vectorvein 0.3.17__py3-none-any.whl → 0.3.19__py3-none-any.whl

vectorvein/chat_clients/utils.py

@@ -21,7 +21,6 @@ from ..types.llm_parameters import (
     NOT_GIVEN,
     ToolParam,
     VectorVeinMessage,
-    VectorVeinTextMessage,
     VectorVeinWorkflowMessage,
 )
 
@@ -159,7 +158,19 @@ def get_token_counts(text: str | dict, model: str = "", use_token_server_first:
             ],
         }
 
-        _, response = Retry(httpx.post).args(url=tokenize_url, headers=headers, json=request_body, timeout=None).retry_times(5).sleep_time(10).run()
+        _, response = (
+            Retry(httpx.post)
+            .args(
+                url=tokenize_url,
+                headers=headers,
+                json=request_body,
+                timeout=None,
+                proxy=endpoint.proxy,
+            )
+            .retry_times(5)
+            .sleep_time(10)
+            .run()
+        )
         if response is None:
             return 1000
         result = response.json()
@@ -179,7 +190,19 @@ def get_token_counts(text: str | dict, model: str = "", use_token_server_first:
                 {"role": "user", "content": text},
             ],
         }
-        _, response = Retry(httpx.post).args(url=tokenize_url, headers=headers, json=request_body, timeout=None).retry_times(5).sleep_time(10).run()
+        _, response = (
+            Retry(httpx.post)
+            .args(
+                url=tokenize_url,
+                headers=headers,
+                json=request_body,
+                timeout=None,
+                proxy=endpoint.proxy,
+            )
+            .retry_times(5)
+            .sleep_time(10)
+            .run()
+        )
         if response is None:
             return 1000
         result = response.json()
@@ -203,7 +226,19 @@ def get_token_counts(text: str | dict, model: str = "", use_token_server_first:
                 ],
             },
         }
-        _, response = Retry(httpx.post).args(base_url, json=request_body, params=params, timeout=None).retry_times(5).sleep_time(10).run()
+        _, response = (
+            Retry(httpx.post)
+            .args(
+                base_url,
+                json=request_body,
+                params=params,
+                timeout=None,
+                proxy=endpoint.proxy,
+            )
+            .retry_times(5)
+            .sleep_time(10)
+            .run()
+        )
         if response is None:
             return 1000
         result = response.json()
@@ -227,10 +262,12 @@ def get_token_counts(text: str | dict, model: str = "", use_token_server_first:
                 if endpoint.is_vertex or endpoint.is_bedrock or endpoint.endpoint_type == "anthropic_vertex" or endpoint.endpoint_type == "anthropic_bedrock":
                     continue
                 elif endpoint.endpoint_type in ("default", "anthropic"):
+                    http_client = httpx.Client(proxy=endpoint.proxy)
                     return (
                         Anthropic(
                             api_key=endpoint.api_key,
                             base_url=endpoint.api_base,
+                            http_client=http_client,
                         )
                         .beta.messages.count_tokens(messages=[{"role": "user", "content": text}], model=model)
                         .input_tokens
@@ -265,7 +302,19 @@ def get_token_counts(text: str | dict, model: str = "", use_token_server_first:
                 {"role": "user", "content": text},
             ],
         }
-        _, response = Retry(httpx.post).args(url=tokenize_url, headers=headers, json=request_body, timeout=None).retry_times(5).sleep_time(10).run()
+        _, response = (
+            Retry(httpx.post)
+            .args(
+                url=tokenize_url,
+                headers=headers,
+                json=request_body,
+                timeout=None,
+                proxy=endpoint.proxy,
+            )
+            .retry_times(5)
+            .sleep_time(10)
+            .run()
+        )
         if response is None:
             return 1000
         result = response.json()
@@ -287,7 +336,19 @@ def get_token_counts(text: str | dict, model: str = "", use_token_server_first:
                 {"role": "user", "content": text},
             ],
         }
-        _, response = Retry(httpx.post).args(url=tokenize_url, headers=headers, json=request_body, timeout=None).retry_times(5).sleep_time(10).run()
+        _, response = (
+            Retry(httpx.post)
+            .args(
+                url=tokenize_url,
+                headers=headers,
+                json=request_body,
+                timeout=None,
+                proxy=endpoint.proxy,
+            )
+            .retry_times(5)
+            .sleep_time(10)
+            .run()
+        )
         if response is None:
             return 1000
         result = response.json()
@@ -578,27 +639,33 @@ def transform_from_openai_message(
                         formatted_content.append(item)
             if tool_calls:
                 for tool_call in tool_calls:
-                    formatted_content.append(
-                        {
-                            "type": "tool_use",
-                            "id": tool_call["id"],
-                            "name": tool_call["function"]["name"],
-                            "input": json.loads(tool_call["function"]["arguments"]) if tool_call["function"]["arguments"] else {},
-                        }
-                    )
+                    if isinstance(tool_call, dict) and "function" in tool_call:
+                        function_info = tool_call["function"]
+                        if isinstance(function_info, dict):
+                            formatted_content.append(
+                                {
+                                    "type": "tool_use",
+                                    "id": tool_call["id"],
+                                    "name": function_info["name"],
+                                    "input": json.loads(function_info["arguments"]) if function_info.get("arguments") else {},
+                                }
+                            )
             return {"role": role, "content": formatted_content}
         else:
             if tool_calls:
                 formatted_content = [{"type": "text", "text": content}] if content else []
                 for tool_call in tool_calls:
-                    formatted_content.append(
-                        {
-                            "type": "tool_use",
-                            "id": tool_call["id"],
-                            "name": tool_call["function"]["name"],
-                            "input": json.loads(tool_call["function"]["arguments"]) if tool_call["function"]["arguments"] else {},
-                        }
-                    )
+                    if isinstance(tool_call, dict) and "function" in tool_call:
+                        function_info = tool_call["function"]
+                        if isinstance(function_info, dict):
+                            formatted_content.append(
+                                {
+                                    "type": "tool_use",
+                                    "id": tool_call["id"],
+                                    "name": function_info["name"],
+                                    "input": json.loads(function_info["arguments"]) if function_info.get("arguments") else {},
+                                }
+                            )
                 return {"role": role, "content": formatted_content}
             else:
                 return {"role": role, "content": content}
@@ -676,24 +743,30 @@ def format_messages(
 
     for message in messages:
         if is_vectorvein_message(message):
-            message = cast(VectorVeinMessage, message)
-            match message["content_type"]:
+            # We know this is a VectorVein message, so we can safely cast it
+            vectorvein_message = message  # type: ignore
+            match vectorvein_message.get("content_type"):
                 case "TXT":
-                    message = cast(VectorVeinTextMessage, message)
+                    # Remove unnecessary cast
+                    pass
                 case "WKF":
-                    message = cast(VectorVeinWorkflowMessage, message)
+                    # Remove unnecessary cast
+                    pass
                 case _:
-                    raise ValueError(f"Unsupported message type: {message['content_type']}")
+                    content_type = vectorvein_message.get("content_type", "unknown")
+                    raise ValueError(f"Unsupported message type: {content_type}")
             # 处理 VectorVein 格式的消息
-            content = message["content"]["text"]
-            if message["content_type"] == "TXT":
-                message = cast(VectorVeinTextMessage, message)
-                role = "user" if message["author_type"] == "U" else "assistant"
-                formatted_message = format_text_message(content, role, message.get("attachments", []), backend, native_multimodal, process_image)
+            content_dict = vectorvein_message.get("content")
+            if isinstance(content_dict, dict) and "text" in content_dict:
+                content = content_dict["text"]
+            else:
+                content = ""
+            if vectorvein_message.get("content_type") == "TXT":
+                role = "user" if vectorvein_message.get("author_type") == "U" else "assistant"
+                formatted_message = format_text_message(content, role, vectorvein_message.get("attachments", []), backend, native_multimodal, process_image)
                 formatted_messages.append(formatted_message)
-            elif message["content_type"] == "WKF" and message["status"] in ("S", "R"):
-                message = cast(VectorVeinWorkflowMessage, message)
-                formatted_messages.extend(format_workflow_messages(message, content, backend))
+            elif vectorvein_message.get("content_type") == "WKF" and vectorvein_message.get("status") in ("S", "R"):
+                formatted_messages.extend(format_workflow_messages(vectorvein_message, content, backend))  # type: ignore
         else:
             # 处理 OpenAI 格式的消息
             message = cast(ChatCompletionMessageParam, message)

vectorvein/settings/__init__.py

@@ -250,7 +250,7 @@ class Settings(BaseModel):
                 delattr(self, backend_name)
 
         # Set version to 2
-        self.VERSION = "2"
+        self.VERSION = "2"  # type: ignore
 
         return self
 

vectorvein/types/defaults.py

@@ -90,6 +90,13 @@ MOONSHOT_MODELS: Final[dict[str, ModelSettingDict]] = {
         "response_format_available": True,
         "native_multimodal": False,
     },
+    "kimi-k2-turbo-preview": {
+        "id": "kimi-k2-turbo-preview",
+        "context_length": 131072,
+        "function_call_available": True,
+        "response_format_available": True,
+        "native_multimodal": False,
+    },
 }
 
 # Deepseek models

vectorvein/workflow/nodes/media_editing.py

@@ -729,7 +729,26 @@ class FfmpegProcess(Node):
                     name="input_files",
                     port_type=PortType.FILE,
                     value=[],
-                    support_file_types=[".mp4", ".avi", ".mov", ".mkv", ".mp3", ".wav", ".ogg", ".m4a", ".jpg", ".jpeg", ".png", ".gif", ".bmp", ".webp", ".webm", ".flv", ".wmv", ".3gp"],
+                    support_file_types=[
+                        ".mp4",
+                        ".avi",
+                        ".mov",
+                        ".mkv",
+                        ".mp3",
+                        ".wav",
+                        ".ogg",
+                        ".m4a",
+                        ".jpg",
+                        ".jpeg",
+                        ".png",
+                        ".gif",
+                        ".bmp",
+                        ".webp",
+                        ".webm",
+                        ".flv",
+                        ".wmv",
+                        ".3gp",
+                    ],
                     multiple=True,
                     required=True,
                     show=True,

vectorvein/workflow/nodes/triggers.py

@@ -26,7 +26,6 @@ class ScheduleTrigger(Node):
             category="triggers",
             task_name="triggers.schedule_trigger",
             node_id=id,
-            has_inputs=False,
             ports={
                 "schedule": InputPort(
                     name="schedule",