Merge branch 'main' into feat/node-execution-retry

.github/workflows/api-tests.yml

@@ -50,6 +50,9 @@ jobs:
       - name: Run ModelRuntime
         run: poetry run -C api bash dev/pytest/pytest_model_runtime.sh
 
+      - name: Run dify config tests
+        run: poetry run -C api python dev/pytest/pytest_config_tests.py
+
       - name: Run Tool
         run: poetry run -C api bash dev/pytest/pytest_tools.sh
 

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/configs/packaging/__init__.py

@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="0.14.0",
+        default="0.14.1",
     )
 
     COMMIT_SHA: str = Field(

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/controllers/console/explore/recommended_app.py

@@ -13,6 +13,7 @@
     "name": fields.String,
     "mode": fields.String,
     "icon": fields.String,
+    "icon_type": fields.String,
     "icon_url": AppIconUrlField,
     "icon_background": fields.String,
 }

api/core/model_runtime/model_providers/azure_openai/_constant.py

@@ -819,6 +819,82 @@ class AzureBaseModel(BaseModel):
             ),
         ),
     ),
+    AzureBaseModel(
+        base_model_name="gpt-4o-2024-11-20",
+        entity=AIModelEntity(
+            model="fake-deployment-name",
+            label=I18nObject(
+                en_US="fake-deployment-name-label",
+            ),
+            model_type=ModelType.LLM,
+            features=[
+                ModelFeature.AGENT_THOUGHT,
+                ModelFeature.VISION,
+                ModelFeature.MULTI_TOOL_CALL,
+                ModelFeature.STREAM_TOOL_CALL,
+            ],
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={
+                ModelPropertyKey.MODE: LLMMode.CHAT.value,
+                ModelPropertyKey.CONTEXT_SIZE: 128000,
+            },
+            parameter_rules=[
+                ParameterRule(
+                    name="temperature",
+                    **PARAMETER_RULE_TEMPLATE[DefaultParameterName.TEMPERATURE],
+                ),
+                ParameterRule(
+                    name="top_p",
+                    **PARAMETER_RULE_TEMPLATE[DefaultParameterName.TOP_P],
+                ),
+                ParameterRule(
+                    name="presence_penalty",
+                    **PARAMETER_RULE_TEMPLATE[DefaultParameterName.PRESENCE_PENALTY],
+                ),
+                ParameterRule(
+                    name="frequency_penalty",
+                    **PARAMETER_RULE_TEMPLATE[DefaultParameterName.FREQUENCY_PENALTY],
+                ),
+                _get_max_tokens(default=512, min_val=1, max_val=16384),
+                ParameterRule(
+                    name="seed",
+                    label=I18nObject(zh_Hans="种子", en_US="Seed"),
+                    type="int",
+                    help=AZURE_DEFAULT_PARAM_SEED_HELP,
+                    required=False,
+                    precision=2,
+                    min=0,
+                    max=1,
+                ),
+                ParameterRule(
+                    name="response_format",
+                    label=I18nObject(zh_Hans="回复格式", en_US="response_format"),
+                    type="string",
+                    help=I18nObject(
+                        zh_Hans="指定模型必须输出的格式", en_US="specifying the format that the model must output"
+                    ),
+                    required=False,
+                    options=["text", "json_object", "json_schema"],
+                ),
+                ParameterRule(
+                    name="json_schema",
+                    label=I18nObject(en_US="JSON Schema"),
+                    type="text",
+                    help=I18nObject(
+                        zh_Hans="设置返回的json schema，llm将按照它返回",
+                        en_US="Set a response json schema will ensure LLM to adhere it.",
+                    ),
+                    required=False,
+                ),
+            ],
+            pricing=PriceConfig(
+                input=5.00,
+                output=15.00,
+                unit=0.000001,
+                currency="USD",
+            ),
+        ),
+    ),
     AzureBaseModel(
         base_model_name="gpt-4-turbo",
         entity=AIModelEntity(

api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml

@@ -171,6 +171,12 @@ model_credential_schema:
           show_on:
             - variable: __model_type
               value: llm
+        - label:
+            en_US: gpt-4o-2024-11-20
+          value: gpt-4o-2024-11-20
+          show_on:
+            - variable: __model_type
+              value: llm
         - label:
             en_US: gpt-4-turbo
           value: gpt-4-turbo

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

@@ -92,7 +92,10 @@ def _invoke(
                 average = embeddings_batch[0]
             else:
                 average = np.average(_result, axis=0, weights=num_tokens_in_batch[i])
-            embeddings[i] = (average / np.linalg.norm(average)).tolist()
+            embedding = (average / np.linalg.norm(average)).tolist()
+            if np.isnan(embedding).any():
+                raise ValueError("Normalized embedding is nan please try again")
+            embeddings[i] = embedding
 
         # calc usage
         usage = self._calc_response_usage(model=model, credentials=credentials, tokens=used_tokens)

api/core/model_runtime/model_providers/bedrock/get_bedrock_client.py

@@ -1,11 +1,19 @@
+from collections.abc import Mapping
+
 import boto3
 from botocore.config import Config
 
+from core.model_runtime.errors.invoke import InvokeBadRequestError
+
+
+def get_bedrock_client(service_name: str, credentials: Mapping[str, str]):
+    region_name = credentials.get("aws_region")
+    if not region_name:
+        raise InvokeBadRequestError("aws_region is required")
+    client_config = Config(region_name=region_name)
+    aws_access_key_id = credentials.get("aws_access_key_id")
+    aws_secret_access_key = credentials.get("aws_secret_access_key")
 
-def get_bedrock_client(service_name, credentials=None):
-    client_config = Config(region_name=credentials["aws_region"])
-    aws_access_key_id = credentials["aws_access_key_id"]
-    aws_secret_access_key = credentials["aws_secret_access_key"]
     if aws_access_key_id and aws_secret_access_key:
         # use aksk to call bedrock
         client = boto3.client(

api/core/model_runtime/model_providers/bedrock/rerank/rerank.py

@@ -62,7 +62,10 @@ def _invoke(
                 }
             )
         modelId = model
-        region = credentials["aws_region"]
+        region = credentials.get("aws_region")
+        # region is a required field
+        if not region:
+            raise InvokeBadRequestError("aws_region is required in credentials")
         model_package_arn = f"arn:aws:bedrock:{region}::foundation-model/{modelId}"
         rerankingConfiguration = {
             "type": "BEDROCK_RERANKING_MODEL",

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

@@ -88,7 +88,10 @@ def _invoke(
                 average = embeddings_batch[0]
             else:
                 average = np.average(_result, axis=0, weights=num_tokens_in_batch[i])
-            embeddings[i] = (average / np.linalg.norm(average)).tolist()
+            embedding = (average / np.linalg.norm(average)).tolist()
+            if np.isnan(embedding).any():
+                raise ValueError("Normalized embedding is nan please try again")
+            embeddings[i] = embedding
 
         # calc usage
         usage = self._calc_response_usage(model=model, credentials=credentials, tokens=used_tokens)

api/core/model_runtime/model_providers/google/llm/_position.yaml

@@ -1,4 +1,5 @@
 - gemini-2.0-flash-exp
+- gemini-2.0-flash-thinking-exp-1219
 - gemini-1.5-pro
 - gemini-1.5-pro-latest
 - gemini-1.5-pro-001

api/core/model_runtime/model_providers/google/llm/gemini-2.0-flash-thinking-exp-1219.yaml

@@ -0,0 +1,39 @@
+model: gemini-2.0-flash-thinking-exp-1219
+label:
+  en_US: Gemini 2.0 Flash Thinking Exp 1219
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - document
+  - video
+  - audio
+model_properties:
+  mode: chat
+  context_size: 32767
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_output_tokens
+    use_template: max_tokens
+    default: 8192
+    min: 1
+    max: 8192
+  - name: json_schema
+    use_template: json_schema
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD

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

@@ -97,7 +97,10 @@ def _invoke(
                 average = embeddings_batch[0]
             else:
                 average = np.average(_result, axis=0, weights=num_tokens_in_batch[i])
-            embeddings[i] = (average / np.linalg.norm(average)).tolist()
+            embedding = (average / np.linalg.norm(average)).tolist()
+            if np.isnan(embedding).any():
+                raise ValueError("Normalized embedding is nan please try again")
+            embeddings[i] = embedding
 
         # calc usage
         usage = self._calc_response_usage(model=model, credentials=credentials, tokens=used_tokens)

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/upstage/text_embedding/text_embedding.py

@@ -100,7 +100,10 @@ def _invoke(
                 average = embeddings_batch[0]
             else:
                 average = np.average(_result, axis=0, weights=num_tokens_in_batch[i])
-            embeddings[i] = (average / np.linalg.norm(average)).tolist()
+            embedding = (average / np.linalg.norm(average)).tolist()
+            if np.isnan(embedding).any():
+                raise ValueError("Normalized embedding is nan please try again")
+            embeddings[i] = embedding
 
         usage = self._calc_response_usage(model=model, credentials=credentials, tokens=used_tokens)
 

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

@@ -40,6 +40,10 @@ 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

@@ -65,6 +65,11 @@ def embed_documents(self, texts: list[str]) -> list[list[float]]:
                     for vector in embedding_result.embeddings:
                         try:
                             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()
@@ -111,6 +116,8 @@ def embed_query(self, text: str) -> list[float]:
 
             embedding_results = embedding_result.embeddings[0]
             embedding_results = (embedding_results / np.linalg.norm(embedding_results)).tolist()
+            if np.isnan(embedding_results).any():
+                raise ValueError("Normalized embedding is nan please try again")
         except Exception as ex:
             if dify_config.DEBUG:
                 logging.exception(f"Failed to embed query text '{text[:10]}...({len(text)} chars)'")

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/tools/provider/builtin/comfyui/tools/comfyui_client.py

@@ -40,7 +40,10 @@ def queue_prompt(self, client_id: str, prompt: dict) -> str:
     def open_websocket_connection(self) -> tuple[WebSocket, str]:
         client_id = str(uuid.uuid4())
         ws = WebSocket()
-        ws_address = f"ws://{self.base_url.authority}/ws?clientId={client_id}"
+        ws_protocol = "ws"
+        if self.base_url.scheme == "https":
+            ws_protocol = "wss"
+        ws_address = f"{ws_protocol}://{self.base_url.authority}/ws?clientId={client_id}"
         ws.connect(ws_address)
         return ws, client_id