PyPI - vectorvein - Versions diffs - 0.2.6__py3-none-any.whl → 0.2.8__py3-none-any.whl - Mend

vectorvein 0.2.6py3-none-any.whl → 0.2.8py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (10) hide show

vectorvein/chat_clients/__init__.py +582 -582
vectorvein/chat_clients/anthropic_client.py +331 -28
vectorvein/chat_clients/base_client.py +302 -24
vectorvein/chat_clients/openai_compatible_client.py +312 -38
vectorvein/types/defaults.py +8 -0
vectorvein/types/llm_parameters.py +4 -1
{vectorvein-0.2.6.dist-info → vectorvein-0.2.8.dist-info}/METADATA +2 -2
{vectorvein-0.2.6.dist-info → vectorvein-0.2.8.dist-info}/RECORD +10 -10
{vectorvein-0.2.6.dist-info → vectorvein-0.2.8.dist-info}/WHEEL +0 -0
{vectorvein-0.2.6.dist-info → vectorvein-0.2.8.dist-info}/entry_points.txt +0 -0

vectorvein/chat_clients/anthropic_client.py CHANGED Viewed

@@ -1,11 +1,30 @@
 # @Author: Bi Ying
 # @Date:   2024-07-26 14:48:55
 import json
-from typing import overload, Generator, AsyncGenerator, Any, Literal, Iterable
+from typing import (
+    Any,
+    Dict,
+    List,
+    Union,
+    Literal,
+    Iterable,
+    overload,
+    Optional,
+    Generator,
+    AsyncGenerator,
+)
 import httpx
 from openai._types import NotGiven as OpenAINotGiven
 from openai._types import NOT_GIVEN as OPENAI_NOT_GIVEN
+from openai._types import Headers, Query, Body
+from openai.types.shared_params.metadata import Metadata
+from openai.types.chat.completion_create_params import ResponseFormat
+from openai.types.chat.chat_completion_modality import ChatCompletionModality
+from openai.types.chat.chat_completion_audio_param import ChatCompletionAudioParam
+from openai.types.chat.chat_completion_reasoning_effort import ChatCompletionReasoningEffort
+from openai.types.chat.chat_completion_stream_options_param import ChatCompletionStreamOptionsParam
+from openai.types.chat.chat_completion_prediction_content_param import ChatCompletionPredictionContentParam
 from anthropic import (
     Anthropic,
     AnthropicVertex,
@@ -17,8 +36,11 @@ from anthropic import (
 from anthropic._types import NOT_GIVEN
 from anthropic.types import (
     TextBlock,
+    ThinkingBlock,
+    RedactedThinkingBlock,
     MessageParam,
     ToolUseBlock,
+    ThinkingConfigParam,
     RawMessageDeltaEvent,
     RawMessageStartEvent,
     RawContentBlockStartEvent,
@@ -43,7 +65,6 @@ from ..types.llm_parameters import (
     ChatCompletionMessage,
     ChatCompletionToolParam,
     ChatCompletionDeltaMessage,
-    ChatCompletionStreamOptionsParam,
 )
@@ -128,6 +149,8 @@ def refactor_into_openai_messages(messages: Iterable[MessageParam]):
             for item in content:
                 if isinstance(item, (TextBlock, ToolUseBlock)):
                     _content.append(item.model_dump())
+                elif isinstance(item, (ThinkingBlock, RedactedThinkingBlock)):
+                    continue
                 elif item.get("type") == "image":
                     image_data = item.get("source", {}).get("data", "")
                     media_type = item.get("source", {}).get("media_type", "")
@@ -257,11 +280,33 @@ class AnthropicChatClient(BaseChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ) -> ChatCompletionMessage:
         pass
@@ -276,11 +321,33 @@ class AnthropicChatClient(BaseChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ) -> Generator[ChatCompletionDeltaMessage, None, None]:
         pass
@@ -295,11 +362,33 @@ class AnthropicChatClient(BaseChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ) -> ChatCompletionMessage | Generator[ChatCompletionDeltaMessage, Any, None]:
         pass
@@ -313,11 +402,33 @@ class AnthropicChatClient(BaseChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ):
         if model is not None:
             self.model = model
@@ -362,7 +473,28 @@ class AnthropicChatClient(BaseChatClient):
                         stream_options=stream_options,
                         top_p=top_p,
                         skip_cutoff=skip_cutoff,
-                        **kwargs,
+                        audio=audio,
+                        frequency_penalty=frequency_penalty,
+                        logit_bias=logit_bias,
+                        logprobs=logprobs,
+                        max_completion_tokens=max_completion_tokens,
+                        metadata=metadata,
+                        modalities=modalities,
+                        n=n,
+                        parallel_tool_calls=parallel_tool_calls,
+                        prediction=prediction,
+                        presence_penalty=presence_penalty,
+                        reasoning_effort=reasoning_effort,
+                        seed=seed,
+                        service_tier=service_tier,
+                        stop=stop,
+                        store=store,
+                        top_logprobs=top_logprobs,
+                        user=user,
+                        extra_headers=extra_headers,
+                        extra_query=extra_query,
+                        extra_body=extra_body,
+                        timeout=timeout,
                     )
                     for chunk in response:
                         yield chunk
@@ -389,7 +521,28 @@ class AnthropicChatClient(BaseChatClient):
                     response_format=response_format,
                     top_p=top_p,
                     skip_cutoff=skip_cutoff,
-                    **kwargs,
+                    audio=audio,
+                    frequency_penalty=frequency_penalty,
+                    logit_bias=logit_bias,
+                    logprobs=logprobs,
+                    max_completion_tokens=max_completion_tokens,
+                    metadata=metadata,
+                    modalities=modalities,
+                    n=n,
+                    parallel_tool_calls=parallel_tool_calls,
+                    prediction=prediction,
+                    presence_penalty=presence_penalty,
+                    reasoning_effort=reasoning_effort,
+                    seed=seed,
+                    service_tier=service_tier,
+                    stop=stop,
+                    store=store,
+                    top_logprobs=top_logprobs,
+                    user=user,
+                    extra_headers=extra_headers,
+                    extra_query=extra_query,
+                    extra_body=extra_body,
+                    timeout=timeout,
                 )
         raw_client = self.raw_client  # 调用完 self.raw_client 后，self.model_id 会被赋值
@@ -403,6 +556,8 @@ class AnthropicChatClient(BaseChatClient):
             top_p = NOT_GIVEN
         if isinstance(self.temperature, NotGiven) or self.temperature is None:
             self.temperature = NOT_GIVEN
+        if isinstance(thinking, NotGiven) or thinking is None:
+            thinking = NOT_GIVEN
         if messages[0].get("role") == "system":
             system_prompt: str = messages[0]["content"]
@@ -447,11 +602,11 @@ class AnthropicChatClient(BaseChatClient):
                 tools=tools_params,
                 tool_choice=tool_choice_param,
                 top_p=top_p,
-                **kwargs,
+                thinking=thinking,
             )
             def generator():
-                result = {"content": "", "usage": {}, "tool_calls": []}
+                result = {"content": "", "reasoning_content": "", "usage": {}, "tool_calls": []}
                 for chunk in stream_response:
                     message = {"content": "", "tool_calls": []}
                     if isinstance(chunk, RawMessageStartEvent):
@@ -472,11 +627,16 @@ class AnthropicChatClient(BaseChatClient):
                             ]
                         elif chunk.content_block.type == "text":
                             message["content"] = chunk.content_block.text
+                        elif chunk.content_block.type == "thinking":
+                            message["reasoning_content"] = chunk.content_block.thinking
                         yield ChatCompletionDeltaMessage(**message)
                     elif isinstance(chunk, RawContentBlockDeltaEvent):
                         if chunk.delta.type == "text_delta":
                             message["content"] = chunk.delta.text
                             result["content"] += chunk.delta.text
+                        elif chunk.delta.type == "thinking_delta":
+                            message["reasoning_content"] = chunk.delta.thinking
+                            result["reasoning_content"] += chunk.delta.thinking
                         elif chunk.delta.type == "input_json_delta":
                             result["tool_calls"][0]["function"]["arguments"] += chunk.delta.partial_json
                             message["tool_calls"] = [
@@ -516,11 +676,12 @@ class AnthropicChatClient(BaseChatClient):
                 tools=tools_params,
                 tool_choice=tool_choice_param,
                 top_p=top_p,
-                **kwargs,
+                thinking=thinking,
             )
             result = {
                 "content": "",
+                "reasoning_content": "",
                 "usage": {
                     "prompt_tokens": response.usage.input_tokens,
                     "completion_tokens": response.usage.output_tokens,
@@ -531,6 +692,8 @@ class AnthropicChatClient(BaseChatClient):
             for content_block in response.content:
                 if isinstance(content_block, TextBlock):
                     result["content"] += content_block.text
+                elif isinstance(content_block, ThinkingBlock):
+                    result["reasoning_content"] = content_block.thinking
                 elif isinstance(content_block, ToolUseBlock):
                     tool_calls.append(content_block.model_dump())
@@ -647,11 +810,33 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ) -> ChatCompletionMessage:
         pass
@@ -666,11 +851,33 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ) -> AsyncGenerator[ChatCompletionDeltaMessage, Any]:
         pass
@@ -685,11 +892,33 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ) -> ChatCompletionMessage | AsyncGenerator[ChatCompletionDeltaMessage, Any]:
         pass
@@ -703,11 +932,33 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         max_tokens: int | None = None,
         tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
         tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
-        response_format: dict | None = None,
+        response_format: ResponseFormat | NotGiven = NOT_GIVEN,
         stream_options: ChatCompletionStreamOptionsParam | None = None,
         top_p: float | NotGiven | None = NOT_GIVEN,
         skip_cutoff: bool = False,
-        **kwargs,
+        audio: Optional[ChatCompletionAudioParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        frequency_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logit_bias: Optional[Dict[str, int]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        logprobs: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        max_completion_tokens: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        metadata: Optional[Metadata] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        modalities: Optional[List[ChatCompletionModality]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        n: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        parallel_tool_calls: bool | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        prediction: Optional[ChatCompletionPredictionContentParam] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        presence_penalty: Optional[float] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        reasoning_effort: Optional[ChatCompletionReasoningEffort] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        thinking: ThinkingConfigParam | None | NotGiven = NOT_GIVEN,
+        seed: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        service_tier: Optional[Literal["auto", "default"]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        stop: Union[Optional[str], List[str]] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        store: Optional[bool] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        top_logprobs: Optional[int] | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        user: str | OpenAINotGiven = OPENAI_NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | OpenAINotGiven = OPENAI_NOT_GIVEN,
     ):
         if model is not None:
             self.model = model
@@ -753,7 +1004,28 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                         stream_options=stream_options,
                         top_p=top_p,
                         skip_cutoff=skip_cutoff,
-                        **kwargs,
+                        audio=audio,
+                        frequency_penalty=frequency_penalty,
+                        logit_bias=logit_bias,
+                        logprobs=logprobs,
+                        max_completion_tokens=max_completion_tokens,
+                        metadata=metadata,
+                        modalities=modalities,
+                        n=n,
+                        parallel_tool_calls=parallel_tool_calls,
+                        prediction=prediction,
+                        presence_penalty=presence_penalty,
+                        reasoning_effort=reasoning_effort,
+                        seed=seed,
+                        service_tier=service_tier,
+                        stop=stop,
+                        store=store,
+                        top_logprobs=top_logprobs,
+                        user=user,
+                        extra_headers=extra_headers,
+                        extra_query=extra_query,
+                        extra_body=extra_body,
+                        timeout=timeout,
                     )
                     async for chunk in response:
                         yield chunk
@@ -781,7 +1053,28 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                     response_format=response_format,
                     top_p=top_p,
                     skip_cutoff=skip_cutoff,
-                    **kwargs,
+                    audio=audio,
+                    frequency_penalty=frequency_penalty,
+                    logit_bias=logit_bias,
+                    logprobs=logprobs,
+                    max_completion_tokens=max_completion_tokens,
+                    metadata=metadata,
+                    modalities=modalities,
+                    n=n,
+                    parallel_tool_calls=parallel_tool_calls,
+                    prediction=prediction,
+                    presence_penalty=presence_penalty,
+                    reasoning_effort=reasoning_effort,
+                    seed=seed,
+                    service_tier=service_tier,
+                    stop=stop,
+                    store=store,
+                    top_logprobs=top_logprobs,
+                    user=user,
+                    extra_headers=extra_headers,
+                    extra_query=extra_query,
+                    extra_body=extra_body,
+                    timeout=timeout,
                 )
         raw_client = self.raw_client  # 调用完 self.raw_client 后，self.model_id 会被赋值
@@ -795,6 +1088,8 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
             top_p = NOT_GIVEN
         if isinstance(self.temperature, NotGiven) or self.temperature is None:
             self.temperature = NOT_GIVEN
+        if isinstance(thinking, NotGiven) or thinking is None:
+            thinking = NOT_GIVEN
         if messages[0].get("role") == "system":
             system_prompt = messages[0]["content"]
@@ -839,11 +1134,11 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                 tools=tools_params,
                 tool_choice=tool_choice_param,
                 top_p=top_p,
-                **kwargs,
+                thinking=thinking,
             )
             async def generator():
-                result = {"content": "", "usage": {}, "tool_calls": []}
+                result = {"content": "", "reasoning_content": "", "usage": {}, "tool_calls": []}
                 async for chunk in stream_response:
                     message = {"content": "", "tool_calls": []}
                     if isinstance(chunk, RawMessageStartEvent):
@@ -864,11 +1159,16 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                             ]
                         elif chunk.content_block.type == "text":
                             message["content"] = chunk.content_block.text
+                        elif chunk.content_block.type == "thinking":
+                            message["reasoning_content"] = chunk.content_block.thinking
                         yield ChatCompletionDeltaMessage(**message)
                     elif isinstance(chunk, RawContentBlockDeltaEvent):
                         if chunk.delta.type == "text_delta":
                             message["content"] = chunk.delta.text
                             result["content"] += chunk.delta.text
+                        elif chunk.delta.type == "thinking_delta":
+                            message["reasoning_content"] = chunk.delta.thinking
+                            result["reasoning_content"] += chunk.delta.thinking
                         elif chunk.delta.type == "input_json_delta":
                             result["tool_calls"][0]["function"]["arguments"] += chunk.delta.partial_json
                             message["tool_calls"] = [
@@ -908,11 +1208,12 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                 tools=tools_params,
                 tool_choice=tool_choice_param,
                 top_p=top_p,
-                **kwargs,
+                thinking=thinking,
             )
             result = {
                 "content": "",
+                "reasoning_content": "",
                 "usage": {
                     "prompt_tokens": response.usage.input_tokens,
                     "completion_tokens": response.usage.output_tokens,
@@ -923,6 +1224,8 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
             for content_block in response.content:
                 if isinstance(content_block, TextBlock):
                     result["content"] += content_block.text
+                elif isinstance(content_block, ThinkingBlock):
+                    result["reasoning_content"] = content_block.thinking
                 elif isinstance(content_block, ToolUseBlock):
                     tool_calls.append(content_block.model_dump())

vectorvein 0.2.6__py3-none-any.whl → 0.2.8__py3-none-any.whl

vectorvein 0.2.6py3-none-any.whl → 0.2.8py3-none-any.whl