vectorvein 0.1.6__py3-none-any.whl → 0.1.8__py3-none-any.whl

vectorvein/chat_clients/anthropic_client.py

@@ -19,8 +19,8 @@ from google.auth.transport.requests import Request
 from google.auth import _helpers
 
 from ..settings import settings
-from .utils import cutoff_messages
 from ..types import defaults as defs
+from .utils import cutoff_messages, get_token_counts
 from .base_client import BaseChatClient, BaseAsyncChatClient
 from ..types.enums import ContextLengthControlType, BackendType
 from ..types.llm_parameters import ChatCompletionMessage, ChatCompletionDeltaMessage
@@ -115,7 +115,7 @@ class AnthropicChatClient(BaseChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ):
@@ -189,6 +189,18 @@ class AnthropicChatClient(BaseChatClient):
                 base_url=self.endpoint.api_base,
             )
 
+        tools_params = refactor_tool_use_params(tools) if tools else tools
+
+        if max_tokens is None:
+            max_output_tokens = self.model_setting.max_output_tokens
+            if max_output_tokens is not None:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+                max_tokens = min(max(max_tokens, 1), max_output_tokens)
+            else:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+
         response = self._client.messages.create(
             model=self.model_setting.id,
             messages=messages,
@@ -196,7 +208,7 @@ class AnthropicChatClient(BaseChatClient):
             stream=self.stream,
             temperature=self.temperature,
             max_tokens=max_tokens,
-            tools=refactor_tool_use_params(tools) if tools else tools,
+            tools=tools_params,
             tool_choice=tool_choice,
         )
 
@@ -303,7 +315,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ):
@@ -376,6 +388,19 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                 api_key=self.endpoint.api_key,
                 base_url=self.endpoint.api_base,
             )
+
+        tools_params = refactor_tool_use_params(tools) if tools else tools
+
+        if max_tokens is None:
+            max_output_tokens = self.model_setting.max_output_tokens
+            if max_output_tokens is not None:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+                max_tokens = min(max(max_tokens, 1), max_output_tokens)
+            else:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+
         response = await self._client.messages.create(
             model=self.model_setting.id,
             messages=messages,
@@ -383,7 +408,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
             stream=self.stream,
             temperature=self.temperature,
             max_tokens=max_tokens,
-            tools=refactor_tool_use_params(tools) if tools else tools,
+            tools=tools_params,
             tool_choice=tool_choice,
         )
 

vectorvein/chat_clients/base_client.py

@@ -46,7 +46,7 @@ class BaseChatClient(ABC):
         model: str | None = None,
         stream: bool = False,
         temperature: float = 0.7,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ) -> ChatCompletionMessage | Generator[ChatCompletionDeltaMessage, Any, None]:
@@ -57,7 +57,7 @@ class BaseChatClient(ABC):
         messages: list,
         model: str | None = None,
         temperature: float = 0.7,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ) -> Generator[ChatCompletionDeltaMessage, Any, None]:
@@ -107,7 +107,7 @@ class BaseAsyncChatClient(ABC):
         model: str | None = None,
         stream: bool = False,
         temperature: float = 0.7,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ) -> ChatCompletionMessage | AsyncGenerator[ChatCompletionDeltaMessage, None]:
@@ -118,7 +118,7 @@ class BaseAsyncChatClient(ABC):
         messages: list,
         model: str | None = None,
         temperature: float = 0.7,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ) -> AsyncGenerator[ChatCompletionDeltaMessage, None]:

vectorvein/chat_clients/gemini_client.py

@@ -43,7 +43,7 @@ class GeminiChatClient(BaseChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | None = None,
         tool_choice: str | None = None,
     ):
@@ -156,7 +156,16 @@ class GeminiChatClient(BaseChatClient):
                 if "text" in part:
                     result["content"] += part["text"]
                 elif "functionCall" in part:
-                    tool_calls.append(part["functionCall"])
+                    tool_call = {
+                        "index": 0,
+                        "id": "call_0",
+                        "function": {
+                            "arguments": json.dumps(part["functionCall"]["args"], ensure_ascii=False),
+                            "name": part["functionCall"]["name"],
+                        },
+                        "type": "function",
+                    }
+                    tool_calls.append(tool_call)
 
             if tool_calls:
                 result["tool_calls"] = tool_calls
@@ -194,7 +203,7 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | None = None,
         tool_choice: str | None = None,
     ):
@@ -310,7 +319,16 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
                     if "text" in part:
                         result["content"] += part["text"]
                     elif "functionCall" in part:
-                        tool_calls.append(part["functionCall"])
+                        tool_call = {
+                            "index": 0,
+                            "id": "call_0",
+                            "function": {
+                                "arguments": json.dumps(part["functionCall"]["args"], ensure_ascii=False),
+                                "name": part["functionCall"]["name"],
+                            },
+                            "type": "function",
+                        }
+                        tool_calls.append(tool_call)
 
                 if tool_calls:
                     result["tool_calls"] = tool_calls

vectorvein/chat_clients/minimax_client.py

@@ -7,8 +7,8 @@ import httpx
 from openai._types import NotGiven
 
 from ..settings import settings
-from .utils import cutoff_messages
 from ..types import defaults as defs
+from .utils import cutoff_messages, get_token_counts
 from .base_client import BaseChatClient, BaseAsyncChatClient
 from ..types.enums import ContextLengthControlType, BackendType
 from ..types.llm_parameters import ChatCompletionMessage, ChatCompletionDeltaMessage
@@ -66,7 +66,7 @@ class MiniMaxChatClient(BaseChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2048,
+        max_tokens: int | None = None,
         tools: list | None = None,
         tool_choice: str = "auto",
     ):
@@ -113,6 +113,16 @@ class MiniMaxChatClient(BaseChatClient):
         else:
             tools_params = {}
 
+        if max_tokens is None:
+            max_output_tokens = self.model_setting.max_output_tokens
+            if max_output_tokens is not None:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+                max_tokens = min(max(max_tokens, 1), max_output_tokens)
+            else:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+
         self.url = self.endpoint.api_base
         self.headers = {"Authorization": f"Bearer {self.endpoint.api_key}", "Content-Type": "application/json"}
 
@@ -145,9 +155,11 @@ class MiniMaxChatClient(BaseChatClient):
                             if "usage" not in chunk_data:
                                 continue
                             else:
+                                if chunk_data["object"] != "chat.completion.chunk":
+                                    continue
                                 yield ChatCompletionDeltaMessage(
                                     **{
-                                        "content": chunk_data["choices"][0]["message"].get("content"),
+                                        "content": chunk_data["choices"][0]["delta"].get("content"),
                                         "role": "assistant",
                                         **tool_calls_params,
                                     }
@@ -211,7 +223,7 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2048,
+        max_tokens: int | None = None,
         tools: list | None = None,
         tool_choice: str = "auto",
     ):
@@ -256,6 +268,16 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
         else:
             tools_params = {}
 
+        if max_tokens is None:
+            max_output_tokens = self.model_setting.max_output_tokens
+            if max_output_tokens is not None:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+                max_tokens = min(max(max_tokens, 1), max_output_tokens)
+            else:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+
         self.url = self.endpoint.api_base
         self.headers = {"Authorization": f"Bearer {self.endpoint.api_key}", "Content-Type": "application/json"}
 
@@ -289,9 +311,11 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
                                 if "usage" not in chunk_data:
                                     continue
                                 else:
+                                    if chunk_data["object"] != "chat.completion.chunk":
+                                        continue
                                     yield ChatCompletionDeltaMessage(
                                         **{
-                                            "content": chunk_data["choices"][0]["message"].get("content"),
+                                            "content": chunk_data["choices"][0]["delta"].get("content"),
                                             "role": "assistant",
                                             **tool_calls_params,
                                         }

vectorvein/chat_clients/openai_compatible_client.py

@@ -11,6 +11,7 @@ from openai import OpenAI, AsyncOpenAI, AzureOpenAI, AsyncAzureOpenAI
 from .base_client import BaseChatClient, BaseAsyncChatClient
 from .utils import (
     cutoff_messages,
+    get_token_counts,
     ToolCallContentProcessor,
     generate_tool_use_system_prompt,
 )
@@ -50,7 +51,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ):
@@ -102,6 +103,16 @@ class OpenAICompatibleChatClient(BaseChatClient):
         else:
             tools_params = {}
 
+        if max_tokens is None:
+            max_output_tokens = self.model_setting.max_output_tokens
+            if max_output_tokens is not None:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+                max_tokens = min(max(max_tokens, 1), max_output_tokens)
+            else:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+
         response: ChatCompletion | Stream[ChatCompletionChunk] = self._client.chat.completions.create(
             model=self.model_setting.id,
             messages=messages,
@@ -122,6 +133,9 @@ class OpenAICompatibleChatClient(BaseChatClient):
                     if not chunk.choices[0].delta:
                         continue
                     if self.model_setting.function_call_available:
+                        if chunk.choices[0].delta.tool_calls:
+                            for index, tool_call in enumerate(chunk.choices[0].delta.tool_calls):
+                                tool_call.index = index
                         yield ChatCompletionDeltaMessage(**chunk.choices[0].delta.model_dump())
                     else:
                         message = chunk.choices[0].delta.model_dump()
@@ -147,7 +161,8 @@ class OpenAICompatibleChatClient(BaseChatClient):
             if tools:
                 if self.model_setting.function_call_available and response.choices[0].message.tool_calls:
                     result["tool_calls"] = [
-                        tool_call.model_dump() for tool_call in response.choices[0].message.tool_calls
+                        {**tool_call.model_dump(), "type": "function"}
+                        for tool_call in response.choices[0].message.tool_calls
                     ]
                 else:
                     tool_call_content_processor = ToolCallContentProcessor(result["content"])
@@ -189,7 +204,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
         model: str | None = None,
         stream: bool | None = None,
         temperature: float | None = None,
-        max_tokens: int = 2000,
+        max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
     ):
@@ -241,6 +256,16 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
         else:
             tools_params = {}
 
+        if max_tokens is None:
+            max_output_tokens = self.model_setting.max_output_tokens
+            if max_output_tokens is not None:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+                max_tokens = min(max(max_tokens, 1), max_output_tokens)
+            else:
+                token_counts = get_token_counts({"messages": messages, "tools_params": tools_params})
+                max_tokens = self.model_setting.context_length - token_counts
+
         response: ChatCompletion | AsyncStream[ChatCompletionChunk] = await self._client.chat.completions.create(
             model=self.model_setting.id,
             messages=messages,
@@ -261,6 +286,9 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
                     if not chunk.choices[0].delta:
                         continue
                     if self.model_setting.function_call_available:
+                        if chunk.choices[0].delta.tool_calls:
+                            for index, tool_call in enumerate(chunk.choices[0].delta.tool_calls):
+                                tool_call.index = index
                         yield ChatCompletionDeltaMessage(**chunk.choices[0].delta.model_dump())
                     else:
                         message = chunk.choices[0].delta.model_dump()
@@ -286,7 +314,8 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
             if tools:
                 if self.model_setting.function_call_available and response.choices[0].message.tool_calls:
                     result["tool_calls"] = [
-                        tool_call.model_dump() for tool_call in response.choices[0].message.tool_calls
+                        {**tool_call.model_dump(), "type": "function"}
+                        for tool_call in response.choices[0].message.tool_calls
                     ]
                 else:
                     tool_call_content_processor = ToolCallContentProcessor(result["content"])

vectorvein/chat_clients/utils.py

@@ -3,8 +3,14 @@
 import re
 import json
 
+import httpx
 import tiktoken
+from anthropic import Anthropic
+from qwen_tokenizer import qwen_tokenizer
+from deepseek_tokenizer import deepseek_tokenizer
 
+from ..settings import settings
+from ..utilities.retry import Retry
 from ..types.enums import BackendType
 from ..utilities.media_processing import ImageProcessor
 
@@ -95,10 +101,88 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
         text = str(text)
     if model == "gpt-3.5-turbo":
         return len(chatgpt_encoding.encode(text))
-    elif model == "gpt-4o":
+    elif model in ("gpt-4o", "gpt-4o-mini"):
         return len(gpt_4o_encoding.encode(text))
     elif model.startswith("abab"):
-        return int(len(text) / 1.33)
+        model_setting = settings.minimax.models[model]
+        if len(model_setting.endpoints) == 0:
+            return int(len(text) / 1.33)
+        endpoint_id = model_setting.endpoints[0]
+        endpoint = settings.get_endpoint(endpoint_id)
+        tokenize_url = "https://api.minimax.chat/v1/tokenize"
+        headers = {"Authorization": f"Bearer {endpoint.api_key}", "Content-Type": "application/json"}
+        request_body = {
+            "model": model,
+            "tokens_to_generate": 128,
+            "temperature": 0.2,
+            "messages": [
+                {"sender_type": "USER", "text": text},
+            ],
+        }
+
+        _, response = (
+            Retry(httpx.post)
+            .args(url=tokenize_url, headers=headers, json=request_body, timeout=None)
+            .retry_times(5)
+            .sleep_time(10)
+            .run()
+        )
+        response = response.json()
+        return response["segments_num"]
+    elif model in ("moonshot-v1-8k", "moonshot-v1-32k", "moonshot-v1-128k"):
+        model_setting = settings.moonshot.models[model]
+        if len(model_setting.endpoints) == 0:
+            return len(chatgpt_encoding.encode(text))
+        endpoint_id = model_setting.endpoints[0]
+        endpoint = settings.get_endpoint(endpoint_id)
+        tokenize_url = "https://api.moonshot.cn/v1/tokenizers/estimate-token-count"
+        headers = {"Content-Type": "application/json", "Authorization": f"Bearer {endpoint.api_key}"}
+        request_body = {
+            "model": model,
+            "messages": [
+                {"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 = response.json()
+        return response["data"]["total_tokens"]
+    elif model.startswith("gemini"):
+        model_setting = settings.gemini.models[model]
+        if len(model_setting.endpoints) == 0:
+            return len(chatgpt_encoding.encode(text))
+        endpoint_id = model_setting.endpoints[0]
+        endpoint = settings.get_endpoint(endpoint_id)
+        url = f"{endpoint.api_base}/models/{model_setting.id}:countTokens"
+        params = {"key": endpoint.api_key}
+        request_body = {
+            "contents": {
+                "role": "USER",
+                "parts": [
+                    {"text": "TEXT"},
+                ],
+            },
+        }
+        _, response = (
+            Retry(httpx.post)
+            .args(url, json=request_body, params=params, timeout=None)
+            .retry_times(5)
+            .sleep_time(10)
+            .run()
+        )
+        result = response.json()
+        return result["totalTokens"]
+    elif model.startswith("claude"):
+        return Anthropic().count_tokens(text)
+    elif model.startswith("deepseek"):
+        return len(deepseek_tokenizer.encode(text))
+    elif model.startswith("qwen"):
+        return len(qwen_tokenizer.encode(text))
     else:
         return len(chatgpt_encoding.encode(text))
 

vectorvein/settings/__init__.py

@@ -14,41 +14,39 @@ class Settings(BaseModel):
         default_factory=list, description="Available endpoints for the LLM service."
     )
 
-    anthropic_models: BackendSettings = Field(
-        default_factory=BackendSettings, description="Anthropic models settings."
-    )
-    deepseek_models: BackendSettings = Field(default_factory=BackendSettings, description="Deepseek models settings.")
-    gemini_models: BackendSettings = Field(default_factory=BackendSettings, description="Gemini models settings.")
-    groq_models: BackendSettings = Field(default_factory=BackendSettings, description="Groq models settings.")
-    local_models: BackendSettings = Field(default_factory=BackendSettings, description="Local models settings.")
-    minimax_models: BackendSettings = Field(default_factory=BackendSettings, description="Minimax models settings.")
-    mistral_models: BackendSettings = Field(default_factory=BackendSettings, description="Mistral models settings.")
-    moonshot_models: BackendSettings = Field(default_factory=BackendSettings, description="Moonshot models settings.")
-    openai_models: BackendSettings = Field(default_factory=BackendSettings, description="OpenAI models settings.")
-    qwen_models: BackendSettings = Field(default_factory=BackendSettings, description="Qwen models settings.")
-    yi_models: BackendSettings = Field(default_factory=BackendSettings, description="Yi models settings.")
-    zhipuai_models: BackendSettings = Field(default_factory=BackendSettings, description="Zhipuai models settings.")
+    anthropic: BackendSettings = Field(default_factory=BackendSettings, description="Anthropic models settings.")
+    deepseek: BackendSettings = Field(default_factory=BackendSettings, description="Deepseek models settings.")
+    gemini: BackendSettings = Field(default_factory=BackendSettings, description="Gemini models settings.")
+    groq: BackendSettings = Field(default_factory=BackendSettings, description="Groq models settings.")
+    local: BackendSettings = Field(default_factory=BackendSettings, description="Local models settings.")
+    minimax: BackendSettings = Field(default_factory=BackendSettings, description="Minimax models settings.")
+    mistral: BackendSettings = Field(default_factory=BackendSettings, description="Mistral models settings.")
+    moonshot: BackendSettings = Field(default_factory=BackendSettings, description="Moonshot models settings.")
+    openai: BackendSettings = Field(default_factory=BackendSettings, description="OpenAI models settings.")
+    qwen: BackendSettings = Field(default_factory=BackendSettings, description="Qwen models settings.")
+    yi: BackendSettings = Field(default_factory=BackendSettings, description="Yi models settings.")
+    zhipuai: BackendSettings = Field(default_factory=BackendSettings, description="Zhipuai models settings.")
 
     def __init__(self, **data):
         model_types = {
-            "anthropic_models": defs.ANTHROPIC_MODELS,
-            "deepseek_models": defs.DEEPSEEK_MODELS,
-            "gemini_models": defs.GEMINI_MODELS,
-            "groq_models": defs.GROQ_MODELS,
-            "local_models": {},
-            "minimax_models": defs.MINIMAX_MODELS,
-            "mistral_models": defs.MISTRAL_MODELS,
-            "moonshot_models": defs.MOONSHOT_MODELS,
-            "openai_models": defs.OPENAI_MODELS,
-            "qwen_models": defs.QWEN_MODELS,
-            "yi_models": defs.YI_MODELS,
-            "zhipuai_models": defs.ZHIPUAI_MODELS,
+            "anthropic": defs.ANTHROPIC_MODELS,
+            "deepseek": defs.DEEPSEEK_MODELS,
+            "gemini": defs.GEMINI_MODELS,
+            "groq": defs.GROQ_MODELS,
+            "local": {},
+            "minimax": defs.MINIMAX_MODELS,
+            "mistral": defs.MISTRAL_MODELS,
+            "moonshot": defs.MOONSHOT_MODELS,
+            "openai": defs.OPENAI_MODELS,
+            "qwen": defs.QWEN_MODELS,
+            "yi": defs.YI_MODELS,
+            "zhipuai": defs.ZHIPUAI_MODELS,
         }
 
         for model_type, default_models in model_types.items():
             if model_type in data:
                 model_settings = BackendSettings()
-                model_settings.update_models(default_models, data[model_type])
+                model_settings.update_models(default_models, data[model_type].get("models", {}))
                 data[model_type] = model_settings
             else:
                 data[model_type] = BackendSettings(models=default_models)
@@ -65,7 +63,7 @@ class Settings(BaseModel):
         return EndpointSetting()
 
     def get_backend(self, backend: BackendType) -> BackendSettings:
-        return getattr(self, f"{backend.value.lower()}_models")
+        return getattr(self, backend.value.lower())
 
 
 settings = Settings()

vectorvein/types/defaults.py

@@ -14,19 +14,19 @@ MODEL_CONTEXT_LENGTH = 32768
 MOONSHOT_MODELS = {
     "moonshot-v1-8k": {
         "id": "moonshot-v1-8k",
-        "context_length": 8000,
+        "context_length": 8192,
         "function_call_available": True,
         "response_format_available": True,
     },
     "moonshot-v1-32k": {
         "id": "moonshot-v1-32k",
-        "context_length": 32000,
+        "context_length": 32768,
         "function_call_available": True,
         "response_format_available": True,
     },
     "moonshot-v1-128k": {
         "id": "moonshot-v1-128k",
-        "context_length": 128000,
+        "context_length": 131072,
         "function_call_available": True,
         "response_format_available": True,
     },
@@ -38,12 +38,14 @@ DEEPSEEK_MODELS = {
     "deepseek-chat": {
         "id": "deepseek-chat",
         "context_length": 128000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
     "deepseek-coder": {
         "id": "deepseek-chat",
         "context_length": 128000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
@@ -85,48 +87,56 @@ QWEN_MODELS = {
     "qwen1.5-1.8b-chat": {
         "id": "qwen1.5-1.8b-chat",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
     "qwen1.5-4b-chat": {
         "id": "qwen1.5-4b-chat",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
     "qwen1.5-7b-chat": {
         "id": "qwen1.5-7b-chat",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
     "qwen1.5-14b-chat": {
         "id": "qwen1.5-14b-chat",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
     "qwen1.5-32b-chat": {
         "id": "qwen1.5-32b-chat",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
     "qwen1.5-72b-chat": {
         "id": "qwen1.5-72b-chat",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
     "qwen1.5-110b-chat": {
         "id": "qwen1.5-110b-chat",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
     "qwen2-72b-instruct": {
         "id": "qwen2-72b-instruct",
         "context_length": 30000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": True,
     },
@@ -138,42 +148,49 @@ YI_MODELS = {
     "yi-large": {
         "id": "yi-large",
         "context_length": 32000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": False,
     },
     "yi-large-turbo": {
         "id": "yi-large-turbo",
         "context_length": 16000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": False,
     },
     "yi-large-fc": {
         "id": "yi-large-fc",
         "context_length": 32000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": False,
     },
     "yi-medium": {
         "id": "yi-medium",
         "context_length": 16000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": False,
     },
     "yi-medium-200k": {
         "id": "yi-medium-200k",
         "context_length": 200000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": False,
     },
     "yi-spark": {
         "id": "yi-spark",
         "context_length": 16000,
+        "max_output_tokens": 4096,
         "function_call_available": False,
         "response_format_available": False,
     },
     "yi-vision": {
         "id": "yi-vision",
         "context_length": 4000,
+        "max_output_tokens": 2000,
         "function_call_available": False,
         "response_format_available": False,
     },
@@ -187,42 +204,56 @@ ZHIPUAI_MODELS = {
         "context_length": 128000,
         "function_call_available": True,
         "response_format_available": False,
+        "max_output_tokens": 4095,
     },
     "glm-4": {
         "id": "glm-4",
         "context_length": 128000,
         "function_call_available": True,
         "response_format_available": False,
+        "max_output_tokens": 4095,
     },
     "glm-4-0520": {
         "id": "glm-4-0520",
         "context_length": 128000,
         "function_call_available": True,
         "response_format_available": False,
+        "max_output_tokens": 4095,
     },
     "glm-4-air": {
         "id": "glm-4-air",
         "context_length": 128000,
         "function_call_available": True,
         "response_format_available": False,
+        "max_output_tokens": 4095,
     },
     "glm-4-airx": {
         "id": "glm-4-airx",
         "context_length": 128000,
         "function_call_available": True,
         "response_format_available": False,
+        "max_output_tokens": 4095,
     },
     "glm-4-flash": {
         "id": "glm-4-flash",
         "context_length": 128000,
         "function_call_available": True,
         "response_format_available": False,
+        "max_output_tokens": 4095,
+    },
+    "glm-4-long": {
+        "id": "glm-4-long",
+        "context_length": 1000000,
+        "function_call_available": True,
+        "response_format_available": False,
+        "max_output_tokens": 4095,
     },
     "glm-4v": {
         "id": "glm-4v",
         "context_length": 2000,
         "function_call_available": False,
         "response_format_available": False,
+        "max_output_tokens": 1024,
     },
 }
 
@@ -287,34 +318,40 @@ OPENAI_MODELS = {
         "context_length": 16385,
         "function_call_available": True,
         "response_format_available": True,
+        "max_output_tokens": 4096,
     },
     "gpt-4-turbo": {
         "id": "gpt-4-turbo",
         "context_length": 128000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
     "gpt-4": {
         "id": "gpt-4",
         "context_length": 8192,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
     "gpt-4o": {
         "id": "gpt-4o",
         "context_length": 128000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
     "gpt-4o-mini": {
         "id": "gpt-4o-mini",
         "context_length": 128000,
+        "max_output_tokens": 16384,
         "function_call_available": True,
         "response_format_available": True,
     },
     "gpt-4v": {
         "id": "gpt-4v",
         "context_length": 128000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
@@ -326,24 +363,28 @@ ANTHROPIC_MODELS = {
     "claude-3-opus-20240229": {
         "id": "claude-3-opus-20240229",
         "context_length": 200000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
     "claude-3-sonnet-20240229": {
         "id": "claude-3-sonnet-20240229",
         "context_length": 200000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
     "claude-3-haiku-20240307": {
         "id": "claude-3-haiku-20240307",
         "context_length": 200000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
     "claude-3-5-sonnet-20240620": {
         "id": "claude-3-5-sonnet-20240620",
         "context_length": 200000,
+        "max_output_tokens": 4096,
         "function_call_available": True,
         "response_format_available": True,
     },
@@ -355,24 +396,28 @@ MINIMAX_MODELS = {
     "abab5-chat": {
         "id": "abab5-chat",
         "context_length": 6144,
+        "max_output_tokens": 6144,
         "function_call_available": True,
         "response_format_available": True,
     },
     "abab5.5-chat": {
         "id": "abab5.5-chat",
         "context_length": 16384,
+        "max_output_tokens": 16384,
         "function_call_available": True,
         "response_format_available": True,
     },
     "abab6-chat": {
         "id": "abab6-chat",
         "context_length": 32768,
+        "max_output_tokens": 32768,
         "function_call_available": True,
         "response_format_available": True,
     },
     "abab6.5s-chat": {
         "id": "abab6.5s-chat",
         "context_length": 245760,
+        "max_output_tokens": 245760,
         "function_call_available": True,
         "response_format_available": True,
     },

vectorvein/utilities/retry.py

@@ -0,0 +1,36 @@
+# @Author: Bi Ying
+# @Date:   2024-08-14 13:03:10
+import time
+
+
+class Retry:
+    def __init__(self, function):
+        self.function = function
+        self.__retry_times = 3
+        self.__sleep_time = 1
+        self.pargs = []
+        self.kwargs = {}
+
+    def args(self, *args, **kwargs):
+        self.pargs = args
+        self.kwargs = kwargs
+        return self
+
+    def retry_times(self, retry_times: int):
+        self.__retry_times = retry_times
+        return self
+
+    def sleep_time(self, sleep_time):
+        self.__sleep_time = sleep_time
+        return self
+
+    def run(self):
+        try_times = 0
+        while try_times < self.__retry_times:
+            try:
+                return True, self.function(*self.pargs, **self.kwargs)
+            except Exception as e:
+                print(f"{self.function.__name__} 函数出错：{e}")
+                try_times += 1
+                time.sleep(self.__sleep_time)
+        return False, None

{vectorvein-0.1.6.dist-info → vectorvein-0.1.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.6
+Version: 0.1.8
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT
@@ -11,6 +11,8 @@ Requires-Dist: httpx>=0.27.0
 Requires-Dist: anthropic[vertex]>=0.31.2
 Requires-Dist: pydantic>=2.8.2
 Requires-Dist: Pillow>=10.4.0
+Requires-Dist: deepseek-tokenizer>=0.1.0
+Requires-Dist: qwen-tokenizer>=0.1.0
 Description-Content-Type: text/markdown
 
 # vectorvein

{vectorvein-0.1.6.dist-info → vectorvein-0.1.8.dist-info}/RECORD

@@ -1,25 +1,26 @@
-vectorvein-0.1.6.dist-info/METADATA,sha256=cfyPkv1PSdCL5Ve9zAIDTNVzrPDLN2GDbiADGewJN0g,423
-vectorvein-0.1.6.dist-info/WHEEL,sha256=rSwsxJWe3vzyR5HCwjWXQruDgschpei4h_giTm0dJVE,90
+vectorvein-0.1.8.dist-info/METADATA,sha256=3gBxjxgbrdsbA4Xwc3MkuCct-9VFuuM0X9hfSj6sGt4,501
+vectorvein-0.1.8.dist-info/WHEEL,sha256=rSwsxJWe3vzyR5HCwjWXQruDgschpei4h_giTm0dJVE,90
 vectorvein/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 vectorvein/chat_clients/__init__.py,sha256=5j7W--jr-l2cqDJp38uXYvkydDK0rnzm7MYGSACHKmU,3976
-vectorvein/chat_clients/anthropic_client.py,sha256=pkk0yPH05WTOnKOAXkm60ZZw1kdT8pCydNkHx7QArh4,18707
-vectorvein/chat_clients/base_client.py,sha256=GjPUAjgd_36-lHr8QXvz2X_-ApjbmAD3KZRjmntJ65U,4247
+vectorvein/chat_clients/anthropic_client.py,sha256=JjigSUsIn06ixIEjnOJhVbcMqy2_MAL3iVUlDFAFMW4,20008
+vectorvein/chat_clients/base_client.py,sha256=wMXpQ1L1KDb2Hg6va3H3GmcVeQB6r6sh7F4IS0DBQWI,4275
 vectorvein/chat_clients/deepseek_client.py,sha256=3qWu01NlJAP2N-Ff62d5-CZXZitlizE1fzb20LNetig,526
-vectorvein/chat_clients/gemini_client.py,sha256=IpkfcqVF38f4kAFnlHyisn4vkiMeM4cICyS4XDD0jJE,12787
+vectorvein/chat_clients/gemini_client.py,sha256=IHcBHTSHkj3f962S5L7Ga-XA-96sq8quIDRZpoqvGss,13653
 vectorvein/chat_clients/groq_client.py,sha256=Uow4pgdmFi93ZQSoOol2-0PhhqkW-S0XuSldvppz5U4,498
 vectorvein/chat_clients/local_client.py,sha256=55nOsxzqUf79q3Y14MKROA71zxhsT7p7FsDZ89rts2M,422
-vectorvein/chat_clients/minimax_client.py,sha256=toe4mFYHphHnPQpOpegaL5n2OxTUu5TJVju61j7hgBw,12100
+vectorvein/chat_clients/minimax_client.py,sha256=iC60QnLQRb5Y0B6_vorL9wkZx1Gs9w9znm8zil4uJ-g,13588
 vectorvein/chat_clients/mistral_client.py,sha256=1aKSylzBDaLYcFnaBIL4-sXSzWmXfBeON9Q0rq-ziWw,534
 vectorvein/chat_clients/moonshot_client.py,sha256=gbu-6nGxx8uM_U2WlI4Wus881rFRotzHtMSoYOcruGU,526
 vectorvein/chat_clients/openai_client.py,sha256=Nz6tV45pWcsOupxjnsRsGTicbQNJWIZyxuJoJ5DGMpg,527
-vectorvein/chat_clients/openai_compatible_client.py,sha256=sN9fq8yZ0XKSVg0eQZDWUE-awrkkJA2lEdMG-WENnUg,11951
+vectorvein/chat_clients/openai_compatible_client.py,sha256=fvg--wFwnFEEhLGS9_u1XzNhtkkDUf4_rq6zYKwnOuI,13738
 vectorvein/chat_clients/qwen_client.py,sha256=-ryh-m9PgsO0fc4ulcCmPTy1155J8YUy15uPoJQOHA0,513
-vectorvein/chat_clients/utils.py,sha256=tAQwfydj46sMxSHeaeOWXrTUY2q0h7482NbvZjbNz9A,17637
+vectorvein/chat_clients/utils.py,sha256=mnAew2Ie3nQHdEyDLKuJvXkQ5QdcSAJ6SpYk5JPbR1Q,20888
 vectorvein/chat_clients/yi_client.py,sha256=RNf4CRuPJfixrwLZ3-DEc3t25QDe1mvZeb9sku2f8Bc,484
 vectorvein/chat_clients/zhipuai_client.py,sha256=Ys5DSeLCuedaDXr3PfG1EW2zKXopt-awO2IylWSwY0s,519
-vectorvein/settings/__init__.py,sha256=1__ILDpBh25S43d32knqLBe05tF4AQwT0ZALdBHw1OY,3396
-vectorvein/types/defaults.py,sha256=Mg-Mj3_eBzKZn1N8x1V2GqyaYgLD13i-NdSYdQC28X4,11437
+vectorvein/settings/__init__.py,sha256=4mpccT7eZC3yI1vVnVViW4wHBnDEH9D2R5EsIP34VgU,3218
+vectorvein/types/defaults.py,sha256=ANIYL0W0bxl2IBxvtkS_WlS_qMQQwpi5TKRdLxdk47M,13027
 vectorvein/types/enums.py,sha256=vzOenCnRlFXBwPh-lfFhjGfM-6yfDj7wZColHODqocI,1550
 vectorvein/types/llm_parameters.py,sha256=nBjStC2zndTY__yhD2WFXB09taxEhDLE3OHA6MICfgE,3494
 vectorvein/utilities/media_processing.py,sha256=BujciRmw1GMmc3ELRvafL8STcy6r5b2rVnh27-uA7so,2256
-vectorvein-0.1.6.dist-info/RECORD,,
+vectorvein/utilities/retry.py,sha256=9ePuJdeUUGx-qMWfaFxmlOvG_lQPwCQ4UB1z3Edlo34,993
+vectorvein-0.1.8.dist-info/RECORD,,