Merge branch 'main' into hy/type_hints_all

api/.ruff.toml

@@ -70,7 +70,6 @@ ignore = [
     "SIM113", # eumerate-for-loop
     "SIM117", # multiple-with-statements
     "SIM210", # if-expr-with-true-false
-    "SIM300", # yoda-conditions,
 ]
 
 [lint.per-file-ignores]

api/controllers/console/admin.py

@@ -31,7 +31,7 @@ def decorated(*args, **kwargs):
         if auth_scheme != "bearer":
             raise Unauthorized("Invalid Authorization header format. Expected 'Bearer <api-key>' format.")
 
-        if dify_config.ADMIN_API_KEY != auth_token:
+        if auth_token != dify_config.ADMIN_API_KEY:
             raise Unauthorized("API key is invalid.")
 
         return view(*args, **kwargs)

api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py

@@ -119,7 +119,7 @@ def _generate_embeddings_by_text_input_key(
                 embeddings.append(result[0].get("embedding"))
 
             return [list(map(float, e)) for e in embeddings]
-        elif "texts" == text_input_key:
+        elif text_input_key == "texts":
             result = client.run(
                 replicate_model_version,
                 input={

api/core/model_runtime/model_providers/siliconflow/siliconflow.py

@@ -18,7 +18,7 @@ def validate_provider_credentials(self, credentials: dict) -> None:
         try:
             model_instance = self.get_model_instance(ModelType.LLM)
 
-            model_instance.validate_credentials(model="deepseek-ai/DeepSeek-V2-Chat", credentials=credentials)
+            model_instance.validate_credentials(model="deepseek-ai/DeepSeek-V2.5", credentials=credentials)
         except CredentialsValidateFailedError as ex:
             raise ex
         except Exception as ex:

api/core/model_runtime/model_providers/volcengine_maas/llm/models.py

@@ -42,6 +42,10 @@ class ModelConfig(BaseModel):
         properties=ModelProperties(context_size=32768, max_tokens=4096, mode=LLMMode.CHAT),
         features=[ModelFeature.TOOL_CALL],
     ),
+    "Doubao-pro-256k": ModelConfig(
+        properties=ModelProperties(context_size=262144, max_tokens=4096, mode=LLMMode.CHAT),
+        features=[],
+    ),
     "Doubao-pro-128k": ModelConfig(
         properties=ModelProperties(context_size=131072, max_tokens=4096, mode=LLMMode.CHAT),
         features=[ModelFeature.TOOL_CALL],

api/core/model_runtime/model_providers/volcengine_maas/text_embedding/models.py

@@ -12,6 +12,7 @@ class ModelConfig(BaseModel):
 
 ModelConfigs = {
     "Doubao-embedding": ModelConfig(properties=ModelProperties(context_size=4096, max_chunks=32)),
+    "Doubao-embedding-large": ModelConfig(properties=ModelProperties(context_size=4096, max_chunks=32)),
 }
 
 
@@ -21,7 +22,7 @@ def get_model_config(credentials: dict) -> ModelConfig:
     if not model_configs:
         return ModelConfig(
             properties=ModelProperties(
-                context_size=int(credentials.get("context_size", 0)),
+                context_size=int(credentials.get("context_size", 4096)),
                 max_chunks=int(credentials.get("max_chunks", 1)),
             )
         )

api/core/model_runtime/model_providers/volcengine_maas/volcengine_maas.yaml

@@ -166,6 +166,12 @@ model_credential_schema:
           show_on:
             - variable: __model_type
               value: llm
+        - label:
+            en_US: Doubao-pro-256k
+          value: Doubao-pro-256k
+          show_on:
+            - variable: __model_type
+              value: llm
         - label:
             en_US: Llama3-8B
           value: Llama3-8B
@@ -220,6 +226,12 @@ model_credential_schema:
           show_on:
             - variable: __model_type
               value: text-embedding
+        - label:
+            en_US: Doubao-embedding-large
+          value: Doubao-embedding-large
+          show_on:
+            - variable: __model_type
+              value: text-embedding
         - label:
             en_US: Custom
             zh_Hans: 自定义

api/core/rag/embedding/cached_embedding.py

@@ -64,8 +64,12 @@ def embed_documents(self, texts: list[str]) -> list[list[float]]:
 
                     for vector in embedding_result.embeddings:
                         try:
-                            vector_array = np.array(vector)
-                            normalized_embedding = (vector_array / np.linalg.norm(vector_array)).tolist()
+                            normalized_embedding = (vector / np.linalg.norm(vector)).tolist()
+                            # stackoverflow best way: https://stackoverflow.com/questions/20319813/how-to-check-list-containing-nan
+                            if np.isnan(normalized_embedding).any():
+                                # for issue #11827  float values are not json compliant
+                                logger.warning(f"Normalized embedding is nan: {normalized_embedding}")
+                                continue
                             embedding_queue_embeddings.append(normalized_embedding)
                         except IntegrityError:
                             db.session.rollback()

api/core/tools/provider/builtin/comfyui/comfyui.py

@@ -11,7 +11,10 @@ class ComfyUIProvider(BuiltinToolProviderController):
     def _validate_credentials(self, credentials: dict[str, Any]) -> None:
         ws = websocket.WebSocket()
         base_url = URL(credentials.get("base_url"))
-        ws_address = f"ws://{base_url.authority}/ws?clientId=test123"
+        ws_protocol = "ws"
+        if base_url.scheme == "https":
+            ws_protocol = "wss"
+        ws_address = f"{ws_protocol}://{base_url.authority}/ws?clientId=test123"
 
         try:
             ws.connect(ws_address)

api/core/workflow/nodes/http_request/node.py

@@ -1,4 +1,5 @@
 import logging
+import mimetypes
 from collections.abc import Mapping, Sequence
 from typing import Any, Optional
 
@@ -156,20 +157,24 @@ def _extract_variable_selector_to_variable_mapping(
 
     def extract_files(self, url: str, response: Response) -> list[File]:
         """
-        Extract files from response
+        Extract files from response by checking both Content-Type header and URL
         """
         files = []
         is_file = response.is_file
         content_type = response.content_type
         content = response.content
 
-        if is_file and content_type:
+        if is_file:
+            # Guess file extension from URL or Content-Type header
+            filename = url.split("?")[0].split("/")[-1] or ""
+            mime_type = content_type or mimetypes.guess_type(filename)[0] or "application/octet-stream"
+
             tool_file = ToolFileManager.create_file_by_raw(
                 user_id=self.user_id,
                 tenant_id=self.tenant_id,
                 conversation_id=None,
                 file_binary=content,
-                mimetype=content_type,
+                mimetype=mime_type,
             )
 
             mapping = {

api/tests/integration_tests/model_runtime/__mock/xinference.py

@@ -21,13 +21,13 @@ def get_chat_model(self: Client, model_uid: str) -> Union[RESTfulGenerateModelHa
         if not re.match(r"https?:\/\/[^\s\/$.?#].[^\s]*$", self.base_url):
             raise RuntimeError("404 Not Found")
 
-        if "generate" == model_uid:
+        if model_uid == "generate":
             return RESTfulGenerateModelHandle(model_uid, base_url=self.base_url, auth_headers={})
-        if "chat" == model_uid:
+        if model_uid == "chat":
             return RESTfulChatModelHandle(model_uid, base_url=self.base_url, auth_headers={})
-        if "embedding" == model_uid:
+        if model_uid == "embedding":
             return RESTfulEmbeddingModelHandle(model_uid, base_url=self.base_url, auth_headers={})
-        if "rerank" == model_uid:
+        if model_uid == "rerank":
             return RESTfulRerankModelHandle(model_uid, base_url=self.base_url, auth_headers={})
         raise RuntimeError("404 Not Found")
 

api/tests/integration_tests/tools/api_tool/test_api_tool.py

@@ -34,9 +34,9 @@ def test_api_tool(setup_http_mock):
     response = tool.do_http_request(tool.api_bundle.server_url, tool.api_bundle.method, headers, parameters)
 
     assert response.status_code == 200
-    assert "/p_param" == response.request.url.path
-    assert b"query_param=q_param" == response.request.url.query
-    assert "h_param" == response.request.headers.get("header_param")
-    assert "application/json" == response.request.headers.get("content-type")
-    assert "cookie_param=c_param" == response.request.headers.get("cookie")
+    assert response.request.url.path == "/p_param"
+    assert response.request.url.query == b"query_param=q_param"
+    assert response.request.headers.get("header_param") == "h_param"
+    assert response.request.headers.get("content-type") == "application/json"
+    assert response.request.headers.get("cookie") == "cookie_param=c_param"
     assert "b_param" in response.content.decode()

api/tests/integration_tests/workflow/nodes/test_http.py

@@ -384,7 +384,7 @@ def test_mock_404(setup_http_mock):
     assert result.outputs is not None
     resp = result.outputs
 
-    assert 404 == resp.get("status_code")
+    assert resp.get("status_code") == 404
     assert "Not Found" in resp.get("body", "")
 
 

web/app/components/tools/add-tool-modal/empty.tsx

@@ -1,13 +1,19 @@
+'use client'
+import { useSearchParams } from 'next/navigation'
 import { useTranslation } from 'react-i18next'
-
 const Empty = () => {
   const { t } = useTranslation()
+  const searchParams = useSearchParams()
 
   return (
     <div className='flex flex-col items-center'>
       <div className="shrink-0 w-[163px] h-[149px] bg-cover bg-no-repeat bg-[url('~@/app/components/tools/add-tool-modal/empty.png')]"></div>
-      <div className='mb-1 text-[13px] font-medium text-text-primary leading-[18px]'>{t('tools.addToolModal.emptyTitle')}</div>
-      <div className='text-[13px] text-text-tertiary leading-[18px]'>{t('tools.addToolModal.emptyTip')}</div>
+      <div className='mb-1 text-[13px] font-medium text-text-primary leading-[18px]'>
+        {t(`tools.addToolModal.${searchParams.get('category') === 'workflow' ? 'emptyTitle' : 'emptyTitleCustom'}`)}
+      </div>
+      <div className='text-[13px] text-text-tertiary leading-[18px]'>
+        {t(`tools.addToolModal.${searchParams.get('category') === 'workflow' ? 'emptyTip' : 'emptyTipCustom'}`)}
+      </div>
     </div>
   )
 }

web/i18n/en-US/tools.ts

@@ -31,6 +31,8 @@ const translation = {
     manageInTools: 'Manage in Tools',
     emptyTitle: 'No workflow tool available',
     emptyTip: 'Go to "Workflow -> Publish as Tool"',
+    emptyTitleCustom: 'No custom tool available',
+    emptyTipCustom: 'Create a custom tool',
   },
   createTool: {
     title: 'Create Custom Tool',

web/i18n/zh-Hans/tools.ts

@@ -31,6 +31,8 @@ const translation = {
     manageInTools: '去工具列表管理',
     emptyTitle: '没有可用的工作流工具',
     emptyTip: '去 “工作流 -> 发布为工具” 添加',
+    emptyTitleCustom: '没有可用的自定义工具',
+    emptyTipCustom: '创建自定义工具',
   },
   createTool: {
     title: '创建自定义工具',