected=False,
-            sourcePosition="right",
-            targetPosition="left",
-            data=start_node,
-        )
-
-    @staticmethod
-    def _create_llm_node(node_desc: NodeDescription) -> CompleteNode:
-        """Create LLM node"""
-        # Build prompt template
-        prompt_template = []
-
-        # Add system prompt
-        system_prompt = node_desc.system_prompt or DEFAULT_SYSTEM_PROMPT
-        prompt_template.append(PromptItem(id=f"{node_desc.id}-system", role=PromptRole.system, text=system_prompt))
-
-        # Add user prompt
-        user_prompt = node_desc.user_prompt or "Please answer these questions:"
-
-        # Build variable list
-        variables = []
-        for var in node_desc.variables or []:
-            source_node = var.source_node or ""
-            source_variable = var.source_variable or ""
-
-            print(
-                f"DEBUG: Processing variable {var.name}, source_node={source_node}, source_variable={source_variable}"
-            )
-
-            # If source node is an LLM node, ensure source_variable is 'text'
-            if source_node:
-                # Check if the source node is an LLM node by checking connections
-                # This is a simple heuristic - if the source node is connected to a node with 'llm' in its ID
-                # or if the source node has 'llm' in its ID, assume it's an LLM node
-                if "llm" in source_node.lower():
-                    print(f"DEBUG: Found LLM node {source_node}")
-                    if source_variable != "text":
-                        old_var = source_variable
-                        source_variable = "text"  # LLM nodes output variable is always 'text'
-                        print(
-                            f"Auto-fixing: Changed source variable from '{old_var}' to 'text' for LLM node {source_node}"  # noqa: E501
-                        )
-
-            # Check if the user prompt already contains correctly formatted variable references
-            # Variable references in LLM nodes should be in the format {{#nodeID.variableName#}}
-            correct_format = f"{{{{#{source_node}.{source_variable}#}}}}"
-            simple_format = f"{{{{{var.name}}}}}"
-
-            # If simple format is used in the prompt, replace it with the correct format
-            if simple_format in user_prompt and source_node and source_variable:
-                user_prompt = user_prompt.replace(simple_format, correct_format)
-
-            variable = Variable(variable=var.name, value_selector=[source_node, source_variable])
-            variables.append(variable)
-
-        # Update user prompt
-        prompt_template.append(PromptItem(id=f"{node_desc.id}-user", role=PromptRole.user, text=user_prompt))
-
-        # Use default model configuration, prioritize configuration in node description
-        provider = node_desc.provider or DEFAULT_MODEL_CONFIG["provider"]
-        model = node_desc.model or DEFAULT_MODEL_CONFIG["model"]
-
-        llm_node = LLMNodeType(
-            title=node_desc.title,
-            desc=node_desc.description or "",
-            type=BlockEnum.llm,
-            model=ModelConfig(
-                provider=provider,
-                name=model,
-                mode=DEFAULT_MODEL_CONFIG["mode"],
-                completion_params=DEFAULT_MODEL_CONFIG["completion_params"],
-            ),
-            prompt_template=prompt_template,
-            variables=variables,
-            context=Context(enabled=False, variable_selector=ValueSelector(value=[])),
-            vision=VisionConfig(enabled=False),
-        )
-
-        return CompleteNode(
-            id=node_desc.id,
-            type="custom",
-            position={"x": 0, "y": 0},  # Temporary position, will be updated later
-            height=126,  # Increase height to match reference file
-            width=244,
-            positionAbsolute={"x": 0, "y": 0},
-            selected=False,
-            sourcePosition="right",
-            targetPosition="left",
-            data=llm_node,
-        )
-
-    @staticmethod
-    def _create_code_node(node_desc: NodeDescription) -> CompleteNode:
-        """Create code node"""
-        # Build variable list and function parameter names
-        variables = []
-        var_names = []
-        var_mapping = {}  # Used to store mapping from variable names to function parameter names
-
-        # First, identify all LLM nodes in the workflow
-        llm_nodes = set()
-        for connection in node_desc.workflow_description.connections:
-            for node in node_desc.workflow_description.nodes:
-                if node.id == connection.source and node.type.lower() == "llm":
-                    llm_nodes.add(node.id)
-
-        for var in node_desc.variables or []:
-            source_node = var.source_node or ""
-            source_variable = var.source_variable or ""
-
-            # Check if source node is an LLM node and warn if source_variable is not 'text'
-            if source_node in llm_nodes and source_variable != "text":
-                print(
-                    f"WARNING: LLM node {source_node} output variable should be 'text', but got '{source_variable}'. This may cause issues in Dify."  # noqa: E501
-                )
-                print("         Consider changing the source_variable to 'text' in your workflow description.")
-                # Auto-fix: Always use 'text' as the source variable for LLM nodes
-                old_var = source_variable
-                source_variable = "text"
-                print(f"Auto-fixing: Changed source variable from '{old_var}' to 'text' for LLM node {source_node}")
-            elif source_node and "llm" in source_node.lower() and source_variable != "text":
-                # Fallback heuristic check based on node ID
-                print(
-                    f"WARNING: Node {source_node} appears to be an LLM node based on its ID, but source_variable is not 'text'."  # noqa: E501
-                )
-                print("         Consider changing the source_variable to 'text' in your workflow description.")
-                # Auto-fix: Always use 'text' as the source variable for LLM nodes
-                old_var = source_variable
-                source_variable = "text"
-                print(f"Auto-fixing: Changed source variable from '{old_var}' to 'text' for LLM node {source_node}")
-
-            # Use variable name as function parameter name
-            variable_name = var.name  # Variable name defined in this node
-            param_name = variable_name  # Function parameter name must match variable name
-
-            # Validate variable name format
-            if not variable_name.replace("_", "").isalnum():
-                raise ValueError(
-                    f"Invalid variable name: {variable_name}. Variable names must only contain letters, numbers, and underscores."  # noqa: E501
-                )
-            if not variable_name[0].isalpha() and variable_name[0] != "_":
-                raise ValueError(
-                    f"Invalid variable name: {variable_name}. Variable names must start with a letter or underscore."
-                )
-
-            var_names.append(param_name)
-            var_mapping[variable_name] = param_name
-
-            variable = Variable(variable=variable_name, value_selector=[source_node, source_variable])
-            variables.append(variable)
-
-        # Build output
-        outputs = {}
-        for output in node_desc.outputs or []:
-            # Validate output variable name format
-            if not output.name.replace("_", "").isalnum():
-                raise ValueError(
-                    f"Invalid output variable name: {output.name}. Output names must only contain letters, numbers, and underscores."  # noqa: E501
-                )
-            if not output.name[0].isalpha() and output.name[0] != "_":
-                raise ValueError(
-                    f"Invalid output variable name: {output.name}. Output names must start with a letter or underscore."
-                )
-
-            outputs[output.name] = OutputVar(type=map_string_to_var_type(output.type))
-
-        # Generate code, ensure function parameters match variable names, return values match output names
-        output_names = [output.name for output in node_desc.outputs or []]
-
-        # Build function parameter list
-        params_str = ", ".join(var_names) if var_names else ""
-
-        # Build return value dictionary
-        return_dict = {}
-        for output_name in output_names:
-            # Use the first variable as the return value by default
-            return_dict[output_name] = var_names[0] if var_names else f'"{output_name}"'
-
-        return_dict_str = ", ".join([f'"{k}": {v}' for k, v in return_dict.items()])
-
-        # Default code template, ensure return dictionary type matches output variable
-        default_code = f"""def main({params_str}):
-    # Write your code here
-    # Process input variables
-    
-    # Return a dictionary, key names must match variable names defined in outputs
-    return {{{return_dict_str}}}"""
-
-        # If custom code is provided, ensure it meets the specifications
-        if node_desc.code:
-            custom_code = node_desc.code
-            # Check if it contains main function definition
-            if not custom_code.strip().startswith("def main("):
-                # Try to fix the code by adding main function with correct parameters
-                custom_code = f"def main({params_str}):\n" + custom_code.strip()
-            else:
-                # Extract function parameters from the existing main function
-                import re
-
-                func_params = re.search(r"def\s+main\s*\((.*?)\)", custom_code)
-                if func_params:
-                    existing_params = [p.strip() for p in func_params.group(1).split(",") if p.strip()]
-                    # Verify that all required parameters are present
-                    missing_params = set(var_names) - set(existing_params)
-                    if missing_params:
-                        # 尝试修复代码，将函数参数替换为正确的参数名
-                        old_params = func_params.group(1)
-                        new_params = params_str
-                        custom_code = custom_code.replace(f"def main({old_params})", f"def main({new_params})")
-                        print(
-                            f"Warning: Fixed missing parameters in code node: {', '.join(missing_params)}. Function parameters must match variable names defined in this node."  # noqa: E501
-                        )
-
-            # Check if the return value is a dictionary and keys match output variables
-            for output_name in output_names:
-                if f'"{output_name}"' not in custom_code and f"'{output_name}'" not in custom_code:
-                    # Code may not meet specifications, use default code
-                    custom_code = default_code
-                    break
-
-            # Use fixed code
-            code = custom_code
-        else:
-            code = default_code
-
-        code_node = CodeNodeType(
-            title=node_desc.title,
-            desc=node_desc.description or "",
-            type=BlockEnum.code,
-            code_language=CodeLanguage.python3,
-            code=code,
-            variables=variables,
-            outputs=outputs,
-        )
-
-        return CompleteNode(
-            id=node_desc.id,
-            type="custom",
-            position={"x": 0, "y": 0},  # Temporary position, will be updated later
-            height=82,  # Increase height to match reference file
-            width=244,
-            positionAbsolute={"x": 0, "y": 0},
-            selected=False,
-            sourcePosition="right",
-            targetPosition="left",
-            data=code_node,
-        )
-
-    @staticmethod
-    def _create_template_node(node_desc: NodeDescription) -> CompleteNode:
-        """Create template node"""
-        # Build variable list
-        variables = []
-        template_text = node_desc.template or ""
-
-        # Collect all node IDs referenced in the template
-        referenced_nodes = set()
-        for var in node_desc.variables or []:
-            source_node = var.source_node or ""
-            source_variable = var.source_variable or ""
-
-            variable = Variable(variable=var.name, value_selector=[source_node, source_variable])
-            variables.append(variable)
-
-            if source_node:
-                referenced_nodes.add(source_node)
-
-            # Modify variable reference format in the template
-            # Replace {{#node_id.variable#}} with {{ variable }}
-            if source_node and source_variable:
-                template_text = template_text.replace(f"{{{{#{source_node}.{source_variable}#}}}}", f"{{ {var.name} }}")
-
-        # Check if a reference to the start node needs to be added
-        # If the template contains a reference to the start node but the variable list does not have a corresponding variable  # noqa: E501
-        import re
-
-        start_node_refs = re.findall(r"{{#(\d+)\.([^#]+)#}}", template_text)
-
-        for node_id, var_name in start_node_refs:
-            # Check if there is already a reference to this variable
-            if not any(v.variable == var_name for v in variables):
-                # Add reference to start node variable
-                variable = Variable(variable=var_name, value_selector=[node_id, var_name])
-                variables.append(variable)
-
-                # Modify variable reference format in the template
-                template_text = template_text.replace(f"{{{{#{node_id}.{var_name}#}}}}", f"{{ {var_name} }}")
-
-        # Get all variable names
-        var_names = [var.variable for var in variables]
-
-        # Simple and crude method: directly replace all possible variable reference formats
-        for var_name in var_names:
-            # Replace {var_name} with {{ var_name }}
-            template_text = template_text.replace("{" + var_name + "}", "{{ " + var_name + " }}")
-            # Replace { var_name } with {{ var_name }}
-            template_text = template_text.replace("{ " + var_name + " }", "{{ " + var_name + " }}")
-            # Replace {var_name } with {{ var_name }}
-            template_text = template_text.replace("{" + var_name + " }", "{{ " + var_name + " }}")
-            # Replace { var_name} with {{ var_name }}
-            template_text = template_text.replace("{ " + var_name + "}", "{{ " + var_name + " }}")
-            # Replace {{{ var_name }}} with {{ var_name }}
-            template_text = template_text.replace("{{{ " + var_name + " }}}", "{{ " + var_name + " }}")
-            # Replace {{{var_name}}} with {{ var_name }}
-            template_text = template_text.replace("{{{" + var_name + "}}}", "{{ " + var_name + " }}")
-
-        # Use regular expression to replace all triple curly braces with double curly braces
-        template_text = re.sub(r"{{{([^}]+)}}}", r"{{ \1 }}", template_text)
-
-        template_node = TemplateTransformNodeType(
-            title=node_desc.title,
-            desc=node_desc.description or "",
-            type=BlockEnum.template_transform,
-            template=template_text,
-            variables=variables,
-        )
-
-        return CompleteNode(
-            id=node_desc.id,
-            type="custom",
-            position={"x": 0, "y": 0},  # Temporary position, will be updated later
-            height=82,  # Increase height to match reference file
-            width=244,
-            positionAbsolute={"x": 0, "y": 0},
-            selected=False,
-            sourcePosition="right",
-            targetPosition="left",
-            data=template_node,
-        )
-
-    @staticmethod
-    def _create_end_node(node_desc: NodeDescription) -> CompleteNode:
-        """Create end node"""
-        # Build output variable list
-        outputs = []
-        for output in node_desc.outputs or []:
-            variable = Variable(
-                variable=output.name, value_selector=[output.source_node or "", output.source_variable or ""]
-            )
-            outputs.append(variable)
-
-        end_node = EndNodeType(
-            title=node_desc.title, desc=node_desc.description or "", type=BlockEnum.end, outputs=outputs
-        )
-
-        return CompleteNode(
-            id=node_desc.id,
-            type="custom",
-            position={"x": 0, "y": 0},  # Temporary position, will be updated later
-            height=90,
-            width=244,
-            positionAbsolute={"x": 0, "y": 0},
-            selected=False,
-            sourcePosition="right",
-            targetPosition="left",
-            data=end_node,
-        )

api/core/auto/workflow_generator/models/__init__.py

@@ -1,7 +0,0 @@
-"""
-模型包
-"""
-
-from .workflow_description import ConnectionDescription, NodeDescription, WorkflowDescription
-
-__all__ = ["ConnectionDescription", "NodeDescription", "WorkflowDescription"]

api/core/auto/workflow_generator/models/workflow_description.py

@@ -1,80 +0,0 @@
-"""
-Workflow Description Model
-Used to represent the simplified workflow description generated by large language models
-"""
-
-from typing import Optional
-
-from pydantic import BaseModel, Field
-
-
-class VariableDescription(BaseModel):
-    """Variable description"""
-
-    name: str
-    type: str
-    description: Optional[str] = None
-    required: bool = True
-    source_node: Optional[str] = None
-    source_variable: Optional[str] = None
-
-
-class OutputDescription(BaseModel):
-    """Output description"""
-
-    name: str
-    type: str = "string"
-    description: Optional[str] = None
-    source_node: Optional[str] = None
-    source_variable: Optional[str] = None
-
-
-class NodeDescription(BaseModel):
-    """Node description"""
-
-    id: str
-    type: str
-    title: str
-    description: Optional[str] = ""
-    variables: Optional[list[VariableDescription]] = Field(default_factory=list)
-    outputs: Optional[list[OutputDescription]] = Field(default_factory=list)
-
-    # LLM node specific fields
-    system_prompt: Optional[str] = None
-    user_prompt: Optional[str] = None
-    provider: Optional[str] = "zhipuai"
-    model: Optional[str] = "glm-4-flash"
-
-    # Code node specific fields
-    code: Optional[str] = None
-
-    # Template node specific fields
-    template: Optional[str] = None
-
-    # Reference to workflow description, used for node relationship analysis
-    workflow_description: Optional["WorkflowDescription"] = Field(default=None, exclude=True)
-
-    class Config:
-        exclude = {"workflow_description"}
-
-
-class ConnectionDescription(BaseModel):
-    """Connection description"""
-
-    source: str
-    target: str
-
-
-class WorkflowDescription(BaseModel):
-    """Workflow description"""
-
-    name: str
-    description: Optional[str] = ""
-    nodes: list[NodeDescription]
-    connections: list[ConnectionDescription]
-
-    def __init__(self, **data):
-        super().__init__(**data)
-        # Add workflow description reference to each node
-        for node in self.nodes:
-            node.workflow_description = self

api/core/auto/workflow_generator/utils/__init__.py

@@ -1,16 +0,0 @@
-"""
-工具函数包
-"""
-
-from .llm_client import LLMClient
-from .prompts import DEFAULT_MODEL_CONFIG, DEFAULT_SYSTEM_PROMPT, build_workflow_prompt
-from .type_mapper import map_string_to_var_type, map_var_type_to_input_type
-
-__all__ = [
-    "DEFAULT_MODEL_CONFIG",
-    "DEFAULT_SYSTEM_PROMPT",
-    "LLMClient",
-    "build_workflow_prompt",
-    "map_string_to_var_type",
-    "map_var_type_to_input_type",
-]

api/core/auto/workflow_generator/utils/config_manager.py

@@ -1,142 +0,0 @@
-"""
-Configuration Manager
-Used to manage configurations and prompts
-"""
-
-import os
-import time
-from pathlib import Path
-from typing import Any, Optional
-
-import yaml
-
-
-class ConfigManager:
-    """Configuration manager for managing configurations"""
-
-    def __init__(self, config_dir: str = "config"):
-        """
-        Initialize configuration manager
-
-        Args:
-            config_dir: Configuration directory path
-        """
-        self.config_dir = Path(config_dir)
-        self.config: dict[str, Any] = {}
-        self.last_load_time: float = 0
-        self.reload_interval: float = 60  # Check every 60 seconds
-        self._load_config()
-
-    def _should_reload(self) -> bool:
-        """Check if configuration needs to be reloaded"""
-        return time.time() - self.last_load_time > self.reload_interval
-
-    def _load_config(self) -> dict[str, Any]:
-        """Load configuration files"""
-        default_config = self._load_yaml(self.config_dir / "default.yaml")
-        custom_config = self._load_yaml(self.config_dir / "custom.yaml")
-        self.config = self._deep_merge(default_config, custom_config)
-        self.last_load_time = time.time()
-        return self.config
-
-    @staticmethod
-    def _load_yaml(path: Path) -> dict[str, Any]:
-        """Load YAML file"""
-        try:
-            with open(path, encoding="utf-8") as f:
-                return yaml.safe_load(f) or {}
-        except FileNotFoundError:
-            print(f"Warning: Config file not found at {path}")
-            return {}
-        except Exception as e:
-            print(f"Error loading config file {path}: {e}")
-            return {}
-
-    @staticmethod
-    def _deep_merge(dict1: dict, dict2: dict) -> dict:
-        """Recursively merge two dictionaries"""
-        result = dict1.copy()
-        for key, value in dict2.items():
-            if key in result and isinstance(result[key], dict) and isinstance(value, dict):
-                result[key] = ConfigManager._deep_merge(result[key], value)
-            else:
-                result[key] = value
-        return result
-
-    def get(self, *keys: str, default: Any = None) -> Any:
-        """
-        Get configuration value
-
-        Args:
-            *keys: Configuration key path
-            default: Default value
-
-        Returns:
-            Configuration value or default value
-        """
-        if self._should_reload():
-            self._load_config()
-
-        current = self.config
-        for key in keys:
-            if isinstance(current, dict) and key in current:
-                current = current[key]
-            else:
-                return default
-        return current
-
-    @property
-    def workflow_generator(self) -> dict[str, Any]:
-        """Get workflow generator configuration"""
-        return self.get("workflow_generator", default={})
-
-    @property
-    def workflow_nodes(self) -> dict[str, Any]:
-        """Get workflow nodes configuration"""
-        return self.get("workflow_nodes", default={})
-
-    @property
-    def output(self) -> dict[str, Any]:
-        """Get output configuration"""
-        return self.get("output", default={})
-
-    def get_output_path(self, filename: Optional[str] = None) -> str:
-        """
-        Get output file path
-
-        Args:
-            filename: Optional filename, uses default filename from config if not specified
-
-        Returns:
-            Complete output file path
-        """
-        output_config = self.output
-        output_dir = output_config.get("dir", "output/")
-        output_filename = filename or output_config.get("filename", "generated_workflow.yml")
-        return os.path.join(output_dir, output_filename)
-
-    def get_workflow_model(self, model_name: Optional[str] = None) -> dict[str, Any]:
-        """
-        Get workflow generation model configuration
-
-        Args:
-            model_name: Model name, uses default model if not specified
-
-        Returns:
-            Model configuration
-        """
-        models = self.workflow_generator.get("models", {})
-
-        if not model_name:
-            model_name = models.get("default")
-
-        return models.get("available", {}).get(model_name, {})
-
-    def get_llm_node_config(self) -> dict[str, Any]:
-        """
-        Get LLM node configuration
-
-        Returns:
-            LLM node configuration
-        """
-        return self.workflow_nodes.get("llm", {})

api/core/auto/workflow_generator/utils/debug_manager.py

@@ -1,151 +0,0 @@
-"""
-Debug Manager
-Used to manage debug information saving
-"""
-
-import datetime
-import json
-import os
-import uuid
-from typing import Any, Optional, Union
-
-
-class DebugManager:
-    """Debug manager for managing debug information saving"""
-
-    _instance = None
-
-    def __new__(cls, *args, **kwargs):
-        """Singleton pattern"""
-        if cls._instance is None:
-            cls._instance = super().__new__(cls)
-            cls._instance._initialized = False
-        return cls._instance
-
-    def __init__(self, config: dict[str, Any] = {}, debug_enabled: bool = False):
-        """
-        Initialize debug manager
-
-        Args:
-            config: Debug configuration
-            debug_enabled: Whether to enable debug mode
-        """
-        # Avoid repeated initialization
-        if self._initialized:
-            return
-
-        self._initialized = True
-        self.config = config or {}
-        self.debug_enabled = debug_enabled or self.config.get("enabled", False)
-        self.debug_dir = self.config.get("dir", "debug/")
-        self.save_options = self.config.get(
-            "save_options", {"prompt": True, "response": True, "json": True, "workflow": True}
-        )
-
-        # Generate run ID
-        self.case_id = self._generate_case_id()
-        self.case_dir = os.path.join(self.debug_dir, self.case_id)
-
-        # If debug is enabled, create debug directory
-        if self.debug_enabled:
-            os.makedirs(self.case_dir, exist_ok=True)
-            print(f"Debug mode enabled, debug information will be saved to: {self.case_dir}")
-
-    def _generate_case_id(self) -> str:
-        """
-        Generate run ID
-
-        Returns:
-            Run ID
-        """
-        # Use format from configuration to generate run ID
-        format_str = self.config.get("case_id_format", "%Y%m%d_%H%M%S_%f")
-        timestamp = datetime.datetime.now().strftime(format_str)
-
-        # Add random string
-        random_str = str(uuid.uuid4())[:8]
-
-        return f"{timestamp}_{random_str}"
-
-    def save_text(self, content: str, filename: str, subdir: Optional[str] = None) -> Optional[str]:
-        """
-        Save text content to file
-
-        Args:
-            content: Text content
-            filename: File name
-            subdir: Subdirectory name
-
-        Returns:
-            Saved file path, returns None if debug is not enabled
-        """
-        if not self.debug_enabled:
-            return None
-
-        try:
-            # Determine save path
-            save_dir = self.case_dir
-            if subdir:
-                save_dir = os.path.join(save_dir, subdir)
-                os.makedirs(save_dir, exist_ok=True)
-
-            file_path = os.path.join(save_dir, filename)
-
-            # Save content
-            with open(file_path, "w", encoding="utf-8") as f:
-                f.write(content)
-
-            print(f"Debug information saved to: {file_path}")
-            return file_path
-        except Exception as e:
-            print(f"Failed to save debug information: {e}")
-            return None
-
-    def save_json(self, data: Union[dict, list], filename: str, subdir: Optional[str] = None) -> Optional[str]:
-        """
-        Save JSON data to file
-
-        Args:
-            data: JSON data
-            filename: File name
-            subdir: Subdirectory name
-
-        Returns:
-            Saved file path, returns None if debug is not enabled
-        """
-        if not self.debug_enabled:
-            return None
-
-        try:
-            # Determine save path
-            save_dir = self.case_dir
-            if subdir:
-                save_dir = os.path.join(save_dir, subdir)
-                os.makedirs(save_dir, exist_ok=True)
-
-            file_path = os.path.join(save_dir, filename)
-
-            # Save content
-            with open(file_path, "w", encoding="utf-8") as f:
-                json.dump(data, f, ensure_ascii=False, indent=2)
-
-            print(f"Debug information saved to: {file_path}")
-            return file_path
-        except Exception as e:
-            print(f"Failed to save debug information: {e}")
-            return None
-
-    def should_save(self, option: str) -> bool:
-        """
-        Check if specified type of debug information should be saved
-
-        Args:
-            option: Debug information type
-
-        Returns:
-            Whether it should be saved
-        """
-        if not self.debug_enabled:
-            return False
-
-        return self.save_options.get(option, False)

api/core/auto/workflow_generator/utils/llm_client.py

@@ -1,438 +0,0 @@
-"""
-LLM Client
-Used to call LLM API
-"""
-
-import json
-import re
-from typing import Any
-
-from core.auto.workflow_generator.utils.debug_manager import DebugManager
-from core.auto.workflow_generator.utils.prompts import DEFAULT_SYSTEM_PROMPT
-from core.model_manager import ModelInstance
-from core.model_runtime.entities.message_entities import SystemPromptMessage, UserPromptMessage
-
-
-class LLMClient:
-    """LLM Client"""
-
-    def __init__(self, model_instance: ModelInstance, debug_manager: DebugManager):
-        """
-        Initialize LLM client
-
-        Args:
-            api_key: API key
-            model: Model name
-            api_base: API base URL
-            max_tokens: Maximum number of tokens to generate
-            debug_manager: Debug manager
-        """
-
-        self.debug_manager = debug_manager or DebugManager()
-        self.model_instance = model_instance
-
-    def generate(self, prompt: str) -> str:
-        """
-        Generate text
-
-        Args:
-            prompt: Prompt text
-
-        Returns:
-            Generated text
-        """
-
-        # Save prompt
-        if self.debug_manager.should_save("prompt"):
-            self.debug_manager.save_text(prompt, "prompt.txt", "llm")
-
-        try:
-            response = self.model_instance.invoke_llm(
-                prompt_messages=[
-                    SystemPromptMessage(content=DEFAULT_SYSTEM_PROMPT),
-                    UserPromptMessage(content=prompt),
-                ],
-                model_parameters={"temperature": 0.7, "max_tokens": 4900},
-            )
-            content = ""
-            for chunk in response:
-                content += chunk.delta.message.content
-            print(f"Generation complete, text length: {len(content)} characters")
-
-            # Save response
-            if self.debug_manager.should_save("response"):
-                self.debug_manager.save_text(content, "response.txt", "llm")
-
-            return content
-        except Exception as e:
-            print(f"Error generating text: {e}")
-            raise e
-
-    def extract_json(self, text: str) -> dict[str, Any]:
-        """
-        Extract JSON from text
-
-        Args:
-            text: Text containing JSON
-
-        Returns:
-            Extracted JSON object
-        """
-        print("Starting JSON extraction from text...")
-
-        # Save original text
-        if self.debug_manager.should_save("json"):
-            self.debug_manager.save_text(text, "original_text.txt", "json")
-
-        # Use regex to extract JSON part
-        json_match = re.search(r"```json\n(.*?)\n```", text, re.DOTALL)
-        if json_match:
-            json_str = json_match.group(1)
-            print("Successfully extracted JSON from code block")
-        else:
-            # Try to match code block without language identifier
-            json_match = re.search(r"```\n(.*?)\n```", text, re.DOTALL)
-            if json_match:
-                json_str = json_match.group(1)
-                print("Successfully extracted JSON from code block without language identifier")
-            else:
-                # Try to match JSON surrounded by curly braces
-                json_match = re.search(r"(\{.*\})", text, re.DOTALL)
-                if json_match:
-                    json_str = json_match.group(1)
-                    print("Successfully extracted JSON from curly braces")
-                else:
-                    # Try to parse entire text
-                    json_str = text
-                    print("No JSON code block found, attempting to parse entire text")
-
-        # Save extracted JSON string
-        if self.debug_manager.should_save("json"):
-            self.debug_manager.save_text(json_str, "extracted_json.txt", "json")
-
-        # Try multiple methods to parse JSON
-        try:
-            # Try direct parsing
-            result = json.loads(json_str)
-        except json.JSONDecodeError as e:
-            print(f"Direct JSON parsing failed: {e}, attempting basic cleaning")
-            try:
-                # Try basic cleaning
-                cleaned_text = self._clean_text(json_str)
-                if self.debug_manager.should_save("json"):
-                    self.debug_manager.save_text(cleaned_text, "cleaned_json_1.txt", "json")
-                result = json.loads(cleaned_text)
-            except json.JSONDecodeError as e:
-                print(f"JSON parsing after basic cleaning failed: {e}, attempting to fix common errors")
-                try:
-                    # Try fixing common errors
-                    fixed_text = self._fix_json_errors(json_str)
-                    if self.debug_manager.should_save("json"):
-                        self.debug_manager.save_text(fixed_text, "cleaned_json_2.txt", "json")
-                    result = json.loads(fixed_text)
-                except json.JSONDecodeError as e:
-                    print(f"JSON parsing after fixing common errors failed: {e}, attempting aggressive cleaning")
-                    try:
-                        # Try aggressive cleaning
-                        aggressive_cleaned = self._aggressive_clean(json_str)
-                        if self.debug_manager.should_save("json"):
-                            self.debug_manager.save_text(aggressive_cleaned, "cleaned_json_3.txt", "json")
-                        result = json.loads(aggressive_cleaned)
-                    except json.JSONDecodeError as e:
-                        print(f"JSON parsing after aggressive cleaning failed: {e}, attempting manual JSON extraction")
-                        # Try manual JSON structure extraction
-                        result = self._manual_json_extraction(json_str)
-                        if self.debug_manager.should_save("json"):
-                            self.debug_manager.save_json(result, "manual_json.json", "json")
-
-        # Check for nested workflow structure
-        if "workflow" in result and isinstance(result["workflow"], dict):
-            print("Detected nested workflow structure, extracting top-level data")
-            # Extract workflow name and description
-            name = result.get("name", "Text Analysis Workflow")
-            description = result.get("description", "")
-
-            # Extract nodes and connections
-            nodes = result["workflow"].get("nodes", [])
-            connections = []
-
-            # If there are connections, extract them
-            if "connections" in result["workflow"]:
-                connections = result["workflow"]["connections"]
-
-            # Build standard format workflow description
-            result = {"name": name, "description": description, "nodes": nodes, "connections": connections}
-
-        # Save final parsed JSON
-        if self.debug_manager.should_save("json"):
-            self.debug_manager.save_json(result, "final_json.json", "json")
-
-        print(
-            f"JSON parsing successful, contains {len(result.get('nodes', []))} nodes and {len(result.get('connections', []))} connections"  # noqa: E501
-        )
-        return result
-
-    def _clean_text(self, text: str) -> str:
-        """
-        Clean text by removing non-JSON characters
-
-        Args:
-            text: Text to clean
-
-        Returns:
-            Cleaned text
-        """
-        print("Starting text cleaning...")
-        # Remove characters that might cause JSON parsing to fail
-        lines = text.split("\n")
-        cleaned_lines = []
-
-        in_json = False
-        for line in lines:
-            if line.strip().startswith("{") or line.strip().startswith("["):
-                in_json = True
-
-            if in_json:
-                cleaned_lines.append(line)
-
-            if line.strip().endswith("}") or line.strip().endswith("]"):
-                in_json = False
-
-        cleaned_text = "\n".join(cleaned_lines)
-        print(f"Text cleaning complete, length before: {len(text)}, length after: {len(cleaned_text)}")
-        return cleaned_text
-
-    def _fix_json_errors(self, text: str) -> str:
-        """
-        Fix common JSON errors
-
-        Args:
-            text: Text to fix
-
-        Returns:
-            Fixed text
-        """
-        print("Attempting to fix JSON errors...")
-
-        # Replace single quotes with double quotes
-        text = re.sub(r"'([^']*)'", r'"\1"', text)
-
-        # Fix missing commas
-        text = re.sub(r"}\s*{", "},{", text)
-        text = re.sub(r"]\s*{", "],{", text)
-        text = re.sub(r"}\s*\[", r"},\[", text)
-        text = re.sub(r"]\s*\[", r"],\[", text)
-
-        # Fix extra commas
-        text = re.sub(r",\s*}", "}", text)
-        text = re.sub(r",\s*]", "]", text)
-
-        # Ensure property names have double quotes
-        text = re.sub(r"([{,]\s*)(\w+)(\s*:)", r'\1"\2"\3', text)
-
-        return text
-
-    def _aggressive_clean(self, text: str) -> str:
-        """
-        More aggressive text cleaning
-
-        Args:
-            text: Text to clean
-
-        Returns:
-            Cleaned text
-        """
-        print("Using aggressive cleaning method...")
-
-        # Try to find outermost curly braces
-        start_idx = text.find("{")
-        end_idx = text.rfind("}")
-
-        if start_idx != -1 and end_idx != -1 and start_idx < end_idx:
-            text = text[start_idx : end_idx + 1]
-
-        # Remove comments
-        text = re.sub(r"//.*?\n", "\n", text)
-        text = re.sub(r"/\*.*?\*/", "", text, flags=re.DOTALL)
-
-        # Fix JSON format
-        text = self._fix_json_errors(text)
-
-        # Remove escape characters
-        text = text.replace("\\n", "\n").replace("\\t", "\t").replace('\\"', '"')
-
-        # Fix potential Unicode escape issues
-        text = re.sub(r"\\u([0-9a-fA-F]{4})", lambda m: chr(int(m.group(1), 16)), text)
-
-        return text
-
-    def _manual_json_extraction(self, text: str) -> dict[str, Any]:
-        """
-        Manual JSON structure extraction
-
-        Args:
-            text: Text to extract from
-
-        Returns:
-            Extracted JSON object
-        """
-        print("Attempting manual JSON structure extraction...")
-
-        # Extract workflow name
-        name_match = re.search(r'"name"\s*:\s*"([^"]*)"', text)
-        name = name_match.group(1) if name_match else "Simple Workflow"
-
-        # Extract workflow description
-        desc_match = re.search(r'"description"\s*:\s*"([^"]*)"', text)
-        description = desc_match.group(1) if desc_match else "Automatically generated workflow"
-
-        # Extract nodes
-        nodes = []
-        node_matches = re.finditer(r'\{\s*"id"\s*:\s*"([^"]*)"\s*,\s*"type"\s*:\s*"([^"]*)"', text)
-
-        for match in node_matches:
-            node_id = match.group(1)
-            node_type = match.group(2)
-
-            # Extract node title
-            title_match = re.search(rf'"id"\s*:\s*"{node_id}".*?"title"\s*:\s*"([^"]*)"', text, re.DOTALL)
-            title = title_match.group(1) if title_match else f"{node_type.capitalize()} Node"
-
-            # Extract node description
-            desc_match = re.search(rf'"id"\s*:\s*"{node_id}".*?"description"\s*:\s*"([^"]*)"', text, re.DOTALL)
-            desc = desc_match.group(1) if desc_match else ""
-
-            # Create basic node based on node type
-            if node_type == "start":
-                # Extract variables
-                variables = []
-                var_section_match = re.search(rf'"id"\s*:\s*"{node_id}".*?"variables"\s*:\s*\[(.*?)\]', text, re.DOTALL)
-
-                if var_section_match:
-                    var_section = var_section_match.group(1)
-                    var_matches = re.finditer(r'\{\s*"name"\s*:\s*"([^"]*)"\s*,\s*"type"\s*:\s*"([^"]*)"', var_section)
-
-                    for var_match in var_matches:
-                        var_name = var_match.group(1)
-                        var_type = var_match.group(2)
-
-                        # Extract variable description
-                        var_desc_match = re.search(
-                            rf'"name"\s*:\s*"{var_name}".*?"description"\s*:\s*"([^"]*)"', var_section, re.DOTALL
-                        )
-                        var_desc = var_desc_match.group(1) if var_desc_match else ""
-
-                        # Extract required status
-                        var_required_match = re.search(
-                            rf'"name"\s*:\s*"{var_name}".*?"required"\s*:\s*(true|false)', var_section, re.DOTALL
-                        )
-                        var_required = var_required_match.group(1).lower() == "true" if var_required_match else True
-
-                        variables.append(
-                            {"name": var_name, "type": var_type, "description": var_desc, "required": var_required}
-                        )
-
-                # If no variables found but this is a greeting workflow, add default user_name variable
-                if not variables and ("greeting" in name.lower()):
-                    variables.append(
-                        {"name": "user_name", "type": "string", "description": "User's name", "required": True}
-                    )
-
-                nodes.append({"id": node_id, "type": "start", "title": title, "desc": desc, "variables": variables})
-            elif node_type == "llm":
-                # Extract system prompt
-                system_prompt_match = re.search(
-                    rf'"id"\s*:\s*"{node_id}".*?"system_prompt"\s*:\s*"([^"]*)"', text, re.DOTALL
-                )
-                system_prompt = system_prompt_match.group(1) if system_prompt_match else "You are a helpful assistant"
-
-                # Extract user prompt
-                user_prompt_match = re.search(
-                    rf'"id"\s*:\s*"{node_id}".*?"user_prompt"\s*:\s*"([^"]*)"', text, re.DOTALL
-                )
-                user_prompt = user_prompt_match.group(1) if user_prompt_match else "Please answer the user's question"
-
-                nodes.append(
-                    {
-                        "id": node_id,
-                        "type": "llm",
-                        "title": title,
-                        "desc": desc,
-                        "provider": "zhipuai",
-                        "model": "glm-4-flash",
-                        "system_prompt": system_prompt,
-                        "user_prompt": user_prompt,
-                        "variables": [],
-                    }
-                )
-            elif node_type in ("template", "template-transform"):
-                # Extract template content
-                template_match = re.search(rf'"id"\s*:\s*"{node_id}".*?"template"\s*:\s*"([^"]*)"', text, re.DOTALL)
-                template = template_match.group(1) if template_match else ""
-
-                # Fix triple curly brace issue in template, replace {{{ with {{ and }}} with }}
-                template = template.replace("{{{", "{{").replace("}}}", "}}")
-
-                nodes.append(
-                    {
-                        "id": node_id,
-                        "type": "template-transform",
-                        "title": title,
-                        "desc": desc,
-                        "template": template,
-                        "variables": [],
-                    }
-                )
-            elif node_type == "end":
-                # Extract outputs
-                outputs = []
-                output_section_match = re.search(
-                    rf'"id"\s*:\s*"{node_id}".*?"outputs"\s*:\s*\[(.*?)\]', text, re.DOTALL
-                )
-
-                if output_section_match:
-                    output_section = output_section_match.group(1)
-                    output_matches = re.finditer(
-                        r'\{\s*"name"\s*:\s*"([^"]*)"\s*,\s*"type"\s*:\s*"([^"]*)"', output_section
-                    )
-
-                    for output_match in output_matches:
-                        output_name = output_match.group(1)
-                        output_type = output_match.group(2)
-
-                        # Extract source node
-                        source_node_match = re.search(
-                            rf'"name"\s*:\s*"{output_name}".*?"source_node"\s*:\s*"([^"]*)"', output_section, re.DOTALL
-                        )
-                        source_node = source_node_match.group(1) if source_node_match else ""
-
-                        # Extract source variable
-                        source_var_match = re.search(
-                            rf'"name"\s*:\s*"{output_name}".*?"source_variable"\s*:\s*"([^"]*)"',
-                            output_section,
-                            re.DOTALL,
-                        )
-                        source_var = source_var_match.group(1) if source_var_match else ""
-
-                        outputs.append(
-                            {
-                                "name": output_name,
-                                "type": output_type,
-                                "source_node": source_node,
-                                "source_variable": source_var,
-                            }
-                        )
-
-                nodes.append({"id": node_id, "type": "end", "title": title, "desc": desc, "outputs": outputs})
-            else:
-                # Other node types
-                nodes.append({"id": node_id, "type": node_type, "title": title, "desc": desc})
-
-        # Extract connections
-        connections = []
-        conn_matches = re.finditer(r'\{\s*"source"\s*:\s*"([^"]*)"\s*,\s*"target"\s*:\s*"([^"]*)"', text)
-
-        for match in conn_matches:
-            connections.append({"source": match.group(1), "target": match.group(2)})
-
-        return {"name": name, "description": description, "nodes": nodes, "connections": connections}

api/core/auto/workflow_generator/utils/prompts.py

@@ -1,171 +0,0 @@
-"""
-Prompt Template Collection
-Contains all prompt templates used for generating workflows
-"""
-
-# Default model configuration
-DEFAULT_MODEL_CONFIG = {
-    "provider": "zhipuai",
-    "model": "glm-4-flash",
-    "mode": "chat",
-    "completion_params": {"temperature": 0.7},
-}
-
-
-# Default system prompt
-DEFAULT_SYSTEM_PROMPT = "You are a workflow design expert who can design Dify workflows based on user requirements."
-
-
-# Code node template
-CODE_NODE_TEMPLATE = """def main(input_var):
-    # Process input variable
-    result = input_var
-
-    # Return a dictionary; keys must exactly match variable names defined in outputs
-    return {"output_var_name": result}"""
-
-
-def build_workflow_prompt(user_requirement: str) -> str:
-    """
-    Build workflow generation prompt
-
-    Args:
-        user_requirement: User requirement description
-
-    Returns:
-        Prompt string
-    """
-    # String concatenation to avoid brace escaping
-    prompt_part1 = (
-        """
-    Please design a Dify workflow based on the following user requirement:
-
-    User requirement: """
-        + user_requirement
-        + """
-
-    The description's language should align consistently with the user's requirements.
-    
-    Generate a concise workflow description containing the following node types:
-    - Start: Start node, defines workflow input parameters
-    - LLM: Large Language Model node for text generation
-    - Code: Code node to execute Python code
-    - Template: Template node for formatting outputs
-    - End: End node, defines workflow output
-
-    【Important Guidelines】:
-    1. When referencing variables in LLM nodes, use the format {{#nodeID.variable_name#}}, e.g., {{#1740019130520.user_question#}}, where 1740019130520 is the source node ID. Otherwise, in most cases, the user prompt should define a template to guide the LLM’s response.
-    2. Code nodes must define a `main` function that directly receives variables from upstream nodes as parameters; do not use template syntax inside the function.
-    3. Dictionary keys returned by Code nodes must exactly match the variable names defined in outputs.
-    4. Variables in Template nodes must strictly use double curly braces format "{{ variable_name }}"; note exactly two curly braces, neither one nor three. For example, "User question is: {{ user_question }}, answer: {{ answer }}". Triple curly braces such as "{{{ variable_name }}}" are strictly forbidden.
-    5. IMPORTANT: In Code nodes, the function parameter names MUST EXACTLY MATCH the variable names defined in that Code node. For example, if a Code node defines a variable with name "input_text" that receives data from an upstream node, the function parameter must also be named "input_text" (e.g., def main(input_text): ...).
-    6. CRITICAL: LLM nodes ALWAYS output their result in a variable named "text". When a Code node receives data from an LLM node, the source_variable MUST be "text". For example, if a Code node has a variable named "llm_output" that receives data from an LLM node, the source_variable should be "text", not "input_text" or any other name.
-
-    Return the workflow description in JSON format as follows:
-    ```json
-    {
-        "name": "Workflow Name",
-        "description": "Workflow description",
-        "nodes": [
-            {
-                "id": "node1",
-                "type": "start",
-                "title": "Start Node",
-                "description": "Description of the start node",
-                "variables": [
-                    {
-                        "name": "variable_name",
-                        "type": "string|number",
-                        "description": "Variable description",
-                        "required": true|false
-                    }
-                ]
-            },
-            {
-                "id": "node2",
-                "type": "llm",
-                "title": "LLM Node",
-                "description": "Description of LLM node",
-                "system_prompt": "System prompt",
-                "user_prompt": "User prompt, variables referenced using {{#nodeID.variable_name#}}, e.g., {{#node1.variable_name#}}",
-                "provider": "zhipuai",
-                "model": "glm-4-flash",
-                "variables": [
-                    {
-                        "name": "variable_name",
-                        "type": "string|number",
-                        "source_node": "node1",
-                        "source_variable": "variable_name"
-                    }
-                ]
-            },
-            {
-                "id": "node3",
-                "type": "code",
-                "title": "Code Node",
-                "description": "Description of the code node",
-                "code": "def main(input_var):\n    import re\n    match = re.search(r'Result[:：](.*?)(?=[.]|$)', input_var)\n    result = match.group(1).strip() if match else 'Not found'\n    return {'output': result}",
-                "variables": [
-                    {
-                        "name": "input_var",
-                        "type": "string|number",
-                        "source_node": "node2",
-                        "source_variable": "text"
-                    }
-                ],
-                "outputs": [
-                    {
-                        "name": "output_var_name",
-                        "type": "string|number|object"
-                    }
-                ]
-            },
-            {
-                "id": "node4",
-                "type": "template",
-                "title": "Template Node",
-                "description": "Description of the template node",
-                "template": "Template content using double curly braces, e.g.: The result is: {{ result }}",
-                "variables": [
-                    {
-                        "name": "variable_name",
-                        "type": "string|number",
-                        "source_node": "node3",
-                        "source_variable": "output_var_name"
-                    }
-                ]
-            },
-            {
-                "id": "node5",
-                "type": "end",
-                "title": "End Node",
-                "description": "Description of the end node",
-                "outputs": [
-                    {
-                        "name": "output_variable_name",
-                        "type": "string|number",
-                        "source_node": "node4",
-                        "source_variable": "output"
-                    }
-                ]
-            }
-        ],
-        "connections": [
-            {"source": "node1", "target": "node2"},
-            {"source": "node2", "target": "node3"},
-            {"source": "node3", "target": "node4"},
-            {"source": "node4", "target": "node5"}
-        ]
-    }
-    ```
-
-    Ensure the workflow logic is coherent, node connections are correct, and variable passing is logical.
-    Generate unique numeric IDs for each node, e.g., 1740019130520.
-    Generate appropriate unique names for each variable across the workflow.
-    Ensure all LLM nodes use provider "zhipuai" and model "glm-4-flash".
-
-    Note: LLM nodes usually return a long text; Code nodes typically require regex to extract relevant information.
-    """  # noqa: E501
-    )
-
-    return prompt_part1

api/core/auto/workflow_generator/utils/type_mapper.py

@@ -1,50 +0,0 @@
-"""
-Type Mapping Utility
-Used to map string types to Dify types
-"""
-
-from core.auto.node_types.common import InputVarType, VarType
-
-
-def map_var_type_to_input_type(var_type: str) -> InputVarType:
-    """
-    Map variable type to input variable type
-
-    Args:
-        var_type: Variable type string
-
-    Returns:
-        Input variable type
-    """
-    type_map = {
-        "string": InputVarType.text_input,
-        "number": InputVarType.number,
-        "boolean": InputVarType.select,
-        "object": InputVarType.json,
-        "array": InputVarType.json,
-        "file": InputVarType.file,
-    }
-
-    return type_map.get(var_type.lower(), InputVarType.text_input)
-
-
-def map_string_to_var_type(type_str: str) -> VarType:
-    """
-    Map string to variable type
-
-    Args:
-        type_str: Type string
-
-    Returns:
-        Variable type
-    """
-    type_map = {
-        "string": VarType.string,
-        "number": VarType.number,
-        "boolean": VarType.boolean,
-        "object": VarType.object,
-        "array": VarType.array,
-        "file": VarType.file,
-    }
-
-    return type_map.get(type_str.lower(), VarType.string)

api/core/auto/workflow_generator/workflow.py

@@ -1,134 +0,0 @@
-import json
-from typing import Any
-
-import yaml
-
-from core.auto.node_types.common import CompleteEdge, CompleteNode
-
-
-class Workflow:
-    """
-    Workflow class
-    """
-
-    def __init__(self, name: str, nodes: list[CompleteNode], edges: list[CompleteEdge]):
-        """
-        Initialize workflow
-
-        Args:
-            name: Workflow name
-            nodes: List of nodes
-            edges: List of edges
-        """
-        self.name = name
-        self.nodes = nodes
-        self.edges = edges
-
-    def to_dict(self) -> dict[str, Any]:
-        """
-        Convert workflow to dictionary
-
-        Returns:
-            Workflow dictionary
-        """
-        # Apply basic information (fixed template)
-        app_info = {
-            "description": "",
-            "icon": "🤖",
-            "icon_background": "#FFEAD5",
-            "mode": "workflow",
-            "name": self.name,
-            "use_icon_as_answer_icon": False,
-        }
-
-        # Feature configuration (fixed template)
-        features = {
-            "file_upload": {
-                "allowed_file_extensions": [".JPG", ".JPEG", ".PNG", ".GIF", ".WEBP", ".SVG"],
-                "allowed_file_types": ["image"],
-                "allowed_file_upload_methods": ["local_file", "remote_url"],
-                "enabled": False,
-                "fileUploadConfig": {
-                    "audio_file_size_limit": 50,
-                    "batch_count_limit": 5,
-                    "file_size_limit": 15,
-                    "image_file_size_limit": 10,
-                    "video_file_size_limit": 100,
-                },
-                "image": {"enabled": False, "number_limits": 3, "transfer_methods": ["local_file", "remote_url"]},
-                "number_limits": 3,
-            },
-            "opening_statement": "",
-            "retriever_resource": {"enabled": True},
-            "sensitive_word_avoidance": {"enabled": False},
-            "speech_to_text": {"enabled": False},
-            "suggested_questions": [],
-            "suggested_questions_after_answer": {"enabled": False},
-            "text_to_speech": {"enabled": False, "language": "", "voice": ""},
-        }
-
-        # View configuration (fixed template)
-        viewport = {"x": 92.96659905656679, "y": 79.13437154762897, "zoom": 0.9002006986311041}
-
-        # Nodes and edges
-        nodes_data = []
-        for node in self.nodes:
-            node_data = node.to_json()
-            nodes_data.append(node_data)
-
-        edges_data = []
-        for edge in self.edges:
-            edge_data = edge.to_json()
-            edges_data.append(edge_data)
-
-        # Build a complete workflow dictionary
-        workflow_dict = {
-            "app": app_info,
-            "kind": "app",
-            "version": "0.1.2",
-            "workflow": {
-                "conversation_variables": [],
-                "environment_variables": [],
-                "features": features,
-                "graph": {"edges": edges_data, "nodes": nodes_data, "viewport": viewport},
-            },
-        }
-
-        return workflow_dict
-
-    def save_to_yaml(self, file_path: str):
-        """
-        Save workflow to YAML file
-
-        Args:
-            file_path: File path
-        """
-        workflow_dict = self.to_dict()
-
-        with open(file_path, "w", encoding="utf-8") as f:
-            yaml.dump(workflow_dict, f, allow_unicode=True, sort_keys=False)
-
-        print(f"Workflow saved to: {file_path}")
-
-    def save_to_json(self, file_path: str):
-        """
-        Save workflow to JSON file
-
-        Args:
-            file_path: File path
-        """
-        workflow_dict = self.to_dict()
-
-        with open(file_path, "w", encoding="utf-8") as f:
-            json.dump(workflow_dict, f, indent=2, ensure_ascii=False)
-
-        print(f"Workflow saved to: {file_path}")
-
-    def to_yaml(self) -> str:
-        """
-        Convert workflow to YAML string
-
-        Returns:
-            YAML string
-        """
-        return yaml.dump(self.to_dict(), allow_unicode=True, sort_keys=False)

api/core/auto/workflow_generator/workflow_generator.py

@@ -1,159 +0,0 @@
-"""
-Workflow Generator
-Used to generate Dify workflows based on user requirements
-"""
-
-from pydantic import ValidationError
-
-from core.auto.workflow_generator.generators.edge_generator import EdgeGenerator
-from core.auto.workflow_generator.generators.layout_engine import LayoutEngine
-from core.auto.workflow_generator.generators.node_generator import NodeGenerator
-from core.auto.workflow_generator.models.workflow_description import WorkflowDescription
-from core.auto.workflow_generator.utils.config_manager import ConfigManager
-from core.auto.workflow_generator.utils.debug_manager import DebugManager
-from core.auto.workflow_generator.utils.llm_client import LLMClient
-from core.auto.workflow_generator.utils.prompts import build_workflow_prompt
-from core.auto.workflow_generator.workflow import Workflow
-from core.model_manager import ModelInstance
-
-
-class WorkflowGenerator:
-    """Workflow generator for creating Dify workflows based on user requirements"""
-
-    def __init__(self, model_instance: ModelInstance, config_dir: str = "config", debug_enabled: bool = False):
-        """
-        Initialize workflow generator
-
-        Args:
-            api_key: LLM API key
-            config_dir: Configuration directory path
-            model_name: Specified model name, uses default model if not specified
-            debug_enabled: Whether to enable debug mode
-        """
-        # Load configuration
-        self.config = ConfigManager(config_dir)
-
-        # Initialize debug manager
-        self.debug_manager = DebugManager(config=self.config.get("debug", default={}), debug_enabled=debug_enabled)
-
-        # Get model configuration
-
-        # Initialize LLM client
-        self.llm_client = LLMClient(model_instance=model_instance, debug_manager=self.debug_manager)
-
-    def generate_workflow(self, user_requirement: str) -> str:
-        """
-        Generate workflow based on user requirements
-
-        Args:
-            user_requirement: User requirement description
-            output_path: Output file path, uses default path from config if None
-
-        Returns:
-            Generated workflow YAML file path
-        """
-        print("\n===== Starting Workflow Generation =====")
-        print(f"User requirement: {user_requirement}")
-
-        # Save user requirement
-        if self.debug_manager.should_save("workflow"):
-            self.debug_manager.save_text(user_requirement, "user_requirement.txt", "workflow")
-
-        # Use default path from config if output path not specified
-
-        # Step 1: Generate simple workflow description
-        print("\n----- Step 1: Generating Simple Workflow Description -----")
-        workflow_description = self._generate_workflow_description(user_requirement)
-        print(f"Workflow name: {workflow_description.name}")
-        print(f"Workflow description: {workflow_description.description}")
-        print(f"Number of nodes: {len(workflow_description.nodes)}")
-        print(f"Number of connections: {len(workflow_description.connections)}")
-
-        # Save workflow description
-        if self.debug_manager.should_save("workflow"):
-            self.debug_manager.save_json(workflow_description.dict(), "workflow_description.json", "workflow")
-
-        # Step 2: Parse description and generate nodes
-        print("\n----- Step 2: Parsing Description, Generating Nodes -----")
-        nodes = NodeGenerator.create_nodes(workflow_description.nodes)
-        print(f"Generated nodes: {len(nodes)}")
-        for i, node in enumerate(nodes):
-            print(f"Node {i + 1}: ID={node.id}, Type={node.data.type.value}, Title={node.data.title}")
-
-        # Save node information
-        if self.debug_manager.should_save("workflow"):
-            nodes_data = [node.dict() for node in nodes]
-            self.debug_manager.save_json(nodes_data, "nodes.json", "workflow")
-
-        # Step 3: Generate edges
-        print("\n----- Step 3: Generating Edges -----")
-        edges = EdgeGenerator.create_edges(nodes, workflow_description.connections)
-        print(f"Generated edges: {len(edges)}")
-        for i, edge in enumerate(edges):
-            print(f"Edge {i + 1}: ID={edge.id}, Source={edge.source}, Target={edge.target}")
-
-        # Save edge information
-        if self.debug_manager.should_save("workflow"):
-            edges_data = [edge.dict() for edge in edges]
-            self.debug_manager.save_json(edges_data, "edges.json", "workflow")
-
-        # Step 4: Apply layout
-        print("\n----- Step 4: Applying Layout -----")
-        LayoutEngine.apply_topological_layout(nodes, edges)
-        print("Applied topological sort layout")
-
-        # Save nodes with layout
-        if self.debug_manager.should_save("workflow"):
-            nodes_with_layout = [node.dict() for node in nodes]
-            self.debug_manager.save_json(nodes_with_layout, "nodes_with_layout.json", "workflow")
-
-        # Step 5: Generate YAML
-        print("\n----- Step 5: Generating YAML -----")
-        workflow = Workflow(name=workflow_description.name, nodes=nodes, edges=edges)
-
-        # Ensure output directory exists
-
-        # Save as YAML
-
-        # Save final YAML
-        print("\n===== Workflow Generation Complete =====")
-        return workflow.to_yaml()
-
-    def _generate_workflow_description(self, user_requirement: str) -> WorkflowDescription:
-        """
-        Generate simple workflow description using LLM
-
-        Args:
-            user_requirement: User requirement description
-
-        Returns:
-            Simple workflow description
-        """
-        # Build prompt
-        print("Building prompt...")
-        prompt = build_workflow_prompt(user_requirement)
-
-        # Call LLM
-        print("Calling LLM to generate workflow description...")
-        response_text = self.llm_client.generate(prompt)
-
-        # Parse LLM response
-        print("Parsing LLM response...")
-        workflow_description_dict = self.llm_client.extract_json(response_text)
-
-        try:
-            # Parse into WorkflowDescription object
-            print("Converting JSON to WorkflowDescription object...")
-            workflow_description = WorkflowDescription.parse_obj(workflow_description_dict)
-            return workflow_description
-        except ValidationError as e:
-            # If parsing fails, print error and raise exception
-            error_msg = f"Failed to parse workflow description: {e}"
-            print(error_msg)
-
-            # Save error information
-            if self.debug_manager.should_save("workflow"):
-                self.debug_manager.save_text(str(e), "validation_error.txt", "workflow")
-                self.debug_manager.save_json(workflow_description_dict, "invalid_workflow_description.json", "workflow")
-
-            raise ValueError(error_msg)

api/core/entities/provider_configuration.py

@@ -7,7 +7,6 @@ from json import JSONDecodeError
 from typing import Optional
 
 from pydantic import BaseModel, ConfigDict, Field
-from sqlalchemy import or_
 
 from constants import HIDDEN_VALUE
 from core.entities.model_entities import ModelStatus, ModelWithProviderEntity, SimpleModelProviderEntity
@@ -180,37 +179,35 @@ class ProviderConfiguration(BaseModel):
             else [],
         )
 
+    def _get_custom_provider_credentials(self) -> Provider | None:
+        """
+        Get custom provider credentials.
+        """
+        # get provider
+        model_provider_id = ModelProviderID(self.provider.provider)
+        provider_names = [self.provider.provider]
+        if model_provider_id.is_langgenius():
+            provider_names.append(model_provider_id.provider_name)
+
+        provider_record = (
+            db.session.query(Provider)
+            .filter(
+                Provider.tenant_id == self.tenant_id,
+                Provider.provider_type == ProviderType.CUSTOM.value,
+                Provider.provider_name.in_(provider_names),
+            )
+            .first()
+        )
+
+        return provider_record
+
     def custom_credentials_validate(self, credentials: dict) -> tuple[Provider | None, dict]:
         """
         Validate custom credentials.
         :param credentials: provider credentials
         :return:
         """
-        # get provider
-        model_provider_id = ModelProviderID(self.provider.provider)
-        if model_provider_id.is_langgenius():
-            provider_record = (
-                db.session.query(Provider)
-                .filter(
-                    Provider.tenant_id == self.tenant_id,
-                    Provider.provider_type == ProviderType.CUSTOM.value,
-                    or_(
-                        Provider.provider_name == model_provider_id.provider_name,
-                        Provider.provider_name == self.provider.provider,
-                    ),
-                )
-                .first()
-            )
-        else:
-            provider_record = (
-                db.session.query(Provider)
-                .filter(
-                    Provider.tenant_id == self.tenant_id,
-                    Provider.provider_type == ProviderType.CUSTOM.value,
-                    Provider.provider_name == self.provider.provider,
-                )
-                .first()
-            )
+        provider_record = self._get_custom_provider_credentials()
 
         # Get provider credential secret variables
         provider_credential_secret_variables = self.extract_secret_variables(
@@ -291,18 +288,7 @@ class ProviderConfiguration(BaseModel):
         :return:
         """
         # get provider
-        provider_record = (
-            db.session.query(Provider)
-            .filter(
-                Provider.tenant_id == self.tenant_id,
-                or_(
-                    Provider.provider_name == ModelProviderID(self.provider.provider).plugin_name,
-                    Provider.provider_name == self.provider.provider,
-                ),
-                Provider.provider_type == ProviderType.CUSTOM.value,
-            )
-            .first()
-        )
+        provider_record = self._get_custom_provider_credentials()
 
         # delete provider
         if provider_record:
@@ -349,6 +335,33 @@ class ProviderConfiguration(BaseModel):
 
         return None
 
+    def _get_custom_model_credentials(
+        self,
+        model_type: ModelType,
+        model: str,
+    ) -> ProviderModel | None:
+        """
+        Get custom model credentials.
+        """
+        # get provider model
+        model_provider_id = ModelProviderID(self.provider.provider)
+        provider_names = [self.provider.provider]
+        if model_provider_id.is_langgenius():
+            provider_names.append(model_provider_id.provider_name)
+
+        provider_model_record = (
+            db.session.query(ProviderModel)
+            .filter(
+                ProviderModel.tenant_id == self.tenant_id,
+                ProviderModel.provider_name.in_(provider_names),
+                ProviderModel.model_name == model,
+                ProviderModel.model_type == model_type.to_origin_model_type(),
+            )
+            .first()
+        )
+
+        return provider_model_record
+
     def custom_model_credentials_validate(
         self, model_type: ModelType, model: str, credentials: dict
     ) -> tuple[ProviderModel | None, dict]:
@@ -361,16 +374,7 @@ class ProviderConfiguration(BaseModel):
         :return:
         """
         # get provider model
-        provider_model_record = (
-            db.session.query(ProviderModel)
-            .filter(
-                ProviderModel.tenant_id == self.tenant_id,
-                ProviderModel.provider_name == self.provider.provider,
-                ProviderModel.model_name == model,
-                ProviderModel.model_type == model_type.to_origin_model_type(),
-            )
-            .first()
-        )
+        provider_model_record = self._get_custom_model_credentials(model_type, model)
 
         # Get provider credential secret variables
         provider_credential_secret_variables = self.extract_secret_variables(
@@ -451,16 +455,7 @@ class ProviderConfiguration(BaseModel):
         :return:
         """
         # get provider model
-        provider_model_record = (
-            db.session.query(ProviderModel)
-            .filter(
-                ProviderModel.tenant_id == self.tenant_id,
-                ProviderModel.provider_name == self.provider.provider,
-                ProviderModel.model_name == model,
-                ProviderModel.model_type == model_type.to_origin_model_type(),
-            )
-            .first()
-        )
+        provider_model_record = self._get_custom_model_credentials(model_type, model)
 
         # delete provider model
         if provider_model_record:
@@ -475,6 +470,26 @@ class ProviderConfiguration(BaseModel):
 
             provider_model_credentials_cache.delete()
 
+    def _get_provider_model_setting(self, model_type: ModelType, model: str) -> ProviderModelSetting | None:
+        """
+        Get provider model setting.
+        """
+        model_provider_id = ModelProviderID(self.provider.provider)
+        provider_names = [self.provider.provider]
+        if model_provider_id.is_langgenius():
+            provider_names.append(model_provider_id.provider_name)
+
+        return (
+            db.session.query(ProviderModelSetting)
+            .filter(
+                ProviderModelSetting.tenant_id == self.tenant_id,
+                ProviderModelSetting.provider_name.in_(provider_names),
+                ProviderModelSetting.model_type == model_type.to_origin_model_type(),
+                ProviderModelSetting.model_name == model,
+            )
+            .first()
+        )
+
     def enable_model(self, model_type: ModelType, model: str) -> ProviderModelSetting:
         """
         Enable model.
@@ -482,16 +497,7 @@ class ProviderConfiguration(BaseModel):
         :param model: model name
         :return:
         """
-        model_setting = (
-            db.session.query(ProviderModelSetting)
-            .filter(
-                ProviderModelSetting.tenant_id == self.tenant_id,
-                ProviderModelSetting.provider_name == self.provider.provider,
-                ProviderModelSetting.model_type == model_type.to_origin_model_type(),
-                ProviderModelSetting.model_name == model,
-            )
-            .first()
-        )
+        model_setting = self._get_provider_model_setting(model_type, model)
 
         if model_setting:
             model_setting.enabled = True
@@ -516,16 +522,7 @@ class ProviderConfiguration(BaseModel):
         :param model: model name
         :return:
         """
-        model_setting = (
-            db.session.query(ProviderModelSetting)
-            .filter(
-                ProviderModelSetting.tenant_id == self.tenant_id,
-                ProviderModelSetting.provider_name == self.provider.provider,
-                ProviderModelSetting.model_type == model_type.to_origin_model_type(),
-                ProviderModelSetting.model_name == model,
-            )
-            .first()
-        )
+        model_setting = self._get_provider_model_setting(model_type, model)
 
         if model_setting:
             model_setting.enabled = False
@@ -550,13 +547,24 @@ class ProviderConfiguration(BaseModel):
         :param model: model name
         :return:
         """
+        return self._get_provider_model_setting(model_type, model)
+
+    def _get_load_balancing_config(self, model_type: ModelType, model: str) -> Optional[LoadBalancingModelConfig]:
+        """
+        Get load balancing config.
+        """
+        model_provider_id = ModelProviderID(self.provider.provider)
+        provider_names = [self.provider.provider]
+        if model_provider_id.is_langgenius():
+            provider_names.append(model_provider_id.provider_name)
+
         return (
-            db.session.query(ProviderModelSetting)
+            db.session.query(LoadBalancingModelConfig)
             .filter(
-                ProviderModelSetting.tenant_id == self.tenant_id,
-                ProviderModelSetting.provider_name == self.provider.provider,
-                ProviderModelSetting.model_type == model_type.to_origin_model_type(),
-                ProviderModelSetting.model_name == model,
+                LoadBalancingModelConfig.tenant_id == self.tenant_id,
+                LoadBalancingModelConfig.provider_name.in_(provider_names),
+                LoadBalancingModelConfig.model_type == model_type.to_origin_model_type(),
+                LoadBalancingModelConfig.model_name == model,
             )
             .first()
         )
@@ -568,11 +576,16 @@ class ProviderConfiguration(BaseModel):
         :param model: model name
         :return:
         """
+        model_provider_id = ModelProviderID(self.provider.provider)
+        provider_names = [self.provider.provider]
+        if model_provider_id.is_langgenius():
+            provider_names.append(model_provider_id.provider_name)
+
         load_balancing_config_count = (
             db.session.query(LoadBalancingModelConfig)
             .filter(
                 LoadBalancingModelConfig.tenant_id == self.tenant_id,
-                LoadBalancingModelConfig.provider_name == self.provider.provider,
+                LoadBalancingModelConfig.provider_name.in_(provider_names),
                 LoadBalancingModelConfig.model_type == model_type.to_origin_model_type(),
                 LoadBalancingModelConfig.model_name == model,
             )
@@ -582,16 +595,7 @@ class ProviderConfiguration(BaseModel):
         if load_balancing_config_count <= 1:
             raise ValueError("Model load balancing configuration must be more than 1.")
 
-        model_setting = (
-            db.session.query(ProviderModelSetting)
-            .filter(
-                ProviderModelSetting.tenant_id == self.tenant_id,
-                ProviderModelSetting.provider_name == self.provider.provider,
-                ProviderModelSetting.model_type == model_type.to_origin_model_type(),
-                ProviderModelSetting.model_name == model,
-            )
-            .first()
-        )
+        model_setting = self._get_provider_model_setting(model_type, model)
 
         if model_setting:
             model_setting.load_balancing_enabled = True
@@ -616,11 +620,16 @@ class ProviderConfiguration(BaseModel):
         :param model: model name
         :return:
         """
+        model_provider_id = ModelProviderID(self.provider.provider)
+        provider_names = [self.provider.provider]
+        if model_provider_id.is_langgenius():
+            provider_names.append(model_provider_id.provider_name)
+
         model_setting = (
             db.session.query(ProviderModelSetting)
             .filter(
                 ProviderModelSetting.tenant_id == self.tenant_id,
-                ProviderModelSetting.provider_name == self.provider.provider,
+                ProviderModelSetting.provider_name.in_(provider_names),
                 ProviderModelSetting.model_type == model_type.to_origin_model_type(),
                 ProviderModelSetting.model_name == model,
             )
@@ -677,11 +686,16 @@ class ProviderConfiguration(BaseModel):
             return
 
         # get preferred provider
+        model_provider_id = ModelProviderID(self.provider.provider)
+        provider_names = [self.provider.provider]
+        if model_provider_id.is_langgenius():
+            provider_names.append(model_provider_id.provider_name)
+
         preferred_model_provider = (
             db.session.query(TenantPreferredModelProvider)
             .filter(
                 TenantPreferredModelProvider.tenant_id == self.tenant_id,
-                TenantPreferredModelProvider.provider_name == self.provider.provider,
+                TenantPreferredModelProvider.provider_name.in_(provider_names),
             )
             .first()
         )

api/core/file/models.py

@@ -63,7 +63,9 @@ class File(BaseModel):
         extension: Optional[str] = None,
         mime_type: Optional[str] = None,
         size: int = -1,
-        storage_key: str,
+        storage_key: Optional[str] = None,
+        dify_model_identity: Optional[str] = FILE_MODEL_IDENTITY,
+        url: Optional[str] = None,
     ):
         super().__init__(
             id=id,
@@ -76,8 +78,10 @@ class File(BaseModel):
             extension=extension,
             mime_type=mime_type,
             size=size,
+            dify_model_identity=dify_model_identity,
+            url=url,
         )
-        self._storage_key = storage_key
+        self._storage_key = str(storage_key)
 
     def to_dict(self) -> Mapping[str, str | int | None]:
         data = self.model_dump(mode="json")

api/core/helper/ssrf_proxy.py

@@ -11,6 +11,19 @@ from configs import dify_config
 
 SSRF_DEFAULT_MAX_RETRIES = dify_config.SSRF_DEFAULT_MAX_RETRIES
 
+HTTP_REQUEST_NODE_SSL_VERIFY = True  # Default value for HTTP_REQUEST_NODE_SSL_VERIFY is True
+try:
+    HTTP_REQUEST_NODE_SSL_VERIFY = dify_config.HTTP_REQUEST_NODE_SSL_VERIFY
+    http_request_node_ssl_verify_lower = str(HTTP_REQUEST_NODE_SSL_VERIFY).lower()
+    if http_request_node_ssl_verify_lower == "true":
+        HTTP_REQUEST_NODE_SSL_VERIFY = True
+    elif http_request_node_ssl_verify_lower == "false":
+        HTTP_REQUEST_NODE_SSL_VERIFY = False
+    else:
+        raise ValueError("Invalid value. HTTP_REQUEST_NODE_SSL_VERIFY should be 'True' or 'False'")
+except NameError:
+    HTTP_REQUEST_NODE_SSL_VERIFY = True
+
 BACKOFF_FACTOR = 0.5
 STATUS_FORCELIST = [429, 500, 502, 503, 504]
 
@@ -39,17 +52,17 @@ def make_request(method, url, max_retries=SSRF_DEFAULT_MAX_RETRIES, **kwargs):
     while retries <= max_retries:
         try:
             if dify_config.SSRF_PROXY_ALL_URL:
-                with httpx.Client(proxy=dify_config.SSRF_PROXY_ALL_URL) as client:
+                with httpx.Client(proxy=dify_config.SSRF_PROXY_ALL_URL, verify=HTTP_REQUEST_NODE_SSL_VERIFY) as client:
                     response = client.request(method=method, url=url, **kwargs)
             elif dify_config.SSRF_PROXY_HTTP_URL and dify_config.SSRF_PROXY_HTTPS_URL:
                 proxy_mounts = {
                     "http://": httpx.HTTPTransport(proxy=dify_config.SSRF_PROXY_HTTP_URL),
                     "https://": httpx.HTTPTransport(proxy=dify_config.SSRF_PROXY_HTTPS_URL),
                 }
-                with httpx.Client(mounts=proxy_mounts) as client:
+                with httpx.Client(mounts=proxy_mounts, verify=HTTP_REQUEST_NODE_SSL_VERIFY) as client:
                     response = client.request(method=method, url=url, **kwargs)
             else:
-                with httpx.Client() as client:
+                with httpx.Client(verify=HTTP_REQUEST_NODE_SSL_VERIFY) as client:
                     response = client.request(method=method, url=url, **kwargs)
 
             if response.status_code not in STATUS_FORCELIST:

api/core/model_runtime/docs/en_US/interfaces.md

@@ -493,7 +493,7 @@ If inputting a combination of text and images, the images need to be constructed
 The base class for all Role message bodies, used only for parameter declaration and cannot be initialized.
 
 ```python
-class PromptMessage(ABC, BaseModel):
+class PromptMessage(BaseModel):
     """
     Model class for prompt message.
     """

api/core/model_runtime/docs/zh_Hans/interfaces.md

@@ -533,7 +533,7 @@ class ImagePromptMessageContent(PromptMessageContent):
 所有 Role 消息体的基类，仅作为参数声明用，不可初始化。
 
 ```python
-class PromptMessage(ABC, BaseModel):
+class PromptMessage(BaseModel):
     """
     Model class for prompt message.
     """

api/core/model_runtime/entities/__init__.py

@@ -31,11 +31,9 @@ __all__ = [
     "ModelPropertyKey",
     "MultiModalPromptMessageContent",
     "PromptMessage",
-    "PromptMessage",
     "PromptMessageContent",
     "PromptMessageContentType",
     "PromptMessageRole",
-    "PromptMessageRole",
     "PromptMessageTool",
     "SystemPromptMessage",
     "TextPromptMessageContent",

api/core/model_runtime/entities/message_entities.py

@@ -1,4 +1,3 @@
-from abc import ABC
 from collections.abc import Sequence
 from enum import Enum, StrEnum
 from typing import Optional
@@ -119,7 +118,7 @@ class DocumentPromptMessageContent(MultiModalPromptMessageContent):
     type: PromptMessageContentType = PromptMessageContentType.DOCUMENT
 
 
-class PromptMessage(ABC, BaseModel):
+class PromptMessage(BaseModel):
     """
     Model class for prompt message.
     """

api/core/model_runtime/model_providers/__base/ai_model.py

@@ -80,7 +80,7 @@ class AIModel(BaseModel):
                         )
                     )
                 elif isinstance(invoke_error, InvokeError):
-                    return invoke_error(description=f"[{self.provider_name}] {invoke_error.description}, {str(error)}")
+                    return InvokeError(description=f"[{self.provider_name}] {invoke_error.description}, {str(error)}")
                 else:
                     return error
 

api/core/ops/ops_trace_manager.py

@@ -214,6 +214,8 @@ class OpsTraceManager:
                 provider_config_map[tracing_provider]["trace_instance"],
                 provider_config_map[tracing_provider]["config_class"],
             )
+            if not decrypt_trace_config:
+                return None
             tracing_instance = trace_instance(config_class(**decrypt_trace_config))
             return tracing_instance
 

api/core/plugin/backwards_invocation/model.py

@@ -3,7 +3,7 @@ from binascii import hexlify, unhexlify
 from collections.abc import Generator
 
 from core.model_manager import ModelManager
-from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk
+from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     PromptMessage,
     SystemPromptMessage,
@@ -46,7 +46,7 @@ class PluginModelBackwardsInvocation(BaseBackwardsInvocation):
             model_parameters=payload.completion_params,
             tools=payload.tools,
             stop=payload.stop,
-            stream=payload.stream or True,
+            stream=True if payload.stream is None else payload.stream,
             user=user_id,
         )
 
@@ -64,7 +64,21 @@ class PluginModelBackwardsInvocation(BaseBackwardsInvocation):
         else:
             if response.usage:
                 LLMNode.deduct_llm_quota(tenant_id=tenant.id, model_instance=model_instance, usage=response.usage)
-            return response
+
+            def handle_non_streaming(response: LLMResult) -> Generator[LLMResultChunk, None, None]:
+                yield LLMResultChunk(
+                    model=response.model,
+                    prompt_messages=response.prompt_messages,
+                    system_fingerprint=response.system_fingerprint,
+                    delta=LLMResultChunkDelta(
+                        index=0,
+                        message=response.message,
+                        usage=response.usage,
+                        finish_reason="",
+                    ),
+                )
+
+            return handle_non_streaming(response)
 
     @classmethod
     def invoke_text_embedding(cls, user_id: str, tenant: Tenant, payload: RequestInvokeTextEmbedding):

api/core/plugin/entities/parameters.py

@@ -147,7 +147,7 @@ def init_frontend_parameter(rule: PluginParameter, type: enum.StrEnum, value: An
     init frontend parameter by rule
     """
     parameter_value = value
-    if not parameter_value and parameter_value != 0:
+    if not parameter_value and parameter_value != 0 and type != PluginParameterType.TOOLS_SELECTOR:
         # get default value
         parameter_value = rule.default
         if not parameter_value and rule.required:

api/core/prompt/advanced_prompt_transform.py

@@ -46,6 +46,7 @@ class AdvancedPromptTransform(PromptTransform):
         memory_config: Optional[MemoryConfig],
         memory: Optional[TokenBufferMemory],
         model_config: ModelConfigWithCredentialsEntity,
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
     ) -> list[PromptMessage]:
         prompt_messages = []
 
@@ -59,6 +60,7 @@ class AdvancedPromptTransform(PromptTransform):
                 memory_config=memory_config,
                 memory=memory,
                 model_config=model_config,
+                image_detail_config=image_detail_config,
             )
         elif isinstance(prompt_template, list) and all(isinstance(item, ChatModelMessage) for item in prompt_template):
             prompt_messages = self._get_chat_model_prompt_messages(
@@ -70,6 +72,7 @@ class AdvancedPromptTransform(PromptTransform):
                 memory_config=memory_config,
                 memory=memory,
                 model_config=model_config,
+                image_detail_config=image_detail_config,
             )
 
         return prompt_messages
@@ -84,6 +87,7 @@ class AdvancedPromptTransform(PromptTransform):
         memory_config: Optional[MemoryConfig],
         memory: Optional[TokenBufferMemory],
         model_config: ModelConfigWithCredentialsEntity,
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
     ) -> list[PromptMessage]:
         """
         Get completion model prompt messages.
@@ -124,7 +128,9 @@ class AdvancedPromptTransform(PromptTransform):
             prompt_message_contents: list[PromptMessageContent] = []
             prompt_message_contents.append(TextPromptMessageContent(data=prompt))
             for file in files:
-                prompt_message_contents.append(file_manager.to_prompt_message_content(file))
+                prompt_message_contents.append(
+                    file_manager.to_prompt_message_content(file, image_detail_config=image_detail_config)
+                )
 
             prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
         else:
@@ -142,6 +148,7 @@ class AdvancedPromptTransform(PromptTransform):
         memory_config: Optional[MemoryConfig],
         memory: Optional[TokenBufferMemory],
         model_config: ModelConfigWithCredentialsEntity,
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
     ) -> list[PromptMessage]:
         """
         Get chat model prompt messages.
@@ -197,7 +204,9 @@ class AdvancedPromptTransform(PromptTransform):
                 prompt_message_contents: list[PromptMessageContent] = []
                 prompt_message_contents.append(TextPromptMessageContent(data=query))
                 for file in files:
-                    prompt_message_contents.append(file_manager.to_prompt_message_content(file))
+                    prompt_message_contents.append(
+                        file_manager.to_prompt_message_content(file, image_detail_config=image_detail_config)
+                    )
                 prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
             else:
                 prompt_messages.append(UserPromptMessage(content=query))
@@ -209,19 +218,25 @@ class AdvancedPromptTransform(PromptTransform):
                     # get last user message content and add files
                     prompt_message_contents = [TextPromptMessageContent(data=cast(str, last_message.content))]
                     for file in files:
-                        prompt_message_contents.append(file_manager.to_prompt_message_content(file))
+                        prompt_message_contents.append(
+                            file_manager.to_prompt_message_content(file, image_detail_config=image_detail_config)
+                        )
 
                     last_message.content = prompt_message_contents
                 else:
                     prompt_message_contents = [TextPromptMessageContent(data="")]  # not for query
                     for file in files:
-                        prompt_message_contents.append(file_manager.to_prompt_message_content(file))
+                        prompt_message_contents.append(
+                            file_manager.to_prompt_message_content(file, image_detail_config=image_detail_config)
+                        )
 
                     prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
             else:
                 prompt_message_contents = [TextPromptMessageContent(data=query)]
                 for file in files:
-                    prompt_message_contents.append(file_manager.to_prompt_message_content(file))
+                    prompt_message_contents.append(
+                        file_manager.to_prompt_message_content(file, image_detail_config=image_detail_config)
+                    )
 
                 prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
         elif query:

api/core/prompt/simple_prompt_transform.py

@@ -9,6 +9,7 @@ from core.app.entities.app_invoke_entities import ModelConfigWithCredentialsEnti
 from core.file import file_manager
 from core.memory.token_buffer_memory import TokenBufferMemory
 from core.model_runtime.entities.message_entities import (
+    ImagePromptMessageContent,
     PromptMessage,
     PromptMessageContent,
     SystemPromptMessage,
@@ -60,6 +61,7 @@ class SimplePromptTransform(PromptTransform):
         context: Optional[str],
         memory: Optional[TokenBufferMemory],
         model_config: ModelConfigWithCredentialsEntity,
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
     ) -> tuple[list[PromptMessage], Optional[list[str]]]:
         inputs = {key: str(value) for key, value in inputs.items()}
 
@@ -74,6 +76,7 @@ class SimplePromptTransform(PromptTransform):
                 context=context,
                 memory=memory,
                 model_config=model_config,
+                image_detail_config=image_detail_config,
             )
         else:
             prompt_messages, stops = self._get_completion_model_prompt_messages(
@@ -85,6 +88,7 @@ class SimplePromptTransform(PromptTransform):
                 context=context,
                 memory=memory,
                 model_config=model_config,
+                image_detail_config=image_detail_config,
             )
 
         return prompt_messages, stops
@@ -175,6 +179,7 @@ class SimplePromptTransform(PromptTransform):
         files: Sequence["File"],
         memory: Optional[TokenBufferMemory],
         model_config: ModelConfigWithCredentialsEntity,
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
     ) -> tuple[list[PromptMessage], Optional[list[str]]]:
         prompt_messages: list[PromptMessage] = []
 
@@ -204,9 +209,9 @@ class SimplePromptTransform(PromptTransform):
             )
 
         if query:
-            prompt_messages.append(self.get_last_user_message(query, files))
+            prompt_messages.append(self.get_last_user_message(query, files, image_detail_config))
         else:
-            prompt_messages.append(self.get_last_user_message(prompt, files))
+            prompt_messages.append(self.get_last_user_message(prompt, files, image_detail_config))
 
         return prompt_messages, None
 
@@ -220,6 +225,7 @@ class SimplePromptTransform(PromptTransform):
         files: Sequence["File"],
         memory: Optional[TokenBufferMemory],
         model_config: ModelConfigWithCredentialsEntity,
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
     ) -> tuple[list[PromptMessage], Optional[list[str]]]:
         # get prompt
         prompt, prompt_rules = self.get_prompt_str_and_rules(
@@ -262,14 +268,21 @@ class SimplePromptTransform(PromptTransform):
         if stops is not None and len(stops) == 0:
             stops = None
 
-        return [self.get_last_user_message(prompt, files)], stops
+        return [self.get_last_user_message(prompt, files, image_detail_config)], stops
 
-    def get_last_user_message(self, prompt: str, files: Sequence["File"]) -> UserPromptMessage:
+    def get_last_user_message(
+        self,
+        prompt: str,
+        files: Sequence["File"],
+        image_detail_config: Optional[ImagePromptMessageContent.DETAIL] = None,
+    ) -> UserPromptMessage:
         if files:
             prompt_message_contents: list[PromptMessageContent] = []
             prompt_message_contents.append(TextPromptMessageContent(data=prompt))
             for file in files:
-                prompt_message_contents.append(file_manager.to_prompt_message_content(file))
+                prompt_message_contents.append(
+                    file_manager.to_prompt_message_content(file, image_detail_config=image_detail_config)
+                )
 
             prompt_message = UserPromptMessage(content=prompt_message_contents)
         else:

api/core/provider_manager.py

@@ -149,6 +149,11 @@ class ProviderManager:
             provider_name = provider_entity.provider
             provider_records = provider_name_to_provider_records_dict.get(provider_entity.provider, [])
             provider_model_records = provider_name_to_provider_model_records_dict.get(provider_entity.provider, [])
+            provider_id_entity = ModelProviderID(provider_name)
+            if provider_id_entity.is_langgenius():
+                provider_model_records.extend(
+                    provider_name_to_provider_model_records_dict.get(provider_id_entity.provider_name, [])
+                )
 
             # Convert to custom configuration
             custom_configuration = self._to_custom_configuration(
@@ -190,6 +195,20 @@ class ProviderManager:
                 provider_name
             )
 
+            provider_id_entity = ModelProviderID(provider_name)
+
+            if provider_id_entity.is_langgenius():
+                if provider_model_settings is not None:
+                    provider_model_settings.extend(
+                        provider_name_to_provider_model_settings_dict.get(provider_id_entity.provider_name, [])
+                    )
+                if provider_load_balancing_configs is not None:
+                    provider_load_balancing_configs.extend(
+                        provider_name_to_provider_load_balancing_model_configs_dict.get(
+                            provider_id_entity.provider_name, []
+                        )
+                    )
+
             # Convert to model settings
             model_settings = self._to_model_settings(
                 provider_entity=provider_entity,
@@ -207,7 +226,7 @@ class ProviderManager:
                 model_settings=model_settings,
             )
 
-            provider_configurations[str(ModelProviderID(provider_name))] = provider_configuration
+            provider_configurations[str(provider_id_entity)] = provider_configuration
 
         # Return the encapsulated object
         return provider_configurations

api/core/rag/datasource/keyword/jieba/jieba.py

@@ -88,16 +88,17 @@ class Jieba(BaseKeyword):
         keyword_table = self._get_dataset_keyword_table()
 
         k = kwargs.get("top_k", 4)
-
+        document_ids_filter = kwargs.get("document_ids_filter")
         sorted_chunk_indices = self._retrieve_ids_by_query(keyword_table or {}, query, k)
 
         documents = []
         for chunk_index in sorted_chunk_indices:
-            segment = (
-                db.session.query(DocumentSegment)
-                .filter(DocumentSegment.dataset_id == self.dataset.id, DocumentSegment.index_node_id == chunk_index)
-                .first()
+            segment_query = db.session.query(DocumentSegment).filter(
+                DocumentSegment.dataset_id == self.dataset.id, DocumentSegment.index_node_id == chunk_index
             )
+            if document_ids_filter:
+                segment_query = segment_query.filter(DocumentSegment.document_id.in_(document_ids_filter))
+            segment = segment_query.first()
 
             if segment:
                 documents.append(

api/core/rag/datasource/retrieval_service.py

@@ -1,5 +1,4 @@
 import concurrent.futures
-import json
 from concurrent.futures import ThreadPoolExecutor
 from typing import Optional
 
@@ -42,6 +41,7 @@ class RetrievalService:
         reranking_model: Optional[dict] = None,
         reranking_mode: str = "reranking_model",
         weights: Optional[dict] = None,
+        document_ids_filter: Optional[list[str]] = None,
     ):
         if not query:
             return []
@@ -65,6 +65,7 @@ class RetrievalService:
                         top_k=top_k,
                         all_documents=all_documents,
                         exceptions=exceptions,
+                        document_ids_filter=document_ids_filter,
                     )
                 )
             if RetrievalMethod.is_support_semantic_search(retrieval_method):
@@ -80,6 +81,7 @@ class RetrievalService:
                         all_documents=all_documents,
                         retrieval_method=retrieval_method,
                         exceptions=exceptions,
+                        document_ids_filter=document_ids_filter,
                     )
                 )
             if RetrievalMethod.is_support_fulltext_search(retrieval_method):
@@ -131,7 +133,14 @@ class RetrievalService:
 
     @classmethod
     def keyword_search(
-        cls, flask_app: Flask, dataset_id: str, query: str, top_k: int, all_documents: list, exceptions: list
+        cls,
+        flask_app: Flask,
+        dataset_id: str,
+        query: str,
+        top_k: int,
+        all_documents: list,
+        exceptions: list,
+        document_ids_filter: Optional[list[str]] = None,
     ):
         with flask_app.app_context():
             try:
@@ -140,7 +149,10 @@ class RetrievalService:
                     raise ValueError("dataset not found")
 
                 keyword = Keyword(dataset=dataset)
-                documents = keyword.search(cls.escape_query_for_search(query), top_k=top_k)
+
+                documents = keyword.search(
+                    cls.escape_query_for_search(query), top_k=top_k, document_ids_filter=document_ids_filter
+                )
                 all_documents.extend(documents)
             except Exception as e:
                 exceptions.append(str(e))
@@ -157,6 +169,7 @@ class RetrievalService:
         all_documents: list,
         retrieval_method: str,
         exceptions: list,
+        document_ids_filter: Optional[list[str]] = None,
     ):
         with flask_app.app_context():
             try:
@@ -171,6 +184,7 @@ class RetrievalService:
                     top_k=top_k,
                     score_threshold=score_threshold,
                     filter={"group_id": [dataset.id]},
+                    document_ids_filter=document_ids_filter,
                 )
 
                 if documents:
@@ -243,7 +257,7 @@ class RetrievalService:
 
     @staticmethod
     def escape_query_for_search(query: str) -> str:
-        return json.dumps(query).strip('"')
+        return query.replace('"', '\\"')
 
     @classmethod
     def format_retrieval_documents(cls, documents: list[Document]) -> list[RetrievalSegments]:
@@ -277,6 +291,8 @@ class RetrievalService:
                     continue
 
                 dataset_document = dataset_documents[document_id]
+                if not dataset_document:
+                    continue
 
                 if dataset_document.doc_form == IndexType.PARENT_CHILD_INDEX:
                     # Handle parent-child documents

api/core/rag/datasource/vdb/analyticdb/analyticdb_vector.py

@@ -53,7 +53,7 @@ class AnalyticdbVector(BaseVector):
         self.analyticdb_vector.delete_by_metadata_field(key, value)
 
     def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
-        return self.analyticdb_vector.search_by_vector(query_vector)
+        return self.analyticdb_vector.search_by_vector(query_vector, **kwargs)
 
     def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
         return self.analyticdb_vector.search_by_full_text(query, **kwargs)

api/core/rag/datasource/vdb/analyticdb/analyticdb_vector_sql.py

@@ -194,6 +194,13 @@ class AnalyticdbVectorBySql:
 
     def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
         top_k = kwargs.get("top_k", 4)
+        if not isinstance(top_k, int) or top_k <= 0:
+            raise ValueError("top_k must be a positive integer")
+        document_ids_filter = kwargs.get("document_ids_filter")
+        where_clause = "WHERE 1=1"
+        if document_ids_filter:
+            document_ids = ", ".join(f"'{id}'" for id in document_ids_filter)
+            where_clause += f"AND metadata_->>'document_id' IN ({document_ids})"
         score_threshold = float(kwargs.get("score_threshold") or 0.0)
         with self._get_cursor() as cur:
             query_vector_str = json.dumps(query_vector)
@@ -202,7 +209,7 @@ class AnalyticdbVectorBySql:
                 f"SELECT t.id AS id, t.vector AS vector, (1.0 - t.score) AS score, "
                 f"t.page_content as page_content, t.metadata_ AS metadata_ "
                 f"FROM (SELECT id, vector, page_content, metadata_, vector <=> %s AS score "
-                f"FROM {self.table_name} ORDER BY score LIMIT {top_k} ) t",
+                f"FROM {self.table_name} {where_clause} ORDER BY score LIMIT {top_k} ) t",
                 (query_vector_str,),
             )
             documents = []
@@ -220,12 +227,19 @@ class AnalyticdbVectorBySql:
 
     def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
         top_k = kwargs.get("top_k", 4)
+        if not isinstance(top_k, int) or top_k <= 0:
+            raise ValueError("top_k must be a positive integer")
+        document_ids_filter = kwargs.get("document_ids_filter")
+        where_clause = ""
+        if document_ids_filter:
+            document_ids = ", ".join(f"'{id}'" for id in document_ids_filter)
+            where_clause += f"AND metadata_->>'document_id' IN ({document_ids})"
         with self._get_cursor() as cur:
             cur.execute(
                 f"""SELECT id, vector, page_content, metadata_, 
                 ts_rank(to_tsvector, to_tsquery_from_text(%s, 'zh_cn'), 32) AS score
                 FROM {self.table_name}
-                WHERE to_tsvector@@to_tsquery_from_text(%s, 'zh_cn')
+                WHERE to_tsvector@@to_tsquery_from_text(%s, 'zh_cn') {where_clause}
                 ORDER BY score DESC
                 LIMIT {top_k}""",
                 (f"'{query}'", f"'{query}'"),

api/core/rag/datasource/vdb/baidu/baidu_vector.py

@@ -123,11 +123,21 @@ class BaiduVector(BaseVector):
 
     def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
         query_vector = [float(val) if isinstance(val, np.float64) else val for val in query_vector]
-        anns = AnnSearch(
-            vector_field=self.field_vector,
-            vector_floats=query_vector,
-            params=HNSWSearchParams(ef=kwargs.get("ef", 10), limit=kwargs.get("top_k", 4)),
-        )
+        document_ids_filter = kwargs.get("document_ids_filter")
+        if document_ids_filter:
+            document_ids = ", ".join(f"'{id}'" for id in document_ids_filter)
+            anns = AnnSearch(
+                vector_field=self.field_vector,
+                vector_floats=query_vector,
+                params=HNSWSearchParams(ef=kwargs.get("ef", 10), limit=kwargs.get("top_k", 4)),
+                filter=f"document_id IN ({document_ids})",
+            )
+        else:
+            anns = AnnSearch(
+                vector_field=self.field_vector,
+                vector_floats=query_vector,
+                params=HNSWSearchParams(ef=kwargs.get("ef", 10), limit=kwargs.get("top_k", 4)),
+            )
         res = self._db.table(self._collection_name).search(
             anns=anns,
             projections=[self.field_id, self.field_text, self.field_metadata],

api/core/rag/datasource/vdb/chroma/chroma_vector.py

@@ -95,7 +95,15 @@ class ChromaVector(BaseVector):
 
     def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
         collection = self._client.get_or_create_collection(self._collection_name)
-        results: QueryResult = collection.query(query_embeddings=query_vector, n_results=kwargs.get("top_k", 4))
+        document_ids_filter = kwargs.get("document_ids_filter")
+        if document_ids_filter:
+            results: QueryResult = collection.query(
+                query_embeddings=query_vector,
+                n_results=kwargs.get("top_k", 4),
+                where={"document_id": {"$in": document_ids_filter}},  # type: ignore
+            )
+        else:
+            results: QueryResult = collection.query(query_embeddings=query_vector, n_results=kwargs.get("top_k", 4))  # type: ignore
         score_threshold = float(kwargs.get("score_threshold") or 0.0)
 
         # Check if results contain data

api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py

@@ -117,6 +117,9 @@ class ElasticSearchVector(BaseVector):
         top_k = kwargs.get("top_k", 4)
         num_candidates = math.ceil(top_k * 1.5)
         knn = {"field": Field.VECTOR.value, "query_vector": query_vector, "k": top_k, "num_candidates": num_candidates}
+        document_ids_filter = kwargs.get("document_ids_filter")
+        if document_ids_filter:
+            knn["filter"] = {"terms": {"metadata.document_id": document_ids_filter}}
 
         results = self._client.search(index=self._collection_name, knn=knn, size=top_k)
 
@@ -145,6 +148,9 @@ class ElasticSearchVector(BaseVector):
 
     def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
         query_str = {"match": {Field.CONTENT_KEY.value: query}}
+        document_ids_filter = kwargs.get("document_ids_filter")
+        if document_ids_filter:
+            query_str["filter"] = {"terms": {"metadata.document_id": document_ids_filter}}  # type: ignore
         results = self._client.search(index=self._collection_name, query=query_str, size=kwargs.get("top_k", 4))
         docs = []
         for hit in results["hits"]["hits"]:

api/core/rag/datasource/vdb/lindorm/lindorm_vector.py

@@ -168,7 +168,12 @@ class LindormVectorStore(BaseVector):
             raise ValueError("All elements in query_vector should be floats")
 
         top_k = kwargs.get("top_k", 10)
-        query = default_vector_search_query(query_vector=query_vector, k=top_k, **kwargs)
+        document_ids_filter = kwargs.get("document_ids_filter")
+        filters = []
+        if document_ids_filter:
+            filters.append({"terms": {"metadata.document_id": document_ids_filter}})
+        query = default_vector_search_query(query_vector=query_vector, k=top_k, filters=filters, **kwargs)
+
         try:
             params = {}
             if self._using_ugc:
@@ -206,7 +211,10 @@ class LindormVectorStore(BaseVector):
         should = kwargs.get("should")
         minimum_should_match = kwargs.get("minimum_should_match", 0)
         top_k = kwargs.get("top_k", 10)
-        filters = kwargs.get("filter")
+        filters = kwargs.get("filter", [])
+        document_ids_filter = kwargs.get("document_ids_filter")
+        if document_ids_filter:
+            filters.append({"terms": {"metadata.document_id": document_ids_filter}})
         routing = self._routing
         full_text_query = default_text_search_query(
             query_text=query,

api/core/rag/datasource/vdb/milvus/milvus_vector.py

@@ -228,12 +228,18 @@ class MilvusVector(BaseVector):
         """
         Search for documents by vector similarity.
         """
+        document_ids_filter = kwargs.get("document_ids_filter")
+        filter = ""
+        if document_ids_filter:
+            document_ids = ", ".join(f"'{id}'" for id in document_ids_filter)
+            filter = f'metadata["document_id"] in ({document_ids})'
         results = self._client.search(
             collection_name=self._collection_name,
             data=[query_vector],
             anns_field=Field.VECTOR.value,
             limit=kwargs.get("top_k", 4),
             output_fields=[Field.CONTENT_KEY.value, Field.METADATA_KEY.value],
+            filter=filter,
         )
 
         return self._process_search_results(
@@ -249,6 +255,11 @@ class MilvusVector(BaseVector):
         if not self._hybrid_search_enabled or not self.field_exists(Field.SPARSE_VECTOR.value):
             logger.warning("Full-text search is not supported in current Milvus version (requires >= 2.5.0)")
             return []
+        document_ids_filter = kwargs.get("document_ids_filter")
+        filter = ""
+        if document_ids_filter:
+            document_ids = ", ".join(f"'{id}'" for id in document_ids_filter)
+            filter = f'metadata["document_id"] in ({document_ids})'
 
         results = self._client.search(
             collection_name=self._collection_name,
@@ -256,6 +267,7 @@ class MilvusVector(BaseVector):
             anns_field=Field.SPARSE_VECTOR.value,
             limit=kwargs.get("top_k", 4),
             output_fields=[Field.CONTENT_KEY.value, Field.METADATA_KEY.value],
+            filter=filter,
         )
 
         return self._process_search_results(