vectorvein 0.1.10__tar.gz → 0.1.12__tar.gz

{vectorvein-0.1.10 → vectorvein-0.1.12}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.10
+Version: 0.1.12
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.1.10 → vectorvein-0.1.12}/pyproject.toml

@@ -16,7 +16,7 @@ description = "Default template for PDM package"
 name = "vectorvein"
 readme = "README.md"
 requires-python = ">=3.10"
-version = "0.1.10"
+version = "0.1.12"
 
 [project.license]
 text = "MIT"

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/chat_clients/__init__.py

@@ -1,5 +1,7 @@
 # @Author: Bi Ying
 # @Date:   2024-07-26 14:48:55
+import httpx
+
 from .base_client import BaseChatClient, BaseAsyncChatClient
 
 from .yi_client import YiChatClient, AsyncYiChatClient
@@ -58,6 +60,9 @@ def create_chat_client(
     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.lower() not in BackendMap["sync"]:
@@ -73,6 +78,9 @@ def create_chat_client(
         stream=stream,
         temperature=temperature,
         context_length_control=context_length_control,
+        random_endpoint=random_endpoint,
+        endpoint_id=endpoint_id,
+        http_client=http_client,
         **kwargs,
     )
 
@@ -83,6 +91,9 @@ def create_async_chat_client(
     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.lower() not in BackendMap["async"]:
@@ -98,6 +109,9 @@ def create_async_chat_client(
         stream=stream,
         temperature=temperature,
         context_length_control=context_length_control,
+        random_endpoint=random_endpoint,
+        endpoint_id=endpoint_id,
+        http_client=http_client,
         **kwargs,
     )
 

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/chat_clients/anthropic_client.py

@@ -3,6 +3,7 @@
 import json
 import random
 
+import httpx
 from openai._types import NotGiven as OpenAINotGiven
 from anthropic import Anthropic, AnthropicVertex, AsyncAnthropic, AsyncAnthropicVertex
 from anthropic._types import NotGiven, NOT_GIVEN
@@ -97,6 +98,7 @@ class AnthropicChatClient(BaseChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.Client | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -106,6 +108,7 @@ class AnthropicChatClient(BaseChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
 
@@ -118,6 +121,7 @@ class AnthropicChatClient(BaseChatClient):
         max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
+        **kwargs,
     ):
         if model is not None:
             self.model = model
@@ -182,11 +186,13 @@ class AnthropicChatClient(BaseChatClient):
                 base_url=base_url,
                 project_id=self.endpoint.credentials.get("quota_project_id"),
                 access_token=self.creds.token,
+                http_client=self.http_client,
             )
         else:
             self._client = Anthropic(
                 api_key=self.endpoint.api_key,
                 base_url=self.endpoint.api_base,
+                http_client=self.http_client,
             )
 
         tools_params = refactor_tool_use_params(tools) if tools else tools
@@ -210,6 +216,7 @@ class AnthropicChatClient(BaseChatClient):
             max_tokens=max_tokens,
             tools=tools_params,
             tool_choice=tool_choice,
+            **kwargs,
         )
 
         if self.stream:
@@ -297,6 +304,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.AsyncClient | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -306,6 +314,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
 
@@ -318,6 +327,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
         max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
+        **kwargs,
     ):
         if model is not None:
             self.model = model
@@ -382,11 +392,13 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
                 base_url=base_url,
                 project_id=self.endpoint.credentials.get("quota_project_id"),
                 access_token=self.creds.token,
+                http_client=self.http_client,
             )
         else:
             self._client = AsyncAnthropic(
                 api_key=self.endpoint.api_key,
                 base_url=self.endpoint.api_base,
+                http_client=self.http_client,
             )
 
         tools_params = refactor_tool_use_params(tools) if tools else tools
@@ -410,6 +422,7 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
             max_tokens=max_tokens,
             tools=tools_params,
             tool_choice=tool_choice,
+            **kwargs,
         )
 
         if self.stream:

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/chat_clients/base_client.py

@@ -3,6 +3,7 @@
 from abc import ABC, abstractmethod
 from typing import Generator, AsyncGenerator, Any
 
+import httpx
 from openai._types import NotGiven, NOT_GIVEN
 
 from ..settings import settings
@@ -23,6 +24,7 @@ class BaseChatClient(ABC):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.Client | None = None,
         **kwargs,
     ):
         self.model = model or self.DEFAULT_MODEL
@@ -31,6 +33,7 @@ class BaseChatClient(ABC):
         self.context_length_control = context_length_control
         self.random_endpoint = random_endpoint
         self.endpoint_id = endpoint_id
+        self.http_client = http_client
 
         self.backend_settings = settings.get_backend(self.BACKEND_NAME)
 
@@ -84,6 +87,7 @@ class BaseAsyncChatClient(ABC):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.AsyncClient | None = None,
         **kwargs,
     ):
         self.model = model or self.DEFAULT_MODEL
@@ -92,6 +96,7 @@ class BaseAsyncChatClient(ABC):
         self.context_length_control = context_length_control
         self.random_endpoint = random_endpoint
         self.endpoint_id = endpoint_id
+        self.http_client = http_client
 
         self.backend_settings = settings.get_backend(self.BACKEND_NAME)
 

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/chat_clients/gemini_client.py

@@ -25,6 +25,7 @@ class GeminiChatClient(BaseChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.Client | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -34,6 +35,7 @@ class GeminiChatClient(BaseChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
 
@@ -107,7 +109,11 @@ class GeminiChatClient(BaseChatClient):
 
             def generator():
                 result = {"content": ""}
-                with httpx.stream("POST", url, headers=headers, params=params, json=request_body) as response:
+                if self.http_client:
+                    client = self.http_client
+                else:
+                    client = httpx.Client()
+                with client.stream("POST", url, headers=headers, params=params, json=request_body) as response:
                     for chunk in response.iter_lines():
                         message = {"content": ""}
                         if not chunk.startswith("data:"):
@@ -142,13 +148,17 @@ class GeminiChatClient(BaseChatClient):
             return generator()
         else:
             url = f"{self.endpoint.api_base}/models/{self.model_setting.id}:generateContent"
-            response = httpx.post(url, json=request_body, headers=headers, params=params, timeout=None).json()
+            if self.http_client:
+                client = self.http_client
+            else:
+                client = httpx.Client()
+            response = client.post(url, json=request_body, headers=headers, params=params, timeout=None).json()
             result = {
                 "content": "",
                 "usage": {
-                    "prompt_tokens": response["usageMetadata"]["promptTokenCount"],
-                    "completion_tokens": response["usageMetadata"]["candidatesTokenCount"],
-                    "total_tokens": response["usageMetadata"]["totalTokenCount"],
+                    "prompt_tokens": response.get("usageMetadata", {}).get("promptTokenCount", 0),
+                    "completion_tokens": response.get("usageMetadata", {}).get("candidatesTokenCount", 0),
+                    "total_tokens": response.get("usageMetadata", {}).get("totalTokenCount", 0),
                 },
             }
             tool_calls = []
@@ -185,6 +195,7 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.AsyncClient | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -194,6 +205,7 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
 
@@ -267,7 +279,10 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
 
             async def generator():
                 result = {"content": ""}
-                client = httpx.AsyncClient()
+                if self.http_client:
+                    client = self.http_client
+                else:
+                    client = httpx.AsyncClient()
                 async with client.stream("POST", url, headers=headers, params=params, json=request_body) as response:
                     async for chunk in response.aiter_lines():
                         message = {"content": ""}
@@ -303,15 +318,19 @@ class AsyncGeminiChatClient(BaseAsyncChatClient):
             return generator()
         else:
             url = f"{self.endpoint.api_base}/models/{self.model_setting.id}:generateContent"
-            async with httpx.AsyncClient(headers=headers, params=params, timeout=None) as client:
-                response = await client.post(url, json=request_body)
+            if self.http_client:
+                client = self.http_client
+            else:
+                client = httpx.AsyncClient()
+            async with client:
+                response = await client.post(url, json=request_body, headers=headers, params=params, timeout=None)
                 response = response.json()
                 result = {
                     "content": "",
                     "usage": {
-                        "prompt_tokens": response["usageMetadata"]["promptTokenCount"],
-                        "completion_tokens": response["usageMetadata"]["candidatesTokenCount"],
-                        "total_tokens": response["usageMetadata"]["totalTokenCount"],
+                        "prompt_tokens": response.get("usageMetadata", {}).get("promptTokenCount", 0),
+                        "completion_tokens": response.get("usageMetadata", {}).get("candidatesTokenCount", 0),
+                        "total_tokens": response.get("usageMetadata", {}).get("totalTokenCount", 0),
                     },
                 }
                 tool_calls = []

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/chat_clients/minimax_client.py

@@ -48,6 +48,7 @@ class MiniMaxChatClient(BaseChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.Client | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -57,9 +58,13 @@ class MiniMaxChatClient(BaseChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
-        self.http_client = httpx.Client()
+        if http_client:
+            self.http_client = http_client
+        else:
+            self.http_client = httpx.Client()
 
     def create_completion(
         self,
@@ -70,6 +75,7 @@ class MiniMaxChatClient(BaseChatClient):
         max_tokens: int | None = None,
         tools: list | None = None,
         tool_choice: str = "auto",
+        **kwargs,
     ):
         if model is not None:
             self.model = model
@@ -135,6 +141,7 @@ class MiniMaxChatClient(BaseChatClient):
             "stream": self.stream,
             "mask_sensitive_info": False,
             **tools_params,
+            **kwargs,
         }
 
         if self.stream:
@@ -206,6 +213,7 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.AsyncClient | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -215,9 +223,13 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
-        self.http_client = httpx.AsyncClient()
+        if http_client:
+            self.http_client = http_client
+        else:
+            self.http_client = httpx.AsyncClient()
 
     async def create_completion(
         self,
@@ -228,6 +240,7 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
         max_tokens: int | None = None,
         tools: list | None = None,
         tool_choice: str = "auto",
+        **kwargs,
     ):
         if model is not None:
             self.model = model
@@ -291,6 +304,7 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
             "stream": self.stream,
             "mask_sensitive_info": False,
             **tools_params,
+            **kwargs,
         }
 
         if self.stream:

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/chat_clients/openai_compatible_client.py

@@ -3,6 +3,7 @@
 import json
 import random
 
+import httpx
 from openai._types import NotGiven, NOT_GIVEN
 from openai._streaming import Stream, AsyncStream
 from openai.types.chat import ChatCompletion, ChatCompletionChunk
@@ -33,6 +34,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.Client | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -42,6 +44,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
 
@@ -54,6 +57,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
         max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
+        **kwargs,
     ):
         if model is not None:
             self.model = model
@@ -73,12 +77,14 @@ class OpenAICompatibleChatClient(BaseChatClient):
             self._client = AzureOpenAI(
                 azure_endpoint=self.endpoint.api_base,
                 api_key=self.endpoint.api_key,
-                api_version="2024-05-01-preview",
+                api_version="2024-08-01-preview",
+                http_client=self.http_client,
             )
         else:
             self._client = OpenAI(
                 api_key=self.endpoint.api_key,
                 base_url=self.endpoint.api_base,
+                http_client=self.http_client,
             )
 
         if self.context_length_control == ContextLengthControlType.Latest:
@@ -120,6 +126,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
             temperature=self.temperature,
             max_tokens=max_tokens,
             **tools_params,
+            **kwargs,
         )
 
         if self.stream:
@@ -186,6 +193,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
         context_length_control: ContextLengthControlType = defs.CONTEXT_LENGTH_CONTROL,
         random_endpoint: bool = True,
         endpoint_id: str = "",
+        http_client: httpx.AsyncClient | None = None,
         **kwargs,
     ):
         super().__init__(
@@ -195,6 +203,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
             context_length_control,
             random_endpoint,
             endpoint_id,
+            http_client,
             **kwargs,
         )
 
@@ -207,6 +216,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
         max_tokens: int | None = None,
         tools: list | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
+        **kwargs,
     ):
         if model is not None:
             self.model = model
@@ -226,12 +236,14 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
             self._client = AsyncAzureOpenAI(
                 azure_endpoint=self.endpoint.api_base,
                 api_key=self.endpoint.api_key,
-                api_version="2024-05-01-preview",
+                api_version="2024-08-01-preview",
+                http_client=self.http_client,
             )
         else:
             self._client = AsyncOpenAI(
                 api_key=self.endpoint.api_key,
                 base_url=self.endpoint.api_base,
+                http_client=self.http_client,
             )
 
         if self.context_length_control == ContextLengthControlType.Latest:
@@ -273,6 +285,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
             temperature=self.temperature,
             max_tokens=max_tokens,
             **tools_params,
+            **kwargs,
         )
 
         if self.stream:

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/types/defaults.py

@@ -213,6 +213,13 @@ ZHIPUAI_MODELS = {
         "response_format_available": False,
         "max_output_tokens": 4095,
     },
+    "glm-4-plus": {
+        "id": "glm-4-plus",
+        "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,
@@ -255,6 +262,13 @@ ZHIPUAI_MODELS = {
         "response_format_available": False,
         "max_output_tokens": 1024,
     },
+    "glm-4v-plus": {
+        "id": "glm-4v-plus",
+        "context_length": 2000,
+        "function_call_available": False,
+        "response_format_available": False,
+        "max_output_tokens": 1024,
+    },
 }
 
 # Mistral models

{vectorvein-0.1.10 → vectorvein-0.1.12}/src/vectorvein/types/llm_parameters.py

@@ -69,6 +69,7 @@ class ChatCompletionMessage(BaseModel):
 
     usage: Optional[Usage] = None
 
+
 class ChatCompletionDeltaMessage(BaseModel):
     content: Optional[str] = None
 

{vectorvein-0.1.10 → vectorvein-0.1.12}/tests/sample_settings.py

@@ -39,7 +39,7 @@ sample_settings = {
         },
         {
             "id": "deepseek-default",
-            "api_base": "https://api.deepseek.com/v1",
+            "api_base": "https://api.deepseek.com/beta",
             "api_key": "",
         },
         {
@@ -80,6 +80,10 @@ sample_settings = {
     },
     "openai": {
         "models": {
+            "gpt-4o-mini": {
+                "id": "gpt-4o-mini",
+                "endpoints": ["azure-openai"],
+            },
             "gpt-4o": {
                 "id": "gpt-4o",
                 "endpoints": ["azure-openai"],
@@ -362,7 +366,7 @@ sample_settings = {
         },
         {
             "id": "deepseek-default",
-            "api_base": "https://api.deepseek.com/v1",
+            "api_base": "https://api.deepseek.com/beta",
             "api_key": "sk-6dad42e7154743cd80b77dff5d0ecaaa",
         },
         {

vectorvein-0.1.12/tests/test_chat_prefix.py

@@ -0,0 +1,23 @@
+# @Author: Bi Ying
+# @Date:   2024-07-27 11:51:28
+import time
+
+from vectorvein.settings import settings
+from vectorvein.types.enums import BackendType
+from vectorvein.chat_clients import create_chat_client
+
+from sample_settings import sample_settings
+
+settings.load(sample_settings)
+messages = [
+    {"role": "user", "content": "Please write quick sort code"},
+    {"role": "assistant", "content": "```python\n", "prefix": True},
+]
+
+
+start_time = time.perf_counter()
+client = create_chat_client(backend=BackendType.DeepSeek, model="deepseek-chat", stream=False)
+response = client.create_completion(messages=messages, stop=["\n```"])
+print(response)
+end_time = time.perf_counter()
+print(f"Stream time elapsed: {end_time - start_time} seconds")

vectorvein-0.1.12/tests/test_http_client.py

@@ -0,0 +1,24 @@
+# @Author: Bi Ying
+# @Date:   2024-07-27 11:51:28
+import time
+
+import httpx
+from vectorvein.settings import settings
+from vectorvein.types.enums import BackendType
+from vectorvein.chat_clients import create_chat_client
+
+from sample_settings import sample_settings
+
+settings.load(sample_settings)
+messages = [
+    {"role": "user", "content": "Please write quick sort code"},
+]
+
+
+start_time = time.perf_counter()
+http_client = httpx.Client()
+client = create_chat_client(backend=BackendType.DeepSeek, model="deepseek-chat", stream=False, http_client=http_client)
+response = client.create_completion(messages=messages)
+print(response)
+end_time = time.perf_counter()
+print(f"Stream time elapsed: {end_time - start_time} seconds")

vectorvein-0.1.12/tests/test_stop.py

@@ -0,0 +1,25 @@
+# @Author: Bi Ying
+# @Date:   2024-07-27 11:51:28
+import time
+
+from vectorvein.settings import settings
+from vectorvein.types.enums import BackendType
+from vectorvein.chat_clients import create_chat_client
+
+from sample_settings import sample_settings
+
+settings.load(sample_settings)
+messages = [
+    {
+        "role": "user",
+        "content": "节点名称是 FileLoader，FileLoader 节点连到 OCR 节点，使用 mermaid 语法表示流程图。直接开始补全，不要有任何解释。\n\n```mermaid\n",
+    }
+]
+
+
+start_time = time.perf_counter()
+client = create_chat_client(backend=BackendType.DeepSeek, model="deepseek-chat", stream=False)
+response = client.create_completion(messages=messages, stop=["\n```"])
+print(response)
+end_time = time.perf_counter()
+print(f"Stream time elapsed: {end_time - start_time} seconds")