vectorvein 0.1.34__py3-none-any.whl → 0.1.36__py3-none-any.whl

vectorvein/chat_clients/anthropic_client.py

@@ -25,6 +25,7 @@ from ..settings import settings
 from ..types import defaults as defs
 from .utils import cutoff_messages, get_message_token_counts
 from .base_client import BaseChatClient, BaseAsyncChatClient
+from .openai_compatible_client import OpenAICompatibleChatClient, AsyncOpenAICompatibleChatClient
 from ..types.enums import ContextLengthControlType, BackendType
 from ..types.llm_parameters import (
     Usage,
@@ -122,7 +123,7 @@ class AnthropicChatClient(BaseChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.Client | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -132,7 +133,7 @@ class AnthropicChatClient(BaseChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
 
     @cached_property
@@ -176,12 +177,23 @@ class AnthropicChatClient(BaseChatClient):
                 access_token=self.creds.token,
                 http_client=self.http_client,
             )
-        else:
+        elif self.endpoint.api_schema_type == "default":
             return Anthropic(
                 api_key=self.endpoint.api_key,
                 base_url=self.endpoint.api_base,
                 http_client=self.http_client,
             )
+        elif self.endpoint.api_schema_type == "openai":
+            return OpenAICompatibleChatClient(
+                model=self.model,
+                stream=self.stream,
+                temperature=self.temperature,
+                context_length_control=self.context_length_control,
+                random_endpoint=self.random_endpoint,
+                endpoint_id=self.endpoint_id,
+                http_client=self.http_client,
+                backend_name=self.BACKEND_NAME,
+            ).raw_client
 
     @overload
     def create_completion(
@@ -213,6 +225,21 @@ class AnthropicChatClient(BaseChatClient):
     ) -> Generator[ChatCompletionDeltaMessage, None, None]:
         pass
 
+    @overload
+    def create_completion(
+        self,
+        messages: list,
+        model: str | None = None,
+        stream: bool | None = None,
+        temperature: float | None = None,
+        max_tokens: int | None = None,
+        tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
+        tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
+        response_format: dict | None = None,
+        **kwargs,
+    ) -> ChatCompletionMessage | Generator[ChatCompletionDeltaMessage, Any, None]:
+        pass
+
     def create_completion(
         self,
         messages: list,
@@ -236,6 +263,41 @@ class AnthropicChatClient(BaseChatClient):
         if isinstance(tool_choice, OpenAINotGiven):
             tool_choice = NOT_GIVEN
 
+        if self.random_endpoint:
+            self.random_endpoint = True
+            self.endpoint_id = random.choice(self.backend_settings.models[self.model].endpoints)
+            self.endpoint = settings.get_endpoint(self.endpoint_id)
+
+        if self.endpoint.api_schema_type == "openai":
+            if self.stream:
+                return OpenAICompatibleChatClient(
+                    model=self.model,
+                    stream=True,
+                    temperature=self.temperature,
+                    context_length_control=self.context_length_control,
+                    random_endpoint=self.random_endpoint,
+                    endpoint_id=self.endpoint_id,
+                    http_client=self.http_client,
+                    backend_name=self.BACKEND_NAME,
+                ).create_completion(
+                    messages, model, True, temperature, max_tokens, tools, tool_choice, response_format, **kwargs
+                )
+            else:
+                return OpenAICompatibleChatClient(
+                    model=self.model,
+                    stream=False,
+                    temperature=self.temperature,
+                    context_length_control=self.context_length_control,
+                    random_endpoint=self.random_endpoint,
+                    endpoint_id=self.endpoint_id,
+                    http_client=self.http_client,
+                    backend_name=self.BACKEND_NAME,
+                ).create_completion(
+                    messages, model, False, temperature, max_tokens, tools, tool_choice, response_format, **kwargs
+                )
+
+        assert isinstance(self.raw_client, Anthropic | AnthropicVertex)
+
         self.model_setting = self.backend_settings.models[self.model]
 
         if messages[0].get("role") == "system":
@@ -377,7 +439,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.AsyncClient | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -387,7 +449,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
 
     @cached_property
@@ -431,12 +493,23 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                 access_token=self.creds.token,
                 http_client=self.http_client,
             )
-        else:
+        elif self.endpoint.api_schema_type == "default":
             return AsyncAnthropic(
                 api_key=self.endpoint.api_key,
                 base_url=self.endpoint.api_base,
                 http_client=self.http_client,
             )
+        elif self.endpoint.api_schema_type == "openai":
+            return AsyncOpenAICompatibleChatClient(
+                model=self.model,
+                stream=self.stream,
+                temperature=self.temperature,
+                context_length_control=self.context_length_control,
+                random_endpoint=self.random_endpoint,
+                endpoint_id=self.endpoint_id,
+                http_client=self.http_client,
+                backend_name=self.BACKEND_NAME,
+            ).raw_client
 
     @overload
     async def create_completion(
@@ -468,6 +541,21 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
     ) -> AsyncGenerator[ChatCompletionDeltaMessage, Any]:
         pass
 
+    @overload
+    async def create_completion(
+        self,
+        messages: list,
+        model: str | None = None,
+        stream: bool | None = None,
+        temperature: float | None = None,
+        max_tokens: int | None = None,
+        tools: Iterable[ToolParam] | NotGiven = NOT_GIVEN,
+        tool_choice: ToolChoice | NotGiven = NOT_GIVEN,
+        response_format: dict | None = None,
+        **kwargs,
+    ) -> ChatCompletionMessage | AsyncGenerator[ChatCompletionDeltaMessage, Any]:
+        pass
+
     async def create_completion(
         self,
         messages: list,
@@ -491,6 +579,41 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         if isinstance(tool_choice, OpenAINotGiven):
             tool_choice = NOT_GIVEN
 
+        if self.random_endpoint:
+            self.random_endpoint = True
+            self.endpoint_id = random.choice(self.backend_settings.models[self.model].endpoints)
+            self.endpoint = settings.get_endpoint(self.endpoint_id)
+
+        if self.endpoint.api_schema_type == "openai":
+            if self.stream:
+                return AsyncOpenAICompatibleChatClient(
+                    model=self.model,
+                    stream=True,
+                    temperature=self.temperature,
+                    context_length_control=self.context_length_control,
+                    random_endpoint=self.random_endpoint,
+                    endpoint_id=self.endpoint_id,
+                    http_client=self.http_client,
+                    backend_name=self.BACKEND_NAME,
+                ).create_completion(
+                    messages, model, True, temperature, max_tokens, tools, tool_choice, response_format, **kwargs
+                )
+            else:
+                return AsyncOpenAICompatibleChatClient(
+                    model=self.model,
+                    stream=False,
+                    temperature=self.temperature,
+                    context_length_control=self.context_length_control,
+                    random_endpoint=self.random_endpoint,
+                    endpoint_id=self.endpoint_id,
+                    http_client=self.http_client,
+                    backend_name=self.BACKEND_NAME,
+                ).create_completion(
+                    messages, model, False, temperature, max_tokens, tools, tool_choice, response_format, **kwargs
+                )
+
+        assert isinstance(self.raw_client, AsyncAnthropic | AsyncAnthropicVertex)
+
         self.model_setting = self.backend_settings.models[self.model]
 
         if messages[0].get("role") == "system":

vectorvein/chat_clients/base_client.py

@@ -34,7 +34,7 @@ class BaseChatClient(ABC):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.Client | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         self.model = model or self.DEFAULT_MODEL
         self.stream = stream
@@ -44,6 +44,9 @@ class BaseChatClient(ABC):
         self.endpoint_id = endpoint_id
         self.http_client = http_client
 
+        if backend_name is not None:
+            self.BACKEND_NAME = BackendType(backend_name)
+
         self.backend_settings = settings.get_backend(self.BACKEND_NAME)
 
         if endpoint_id:
@@ -156,7 +159,7 @@ class BaseAsyncChatClient(ABC):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.AsyncClient | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         self.model = model or self.DEFAULT_MODEL
         self.stream = stream
@@ -166,6 +169,9 @@ class BaseAsyncChatClient(ABC):
         self.endpoint_id = endpoint_id
         self.http_client = http_client
 
+        if backend_name is not None:
+            self.BACKEND_NAME = BackendType(backend_name)
+
         self.backend_settings = settings.get_backend(self.BACKEND_NAME)
 
         if endpoint_id:

vectorvein/chat_clients/gemini_client.py

@@ -35,7 +35,7 @@ class GeminiChatClient(BaseChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.Client | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -45,7 +45,7 @@ class GeminiChatClient(BaseChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
 
     @cached_property
@@ -263,7 +263,7 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.AsyncClient | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -273,7 +273,7 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
 
     @cached_property

vectorvein/chat_clients/minimax_client.py

@@ -56,7 +56,7 @@ class MiniMaxChatClient(BaseChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.Client | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -66,7 +66,7 @@ class MiniMaxChatClient(BaseChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
         if http_client:
             self.http_client = http_client
@@ -271,7 +271,7 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.AsyncClient | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -281,7 +281,7 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
         if http_client:
             self.http_client = http_client

vectorvein/chat_clients/openai_compatible_client.py

@@ -25,6 +25,7 @@ from ..types.llm_parameters import (
     NOT_GIVEN,
     ToolParam,
     ToolChoice,
+    BackendSettings,
     ChatCompletionMessage,
     ChatCompletionDeltaMessage,
 )
@@ -43,7 +44,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.Client | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -53,7 +54,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
 
     @cached_property
@@ -265,7 +266,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
         random_endpoint: bool = True,
         endpoint_id: str = "",
         http_client: httpx.AsyncClient | None = None,
-        **kwargs,
+        backend_name: str | None = None,
     ):
         super().__init__(
             model,
@@ -275,7 +276,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
             random_endpoint,
             endpoint_id,
             http_client,
-            **kwargs,
+            backend_name,
         )
 
     @cached_property

vectorvein/chat_clients/utils.py

@@ -7,7 +7,7 @@ from typing import Iterable
 import httpx
 import tiktoken
 from anthropic import Anthropic
-from qwen_tokenizer import qwen_tokenizer
+from qwen_tokenizer import get_tokenizer
 from deepseek_tokenizer import deepseek_tokenizer
 
 from ..settings import settings
@@ -208,6 +208,7 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
     elif model.startswith("deepseek"):
         return len(deepseek_tokenizer.encode(text))
     elif model.startswith("qwen"):
+        qwen_tokenizer = get_tokenizer(model)
         return len(qwen_tokenizer.encode(text))
     elif model.startswith("stepfun"):
         model_setting = settings.moonshot.models[model]

vectorvein/types/llm_parameters.py

@@ -23,6 +23,10 @@ class EndpointSetting(BaseModel):
     region: Optional[str] = Field(None, description="The region for the endpoint.")
     api_base: str = Field(None, description="The base URL for the API.")
     api_key: Optional[str] = Field(None, description="The API key for authentication.")
+    api_schema_type: Optional[str] = Field(
+        "default",
+        description="The type of client for the endpoint. Set to 'default' will determine the type automatically.",
+    )
     credentials: Optional[dict] = Field(None, description="Additional credentials if needed.")
     is_azure: bool = Field(False, description="Indicates if the endpoint is for Azure.")
     is_vertex: bool = Field(False, description="Indicates if the endpoint is for Vertex.")

{vectorvein-0.1.34.dist-info → vectorvein-0.1.36.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.34
+Version: 0.1.36
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.1.34.dist-info → vectorvein-0.1.36.dist-info}/RECORD

@@ -1,24 +1,24 @@
-vectorvein-0.1.34.dist-info/METADATA,sha256=vU6f2zVHj_35BGEKhGfkeJ_Cj4uAE1MzTIHu25n5WOI,502
-vectorvein-0.1.34.dist-info/WHEEL,sha256=pM0IBB6ZwH3nkEPhtcp50KvKNX-07jYtnb1g1m6Z4Co,90
-vectorvein-0.1.34.dist-info/entry_points.txt,sha256=6OYgBcLyFCUgeqLgnvMyOJxPCWzgy7se4rLPKtNonMs,34
+vectorvein-0.1.36.dist-info/METADATA,sha256=gRwf2-Irb4TOVXjFue0Vb4ofR_zIM6AS0q4IRv4eIl0,502
+vectorvein-0.1.36.dist-info/WHEEL,sha256=thaaA2w1JzcGC48WYufAs8nrYZjJm8LqNfnXFOFyCC4,90
+vectorvein-0.1.36.dist-info/entry_points.txt,sha256=6OYgBcLyFCUgeqLgnvMyOJxPCWzgy7se4rLPKtNonMs,34
 vectorvein/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 vectorvein/chat_clients/__init__.py,sha256=dW169oK1n3v8Z0uD8itghzlCP72rxiaS-XYn6fvI2xM,16788
-vectorvein/chat_clients/anthropic_client.py,sha256=h82GxBi7h22B7leBuPofwBstxH_c12tEgGjpnKg6UDc,25007
+vectorvein/chat_clients/anthropic_client.py,sha256=FsiAe4NAP7tC6EPrAqSorc5eaHkOfGKJjE6pEmuQAaw,30619
 vectorvein/chat_clients/baichuan_client.py,sha256=CVMvpgjdrZGv0BWnTOBD-f2ufZ3wq3496wqukumsAr4,526
-vectorvein/chat_clients/base_client.py,sha256=wxh7WkzFG4cD4I4t4e6RGe1KiFZc8Z5llh2iVblXEZE,8415
+vectorvein/chat_clients/base_client.py,sha256=7i456Yn-tqY0oPeIj_wHWNGGzCKPAbX5Ufxy2wWGMNY,8653
 vectorvein/chat_clients/deepseek_client.py,sha256=3qWu01NlJAP2N-Ff62d5-CZXZitlizE1fzb20LNetig,526
-vectorvein/chat_clients/gemini_client.py,sha256=VxII45fMjE9JTlOuq4n7R0lNVQRoHTUyTNZE4ICXNrM,18685
+vectorvein/chat_clients/gemini_client.py,sha256=6LKy03iVDXtj0Bn99o9M8rRURMzozX-kf0FoazxPQt0,18739
 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=ljnT9QtVUiySSQSECEv9g2vRfv88K2pPPNZH4sCh838,17204
+vectorvein/chat_clients/minimax_client.py,sha256=0MVMb4g0K_VKnPGHYX81jHiBaQUGWFG5vV8wyClCT_8,17258
 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=gfCTXji8pgFUiultiNDKcmPIGu7lFfQ9VmA8o2_Mm6c,18823
+vectorvein/chat_clients/openai_compatible_client.py,sha256=H7W7BEA71Dh6N42cFuH-nXN9en6oU14OU6RyMmefbnM,18899
 vectorvein/chat_clients/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 vectorvein/chat_clients/qwen_client.py,sha256=-ryh-m9PgsO0fc4ulcCmPTy1155J8YUy15uPoJQOHA0,513
 vectorvein/chat_clients/stepfun_client.py,sha256=zsD2W5ahmR4DD9cqQTXmJr3txrGuvxbRWhFlRdwNijI,519
-vectorvein/chat_clients/utils.py,sha256=1LddLLVf8r8_Hj5LEYrQRus2qfsuXkJPMOu9VsiKMys,24338
+vectorvein/chat_clients/utils.py,sha256=7aSukIJrjFr50o160tgV7t2gUdZUdeNJnJaeJSwlBtQ,24383
 vectorvein/chat_clients/yi_client.py,sha256=RNf4CRuPJfixrwLZ3-DEc3t25QDe1mvZeb9sku2f8Bc,484
 vectorvein/chat_clients/zhipuai_client.py,sha256=Ys5DSeLCuedaDXr3PfG1EW2zKXopt-awO2IylWSwY0s,519
 vectorvein/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -27,8 +27,8 @@ vectorvein/settings/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,
 vectorvein/types/defaults.py,sha256=xefmRNYBGbnWA5kjLLFKN91UM5gnHZ5-kcCNlQRfznk,22095
 vectorvein/types/enums.py,sha256=x_S0IJiEWijOAEiMNdiGDGEWGtmt7TwMriJVDqrDmTo,1637
 vectorvein/types/exception.py,sha256=gnW4GnJ76jND6UGnodk9xmqkcbeS7Cz2rvncA2HpD5E,69
-vectorvein/types/llm_parameters.py,sha256=5Q_NWVjbEhEcG7lYLebiQZ9uQU9rZznFmrUxDZ17yqY,4714
+vectorvein/types/llm_parameters.py,sha256=uvVS3dqqvGMJoiz79LxdKjGj3YSmGD3U_yA0f7mLoTM,4903
 vectorvein/types/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 vectorvein/utilities/media_processing.py,sha256=BujciRmw1GMmc3ELRvafL8STcy6r5b2rVnh27-uA7so,2256
 vectorvein/utilities/retry.py,sha256=9ePuJdeUUGx-qMWfaFxmlOvG_lQPwCQ4UB1z3Edlo34,993
-vectorvein-0.1.34.dist-info/RECORD,,
+vectorvein-0.1.36.dist-info/RECORD,,

{vectorvein-0.1.34.dist-info → vectorvein-0.1.36.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: pdm-backend (2.4.2)
+Generator: pdm-backend (2.4.3)
 Root-Is-Purelib: true
 Tag: py3-none-any