vectorvein 0.1.28__tar.gz → 0.1.30__tar.gz

{vectorvein-0.1.28 → vectorvein-0.1.30}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.28
+Version: 0.1.30
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.1.28 → vectorvein-0.1.30}/pyproject.toml

@@ -16,7 +16,7 @@ description = "Default template for PDM package"
 name = "vectorvein"
 readme = "README.md"
 requires-python = ">=3.10"
-version = "0.1.28"
+version = "0.1.30"
 
 [project.license]
 text = "MIT"

vectorvein-0.1.30/src/vectorvein/chat_clients/__init__.py

@@ -0,0 +1,522 @@
+# @Author: Bi Ying
+# @Date:   2024-07-26 14:48:55
+import httpx
+from typing import overload, Literal
+
+from .base_client import BaseChatClient, BaseAsyncChatClient
+
+from .yi_client import YiChatClient, AsyncYiChatClient
+from .groq_client import GroqChatClient, AsyncGroqChatClient
+from .qwen_client import QwenChatClient, AsyncQwenChatClient
+from .local_client import LocalChatClient, AsyncLocalChatClient
+from .gemini_client import GeminiChatClient, AsyncGeminiChatClient
+from .openai_client import OpenAIChatClient, AsyncOpenAIChatClient
+from .zhipuai_client import ZhiPuAIChatClient, AsyncZhiPuAIChatClient
+from .minimax_client import MiniMaxChatClient, AsyncMiniMaxChatClient
+from .mistral_client import MistralChatClient, AsyncMistralChatClient
+from .stepfun_client import StepFunChatClient, AsyncStepFunChatClient
+from .baichuan_client import BaichuanChatClient, AsyncBaichuanChatClient
+from .moonshot_client import MoonshotChatClient, AsyncMoonshotChatClient
+from .deepseek_client import DeepSeekChatClient, AsyncDeepSeekChatClient
+
+from ..types import defaults as defs
+from ..types.enums import BackendType, ContextLengthControlType
+from .anthropic_client import AnthropicChatClient, AsyncAnthropicChatClient
+from .utils import format_messages, get_token_counts, get_message_token_counts, ToolCallContentProcessor
+
+# 后端映射
+BackendMap = {
+    "sync": {
+        BackendType.Anthropic: AnthropicChatClient,
+        BackendType.DeepSeek: DeepSeekChatClient,
+        BackendType.Gemini: GeminiChatClient,
+        BackendType.Groq: GroqChatClient,
+        BackendType.Local: LocalChatClient,
+        BackendType.MiniMax: MiniMaxChatClient,
+        BackendType.Mistral: MistralChatClient,
+        BackendType.Moonshot: MoonshotChatClient,
+        BackendType.OpenAI: OpenAIChatClient,
+        BackendType.Qwen: QwenChatClient,
+        BackendType.Yi: YiChatClient,
+        BackendType.ZhiPuAI: ZhiPuAIChatClient,
+        BackendType.Baichuan: BaichuanChatClient,
+        BackendType.StepFun: StepFunChatClient,
+    },
+    "async": {
+        BackendType.Anthropic: AsyncAnthropicChatClient,
+        BackendType.DeepSeek: AsyncDeepSeekChatClient,
+        BackendType.Gemini: AsyncGeminiChatClient,
+        BackendType.Groq: AsyncGroqChatClient,
+        BackendType.Local: AsyncLocalChatClient,
+        BackendType.MiniMax: AsyncMiniMaxChatClient,
+        BackendType.Mistral: AsyncMistralChatClient,
+        BackendType.Moonshot: AsyncMoonshotChatClient,
+        BackendType.OpenAI: AsyncOpenAIChatClient,
+        BackendType.Qwen: AsyncQwenChatClient,
+        BackendType.Yi: AsyncYiChatClient,
+        BackendType.ZhiPuAI: AsyncZhiPuAIChatClient,
+        BackendType.Baichuan: AsyncBaichuanChatClient,
+        BackendType.StepFun: AsyncStepFunChatClient,
+    },
+}
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Anthropic],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> AnthropicChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.DeepSeek],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> DeepSeekChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Gemini],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> GeminiChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Groq],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> GroqChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Local],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> LocalChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.MiniMax],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> MiniMaxChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Mistral],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> MistralChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Moonshot],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> MoonshotChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.OpenAI],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> OpenAIChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Qwen],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> QwenChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Yi],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> YiChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.ZhiPuAI],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> ZhiPuAIChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.Baichuan],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> BaichuanChatClient: ...
+
+
+@overload
+def create_chat_client(
+    backend: Literal[BackendType.StepFun],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> StepFunChatClient: ...
+
+
+def create_chat_client(
+    backend: BackendType,
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.Client | None = None,
+    **kwargs,
+) -> BaseChatClient:
+    if backend not in BackendMap["sync"]:
+        raise ValueError(f"Unsupported backend: {backend}")
+
+    ClientClass = BackendMap["sync"][backend]
+    if model is None:
+        model = ClientClass.DEFAULT_MODEL
+    return ClientClass(
+        model=model,
+        stream=stream,
+        temperature=temperature,
+        context_length_control=context_length_control,
+        random_endpoint=random_endpoint,
+        endpoint_id=endpoint_id,
+        http_client=http_client,
+        **kwargs,
+    )
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Anthropic],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncAnthropicChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.DeepSeek],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncDeepSeekChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Gemini],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncGeminiChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Groq],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncGroqChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Local],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncLocalChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.MiniMax],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncMiniMaxChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Mistral],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncMistralChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Moonshot],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncMoonshotChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.OpenAI],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncOpenAIChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Qwen],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncQwenChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Yi],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncYiChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.ZhiPuAI],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncZhiPuAIChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.Baichuan],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncBaichuanChatClient: ...
+
+
+@overload
+def create_async_chat_client(
+    backend: Literal[BackendType.StepFun],
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> AsyncStepFunChatClient: ...
+
+
+def create_async_chat_client(
+    backend: BackendType,
+    model: str | None = None,
+    stream: bool = False,
+    temperature: float = 0.7,
+    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
+    random_endpoint: bool = True,
+    endpoint_id: str = "",
+    http_client: httpx.AsyncClient | None = None,
+    **kwargs,
+) -> BaseAsyncChatClient:
+    if backend not in BackendMap["async"]:
+        raise ValueError(f"Unsupported backend: {backend}")
+
+    ClientClass = BackendMap["async"][backend]
+    if model is None:
+        model = ClientClass.DEFAULT_MODEL
+    return ClientClass(
+        model=model,
+        stream=stream,
+        temperature=temperature,
+        context_length_control=context_length_control,
+        random_endpoint=random_endpoint,
+        endpoint_id=endpoint_id,
+        http_client=http_client,
+        **kwargs,
+    )
+
+
+__all__ = [
+    "BackendType",
+    "format_messages",
+    "get_token_counts",
+    "create_chat_client",
+    "create_async_chat_client",
+    "get_message_token_counts",
+    "ToolCallContentProcessor",
+]

{vectorvein-0.1.28 → vectorvein-0.1.30}/src/vectorvein/chat_clients/utils.py

@@ -21,8 +21,22 @@ from ..types.llm_parameters import (
 )
 
 
-chatgpt_encoding = tiktoken.encoding_for_model("gpt-3.5-turbo")
-gpt_4o_encoding = tiktoken.encoding_for_model("gpt-4o")
+gpt_35_encoding = None
+gpt_4o_encoding = None
+
+
+def get_gpt_35_encoding():
+    global gpt_35_encoding
+    if gpt_35_encoding is None:
+        gpt_35_encoding = tiktoken.encoding_for_model("gpt-3.5-turbo")
+    return gpt_35_encoding
+
+
+def get_gpt_4o_encoding():
+    global gpt_4o_encoding
+    if gpt_4o_encoding is None:
+        gpt_4o_encoding = tiktoken.encoding_for_model("gpt-4o")
+    return gpt_4o_encoding
 
 
 class ToolCallContentProcessor:
@@ -106,9 +120,9 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
     if not isinstance(text, str):
         text = str(text)
     if model == "gpt-3.5-turbo":
-        return len(chatgpt_encoding.encode(text))
+        return len(get_gpt_35_encoding().encode(text))
     elif model in ("gpt-4o", "gpt-4o-mini"):
-        return len(gpt_4o_encoding.encode(text))
+        return len(get_gpt_4o_encoding().encode(text))
     elif model.startswith("abab"):
         model_setting = settings.minimax.models[model]
         if len(model_setting.endpoints) == 0:
@@ -140,7 +154,7 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
     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))
+            return len(get_gpt_35_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"
@@ -165,7 +179,7 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
     elif model.startswith("gemini"):
         model_setting = settings.gemini.models[model]
         if len(model_setting.endpoints) == 0:
-            return len(chatgpt_encoding.encode(text))
+            return len(get_gpt_35_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"
@@ -198,7 +212,7 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
     elif model.startswith("stepfun"):
         model_setting = settings.moonshot.models[model]
         if len(model_setting.endpoints) == 0:
-            return len(chatgpt_encoding.encode(text))
+            return len(get_gpt_35_encoding().encode(text))
         endpoint_id = model_setting.endpoints[0]
         endpoint = settings.get_endpoint(endpoint_id)
         tokenize_url = "https://api.stepfun.com/v1/token/count"
@@ -221,7 +235,7 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
         result = response.json()
         return result["data"]["total_tokens"]
     else:
-        return len(chatgpt_encoding.encode(text))
+        return len(get_gpt_35_encoding().encode(text))
 
 
 def calculate_image_tokens(width: int, height: int, model: str = "gpt-4o"):

vectorvein-0.1.28/src/vectorvein/chat_clients/__init__.py

@@ -1,129 +0,0 @@
-# @Author: Bi Ying
-# @Date:   2024-07-26 14:48:55
-import httpx
-
-from .base_client import BaseChatClient, BaseAsyncChatClient
-
-from .yi_client import YiChatClient, AsyncYiChatClient
-from .groq_client import GroqChatClient, AsyncGroqChatClient
-from .qwen_client import QwenChatClient, AsyncQwenChatClient
-from .local_client import LocalChatClient, AsyncLocalChatClient
-from .gemini_client import GeminiChatClient, AsyncGeminiChatClient
-from .openai_client import OpenAIChatClient, AsyncOpenAIChatClient
-from .zhipuai_client import ZhiPuAIChatClient, AsyncZhiPuAIChatClient
-from .minimax_client import MiniMaxChatClient, AsyncMiniMaxChatClient
-from .mistral_client import MistralChatClient, AsyncMistralChatClient
-from .stepfun_client import StepFunChatClient, AsyncStepFunChatClient
-from .baichuan_client import BaichuanChatClient, AsyncBaichuanChatClient
-from .moonshot_client import MoonshotChatClient, AsyncMoonshotChatClient
-from .deepseek_client import DeepSeekChatClient, AsyncDeepSeekChatClient
-
-from ..types import defaults as defs
-from ..types.enums import BackendType, ContextLengthControlType
-from .anthropic_client import AnthropicChatClient, AsyncAnthropicChatClient
-from .utils import format_messages, get_token_counts, get_message_token_counts, ToolCallContentProcessor
-
-
-BackendMap = {
-    "sync": {
-        BackendType.Anthropic: AnthropicChatClient,
-        BackendType.DeepSeek: DeepSeekChatClient,
-        BackendType.Gemini: GeminiChatClient,
-        BackendType.Groq: GroqChatClient,
-        BackendType.Local: LocalChatClient,
-        BackendType.MiniMax: MiniMaxChatClient,
-        BackendType.Mistral: MistralChatClient,
-        BackendType.Moonshot: MoonshotChatClient,
-        BackendType.OpenAI: OpenAIChatClient,
-        BackendType.Qwen: QwenChatClient,
-        BackendType.Yi: YiChatClient,
-        BackendType.ZhiPuAI: ZhiPuAIChatClient,
-        BackendType.Baichuan: BaichuanChatClient,
-        BackendType.StepFun: StepFunChatClient,
-    },
-    "async": {
-        BackendType.Anthropic: AsyncAnthropicChatClient,
-        BackendType.DeepSeek: AsyncDeepSeekChatClient,
-        BackendType.Gemini: AsyncGeminiChatClient,
-        BackendType.Groq: AsyncGroqChatClient,
-        BackendType.Local: AsyncLocalChatClient,
-        BackendType.MiniMax: AsyncMiniMaxChatClient,
-        BackendType.Mistral: AsyncMistralChatClient,
-        BackendType.Moonshot: AsyncMoonshotChatClient,
-        BackendType.OpenAI: AsyncOpenAIChatClient,
-        BackendType.Qwen: AsyncQwenChatClient,
-        BackendType.Yi: AsyncYiChatClient,
-        BackendType.ZhiPuAI: AsyncZhiPuAIChatClient,
-        BackendType.Baichuan: AsyncBaichuanChatClient,
-        BackendType.StepFun: AsyncStepFunChatClient,
-    },
-}
-
-
-def create_chat_client(
-    backend: BackendType,
-    model: str | None = None,
-    stream: bool = False,
-    temperature: float = 0.7,
-    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
-    random_endpoint: bool = True,
-    endpoint_id: str = "",
-    http_client: httpx.Client | None = None,
-    **kwargs,
-) -> BaseChatClient:
-    if backend not in BackendMap["sync"]:
-        raise ValueError(f"Unsupported backend: {backend}")
-
-    ClientClass = BackendMap["sync"][backend]
-    if model is None:
-        model = ClientClass.DEFAULT_MODEL
-    return BackendMap["sync"][backend](
-        model=model,
-        stream=stream,
-        temperature=temperature,
-        context_length_control=context_length_control,
-        random_endpoint=random_endpoint,
-        endpoint_id=endpoint_id,
-        http_client=http_client,
-        **kwargs,
-    )
-
-
-def create_async_chat_client(
-    backend: BackendType,
-    model: str | None = None,
-    stream: bool = False,
-    temperature: float = 0.7,
-    context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
-    random_endpoint: bool = True,
-    endpoint_id: str = "",
-    http_client: httpx.AsyncClient | None = None,
-    **kwargs,
-) -> BaseAsyncChatClient:
-    if backend not in BackendMap["async"]:
-        raise ValueError(f"Unsupported backend: {backend}")
-
-    ClientClass = BackendMap["async"][backend]
-    if model is None:
-        model = ClientClass.DEFAULT_MODEL
-    return BackendMap["async"][backend](
-        model=model,
-        stream=stream,
-        temperature=temperature,
-        context_length_control=context_length_control,
-        random_endpoint=random_endpoint,
-        endpoint_id=endpoint_id,
-        http_client=http_client,
-        **kwargs,
-    )
-
-
-__all__ = [
-    "BackendType",
-    "format_messages",
-    "get_token_counts",
-    "create_chat_client",
-    "create_async_chat_client",
-    "get_message_token_counts",
-    "ToolCallContentProcessor",
-]