openrouter-provider 0.1.0__py3-none-any.whl → 1.0.0__py3-none-any.whl

OpenRouterProvider/Tool.py → tool.py

@@ -1,10 +1,11 @@
+from __future__ import annotations
 import inspect
 from functools import wraps
-from typing import get_type_hints, get_origin, get_args
+from typing import get_type_hints, get_origin, get_args, Callable, Any, Dict
 
 
 class tool_model:
-    def __init__(self, func):
+    def __init__(self, func: Callable) -> None:
         self.func = func
         self.name = func.__name__
 
@@ -12,22 +13,20 @@ class tool_model:
         type_hints = get_type_hints(func)
 
         type_map = {
-            int:   "integer",
-            str:   "string",
+            int: "integer",
+            str: "string",
             float: "number",
-            bool:  "boolean",
+            bool: "boolean",
         }
 
-        properties = {}
-        required = []
+        properties: Dict[str, Dict[str, Any]] = {}
+        required: list[str] = []
 
-        # Build JSON schema properties
         for name, param in sig.parameters.items():
             anno = type_hints.get(name, None)
             origin = get_origin(anno)
 
             if origin is list:
-                # Handle list element types
                 (elem_type,) = get_args(anno)
                 json_item_type = type_map.get(elem_type, "string")
                 schema = {
@@ -36,7 +35,6 @@ class tool_model:
                     "description": name,
                 }
             else:
-                # Primitive types
                 json_type = type_map.get(anno, "string")
                 schema = {
                     "type": json_type,
@@ -47,7 +45,6 @@ class tool_model:
             if param.default is inspect._empty:
                 required.append(name)
 
-        # Attach tool definition metadata
         self.tool_definition = {
             "type": "function",
             "function": {
@@ -65,7 +62,7 @@ class tool_model:
 
         wraps(func)(self)
 
-    def __call__(self, *args, **kwargs):
+    def __call__(self, *args: Any, **kwargs: Any) -> Any:
         return self.func(*args, **kwargs)
     
     

OpenRouterProvider/Chat_message.py

@@ -1,142 +0,0 @@
-from .LLMs import LLMModel
-
-from enum import Enum
-from PIL import Image
-import base64
-from io import BytesIO
-from dataclasses import dataclass
-
-from openai.types.chat import ChatCompletion
-
-
-class Role(Enum):
-    system = "system"
-    user = "user"
-    ai = "assistant"
-    agent = "agent"
-    tool = "tool"
-
-
-@dataclass
-class ToolCall:
-    id: str
-    name: str
-    arguments: dict
-    result: any = ""
-
-
-class Chat_message:
-    def __init__(self, 
-                 text: str, 
-                 images: list[Image.Image]=None, 
-                 role: Role=Role.user, 
-                 answerdBy: LLMModel=None, 
-                 raw_response: ChatCompletion=None
-                ) -> None:
-        self.role = role
-        self.text = text
-        self.images = self._process_image(images=images)
-        self.answeredBy: LLMModel = answerdBy
-        
-        self.tool_calls: list[ToolCall] = []
-        self.raw_resoonse: ChatCompletion = raw_response
-
-
-    def __str__(self) -> str:
-        # ANSI color codes for blue, green, and reset (to default)
-        BLUE = "\033[34m"
-        GREEN = "\033[32m"
-        RESET = "\033[0m"
-        
-        message = ""
-        
-        if self.role == Role.system:
-            message = "---------------------- System ----------------------\n"
-        elif self.role == Role.user:
-            message = BLUE + "----------------------- User -----------------------\n" + RESET
-        elif self.role == Role.ai:
-            message = GREEN + "--------------------- Assistant --------------------\n" + RESET
-        
-        # Append text and reset color formatting at the end
-        message += self.text + RESET + "\n"
-        
-        return message
-        
-    def _process_image(self, images: list):
-        """
-        Process a list of images by resizing them to maintain aspect ratio and then converting them to base64 format.
-        
-        Args:
-            images (list): A list of image objects to be processed.
-
-        Returns:
-            list: A list of base64-encoded image strings if input is not None/empty, otherwise `None`.
-        
-        Note:
-            - Images should be provided as a "list" even if there is only a single image to process.
-        """
-        if images == None:
-            return None
-
-        base64_images = []
-        for image in images:
-            if image.mode == "RGBA":
-                image = image.convert("RGB")
-            
-            image = self._resize_image_aspect_ratio(image=image)
-            image = self._convert_to_base64(image=image)
-            base64_images.append(image)
-
-        return base64_images
-        
-    def _convert_to_base64(self, image: Image) -> str:
-        """
-        Convert an image to a base64-encoded string.
-
-        Args:
-            image (Image): The image object to be converted to base64 format.
-
-        Returns:
-            str: The base64-encoded string representation of the image.
-
-        Note:
-            - The image format will default to 'JPEG' if the format is not specified.
-        """
-        buffered = BytesIO()
-        format = image.format if image.format else 'JPEG'
-        image.save(buffered, format=format)
-        img_bytes = buffered.getvalue()
-        img_base64 = base64.b64encode(img_bytes).decode('utf-8')
-        
-        return img_base64
-    
-    def _resize_image_aspect_ratio(self, image: Image, target_length=1024):
-        """
-        Resize an image to a target length while maintaining its aspect ratio.
-
-        Args:
-            image (Image): The image object to be resized.
-            target_length (int, optional): The target length for the larger dimension (default is 1024).
-
-        Returns:
-            Image: The resized image object with maintained aspect ratio.
-
-        Note:
-            - The smaller dimension is scaled proportionally based on the larger dimension to maintain aspect ratio.
-            - If the image's aspect ratio is non-square, the target_length is applied to the larger dimension.
-        """
-        
-        width, height = image.size
-        
-        if width > height:
-            new_width = target_length
-            new_height = int((target_length / width) * height)
-        else:
-            new_height = target_length
-            new_width = int((target_length / height) * width)
-
-        resized_image = image.resize((new_width, new_height))
-        
-        return resized_image
-    
-        

OpenRouterProvider/OpenRouterProvider.py

@@ -1,266 +0,0 @@
-
-# structured output
-# https://note.com/brave_quince241/n/n60a5759c8f05
-
-import logging
-from .Chat_message import *
-from .Tool import tool_model
-from .LLMs import *
-
-from openai import OpenAI, AsyncOpenAI
-from openai.types.chat import ChatCompletionChunk
-from dotenv import load_dotenv
-import os, time
-from dataclasses import dataclass, field, asdict
-from typing import List, Optional, Literal, Iterator, AsyncIterator
-from pprint import pprint
-from pydantic import BaseModel
-
-
-# エラーのみ表示、詳細なトレースバック付き
-logging.basicConfig(level=logging.ERROR, format="%(asctime)s - %(levelname)s - %(message)s")
-logger = logging.getLogger(__name__)
-
-
-@dataclass
-class ProviderConfig:
-    order: Optional[List[str]] = None
-    allow_fallbacks: bool = None
-    require_parameters: bool = None
-    data_collection: Literal["allow", "deny"] = None
-    only: Optional[List[str]] = None
-    ignore: Optional[List[str]] = None
-    quantizations: Optional[List[str]] = None
-    sort: Optional[Literal["price", "throughput"]] = None
-    max_price: Optional[dict] = None
-
-    def to_dict(self) -> dict:
-        return {k: v for k, v in asdict(self).items() if v is not None}
-
-
-class OpenRouterProvider:
-    def __init__(self) -> None:
-        load_dotenv()
-        api_key = os.getenv("OPENROUTER_API_KEY")
-        if not api_key:
-            logger.error("OPENROUTER_API_KEY is not set in environment variables.")
-        self.client = OpenAI(
-            base_url="https://openrouter.ai/api/v1",
-            api_key=api_key,
-        )
-        self.async_client = AsyncOpenAI(
-            base_url="https://openrouter.ai/api/v1",
-            api_key=api_key,
-        )
-
-    def make_prompt(
-        self, 
-        system_prompt: Chat_message,
-        querys: list[Chat_message]
-    ) -> list[dict]:
-        messages = [{"role": "system", "content": system_prompt.text}]
-
-        for query in querys:
-            if query.role == Role.user:
-                if query.images is None:
-                    messages.append({"role": "user", "content": query.text})
-                else:
-                    content = [{"type": "text", "text": query.text}]
-                    for img in query.images[:50]:
-                        content.append(
-                            {"type": "image_url",
-                             "image_url": {"url": f"data:image/jpeg;base64,{img}"}})
-                    messages.append({"role": "user", "content": content})
-
-            elif query.role == Role.ai or query.role == Role.tool:
-                assistant_msg = {"role": "assistant"}
-                assistant_msg["content"] = query.text or None
-
-                if query.tool_calls:
-                    assistant_msg["tool_calls"] = [
-                        {
-                            "id": str(t.id),
-                            "type": "function",
-                            "function": {
-                                "name": t.name,
-                                "arguments": t.arguments
-                            }
-                        }
-                        for t in query.tool_calls
-                    ]
-                messages.append(assistant_msg)
-
-            for t in query.tool_calls:
-                messages.append({
-                    "role": "tool",
-                    "tool_call_id": str(t.id),
-                    "content": str(t.result)
-                })
-            
-        return messages
-
-    def invoke(
-        self, 
-        model: LLMModel, 
-        system_prompt: Chat_message, 
-        querys: list[Chat_message], 
-        tools: list[tool_model] = [], 
-        provider: ProviderConfig = None,
-        temperature: float = 0.3
-    ) -> Chat_message:
-        try:
-            messages = self.make_prompt(system_prompt, querys)
-
-            tool_defs = [tool.tool_definition for tool in tools] if tools else None
-            provider_dict = provider.to_dict() if provider else None
-            
-            response = self.client.chat.completions.create(
-                model=model.name,
-                temperature=temperature,
-                messages=messages,
-                tools=tool_defs,
-                extra_body={"provider": provider_dict},
-            )
-
-            reply = Chat_message(text=response.choices[0].message.content, role=Role.ai, raw_response=response)
-
-            if response.choices[0].message.tool_calls:
-                reply.role = Role.tool
-                for tool in response.choices[0].message.tool_calls:
-                    reply.tool_calls.append(ToolCall(id=tool.id, name=tool.function.name, arguments=tool.function.arguments))
-            return reply
-
-        except Exception as e:
-            logger.exception(f"An error occurred while invoking the model: {e.__class__.__name__}: {str(e)}")
-            return Chat_message(text="Fail to get response. Please see the error message.", role=Role.ai, raw_response=None)
-        
-    def invoke_stream(
-        self, 
-        model: LLMModel, 
-        system_prompt: Chat_message, 
-        querys: list[Chat_message], 
-        tools: list[tool_model] = [], 
-        provider: ProviderConfig = None,
-        temperature: float = 0.3
-    ) -> Iterator[ChatCompletionChunk]:
-        # chunk example
-        # ChatCompletionChunk(id='gen-1746748260-mdKZLTs9QY7MmUxWKb8V', choices=[Choice(delta=ChoiceDelta(content='!', function_call=None, refusal=None, role='assistant', tool_calls=None), finish_reason=None, index=0, logprobs=None, native_finish_reason=None)], created=1746748260, model='openai/gpt-4o-mini', object='chat.completion.chunk', service_tier=None, system_fingerprint='fp_e2f22fdd96', usage=None, provider='OpenAI')
-        
-        # ChatCompletionChunk(id='gen-1746748260-mdKZLTs9QY7MmUxWKb8V', choices=[Choice(delta=ChoiceDelta(content='', function_call=None, refusal=None, role='assistant', tool_calls=None), finish_reason='stop', index=0, logprobs=None, native_finish_reason='stop')], created=1746748260, model='openai/gpt-4o-mini', object='chat.completion.chunk', service_tier=None, system_fingerprint='fp_e2f22fdd96', usage=None, provider='OpenAI')
-        
-        # ChatCompletionChunk(id='gen-1746748260-mdKZLTs9QY7MmUxWKb8V', choices=[Choice(delta=ChoiceDelta(content='', function_call=None, refusal=None, role='assistant', tool_calls=None), finish_reason=None, index=0, logprobs=None, native_finish_reason=None)], created=1746748260, model='openai/gpt-4o-mini', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=CompletionUsage(completion_tokens=54, prompt_tokens=61, total_tokens=115, completion_tokens_details=CompletionTokensDetails(reasoning_tokens=0), prompt_tokens_details={'cached_tokens': 0}), provider='OpenAI')
-        
-        try:
-            messages = self.make_prompt(system_prompt, querys)
-
-            tool_defs = [tool.tool_definition for tool in tools] if tools else None
-            provider_dict = provider.to_dict() if provider else None
-
-            response = self.client.chat.completions.create(
-                model=model.name,
-                temperature=temperature,
-                messages=messages,
-                tools=tool_defs,
-                extra_body={"provider": provider_dict},
-                stream=True
-            )
-            
-            return response
-
-        except Exception as e:
-            logger.exception(f"An error occurred while invoking the model: {e.__class__.__name__}: {str(e)}")
-            return Chat_message(text="Fail to get response. Please see the error message.", role=Role.ai, raw_response=None)
-
-    async def async_invoke(
-        self, model: LLMModel, 
-        system_prompt: Chat_message, 
-        querys: list[Chat_message], 
-        tools: list[tool_model] = [], 
-        provider: ProviderConfig = None,
-        temperature: float = 0.3
-    ) -> Chat_message:
-        try:
-            messages = self.make_prompt(system_prompt, querys)
-
-            tool_defs = [tool.tool_definition for tool in tools] if tools else None
-            provider_dict = provider.to_dict() if provider else None
-
-            response = await self.async_client.chat.completions.create(
-                model=model.name,
-                temperature=temperature,
-                messages=messages,
-                tools=tool_defs,
-                extra_body={"provider": provider_dict}
-            )
-
-            reply = Chat_message(text=response.choices[0].message.content, role=Role.ai, raw_response=response)
-
-            if response.choices[0].message.tool_calls:
-                reply.role = Role.tool
-                for tool in response.choices[0].message.tool_calls:
-                    reply.tool_calls.append(ToolCall(id=tool.id, name=tool.function.name, arguments=tool.function.arguments))
-            return reply
-
-        except Exception as e:
-            logger.exception(f"An error occurred while asynchronously invoking the model: {e.__class__.__name__}: {str(e)}")
-            return Chat_message(text="Fail to get response. Please see the error message.", role=Role.ai, raw_response=None)
-        
-    async def async_invoke_stream(
-        self,
-        model: LLMModel,
-        system_prompt: Chat_message,
-        querys: list[Chat_message],
-        tools: list[tool_model] = [],
-        provider: ProviderConfig = None,
-        temperature: float = 0.3
-    ) -> AsyncIterator[ChatCompletionChunk]:
-        try:
-            messages = self.make_prompt(system_prompt, querys)
-
-            tool_defs = [tool.tool_definition for tool in tools] if tools else None
-            provider_dict = provider.to_dict() if provider else None
-
-            response = await self.async_client.chat.completions.create(
-                model=model.name,
-                temperature=temperature,
-                messages=messages,
-                tools=tool_defs,
-                extra_body={"provider": provider_dict},
-                stream=True
-            )
-
-            async for chunk in response:
-                yield chunk
-
-        except Exception as e:
-            logger.exception(f"An error occurred while asynchronously streaming the model: {e.__class__.__name__}: {str(e)}")
-            return
-        
-    def structured_output(
-        self, 
-        model: LLMModel, 
-        system_prompt: Chat_message, 
-        querys: list[Chat_message], 
-        provider: ProviderConfig = None, 
-        json_schema: BaseModel = None,
-        temperature: float = 0.3
-    ) -> BaseModel:
-        try:
-            messages = self.make_prompt(system_prompt, querys)
-            provider_dict = provider.to_dict() if provider else None
-            
-            response = self.client.chat.completions.create(
-                model=model.name,
-                temperature=temperature,
-                messages=messages,
-                response_format={"type": "json_schema", "json_schema": {"name": json_schema.__name__, "schema": json_schema.model_json_schema()}},
-                extra_body={"provider": provider_dict},
-            )
-
-            return json_schema.model_validate_json(response.choices[0].message.content)
-
-        except Exception as e:
-            logger.exception(f"An error occurred while invoking structured output: {e.__class__.__name__}: {str(e)}")
-            return Chat_message(text="Fail to get response. Please see the error message.", role=Role.ai, raw_response=None)
-        
-    

openrouter_provider-0.1.0.dist-info/METADATA

@@ -1,232 +0,0 @@
-Metadata-Version: 2.4
-Name: openrouter-provider
-Version: 0.1.0
-Summary: This is an unofficial wrapper of OpenRouter.
-Author-email: Keisuke Miyamto <aichiboyhighschool@gmail.com>
-Requires-Python: >=3.7
-Description-Content-Type: text/markdown
-Requires-Dist: annotated-types
-Requires-Dist: anyio
-Requires-Dist: certifi
-Requires-Dist: distro
-Requires-Dist: h11
-Requires-Dist: httpcore
-Requires-Dist: httpx
-Requires-Dist: idna
-Requires-Dist: jiter
-Requires-Dist: openai
-Requires-Dist: pillow
-Requires-Dist: pydantic
-Requires-Dist: pydantic_core
-Requires-Dist: python-dotenv
-Requires-Dist: sniffio
-Requires-Dist: tqdm
-Requires-Dist: typing-inspection
-Requires-Dist: typing_extensions
-
-## Introduction
-
-Welcome to **openrouter-provider**, an unofficial Python wrapper for the OpenRouter API. This library lets you easily integrate with OpenRouter models, manage chat sessions, process images, and call tools within your Python application.
-
-
-## Features
-
-* Simple chat interface with system, user, assistant, and tool roles
-* Automatic image resizing and Base64 encoding
-* Built-in tool decorator for defining custom functions
-
-
-## Installation
-
-### From PyPI
-
-```bash
-pip3 install openrouter-provider
-```
-
-### From Source
-
-```bash
-git clone https://github.com/yourusername/openrouter-provider.git
-cd openrouter-provider
-pip3 install .
-```
-
-
-
-## Configuration
-
-1. Create a `.env` file in your project root.
-2. Add your OpenRouter API key:
-
-   ```bash
-   OPENROUTER_API_KEY=your_api_key_here
-   ```
-
-
-
-## Usage
-
-### Basic chat bot
-Chat history is automatically sent, by Chatbot_manager. If you want to delete chat history, use `clear_memory` method.
-
-```python
-from OpenRouterProvider.Chatbot_manager import Chat_message, Chatbot_manager
-from OpenRouterProvider.LLMs import gpt_4o_mini
-
-# Declare chat bot
-ai = Chatbot_manager(system_prompt="Please answer in English.")
-
-# Send query
-query = Chat_message(text="Introduce yourself, please.")
-response = ai.invoke(model=gpt_4o_mini, query=query)
-print(response.text)
-
-# Send next query. Chatbot_manager automatically handle chat history.
-query = Chat_message(text="Tell me a short story.")
-response = ai.invoke(model=gpt_4o_mini, query=query)
-print(response.text)
-
-# Print all chat history
-ai.print_memory()  
-
-# Delete all chat history
-ai.clear_memory()
-```
-
-### Chat bot with images
-You can use images in the chat.
-
-```python
-from OpenRouterProvider.Chatbot_manager import Chat_message, Chatbot_manager
-from OpenRouterProvider.LLMs import gpt_4o_mini
-from PIL import Image
-
-dog = Image.open("dog.jpg")
-cat = Image.open("cat.jpg")
-
-# Send query with images
-ai = Chatbot_manager(system_prompt="Please answer in English.")
-query = Chat_message(text="What can you see in the images?", images=[dog, cat])
-response = ai.invoke(model=gpt_4o_mini, query=query)
-print(response.text) 
-```
-
-### With tools
-
-Use the `@tool_model` decorator to expose Python functions as callable tools in the chat. Tools are automatically processed by Chat_manager, so you don't need to care it.
-
-```python
-from OpenRouterProvider.Chatbot_manager import Chat_message, Chatbot_manager
-from OpenRouterProvider.LLMs import gpt_4o_mini
-from OpenRouterProvider.Tool import tool_model
-
-@tool_model
-def get_user_info():
-    """
-    Return user's personal info: name, age, and address.
-    """
-    return "name: Alice\nage: 30\naddress: Wonderland"
-
-ai = Chatbot_manager(system_prompt="Please answer in English.", tools=[get_user_info])
-query = Chat_message(text="What is the name, age, address of the user?")
-response = ai.invoke(model=gpt_4o_mini, query=query)
-ai.print_memory()
-```
-
-## Advanced Usage
-### Prebuilt and Custom Model Usage
-
-You can use prebuilt models defined or declare your own custom models easily.
-This library provides many ready-to-use models from OpenAI, Anthropic, Google, and others.
-
-```python
-from OpenRouterProvider.Chatbot_manager import Chat_message, Chatbot_manager
-from OpenRouterProvider.LLMs import gpt_4o, claude_3_7_sonnet
-
-# Use OpenAI GPT-4o
-ai = Chatbot_manager(system_prompt="Please answer in English.")
-query = Chat_message(text="Tell me a joke.")
-response = ai.invoke(model=gpt_4o, query=query)
-print(response.text)
-
-# Use Anthropic Claude 3.7 Sonnet
-query = Chat_message(text="Summarize the story of Hamlet.")
-response = ai.invoke(model=claude_3_7_sonnet, query=query)
-print(response.text)
-```
-
-Available prebuilt models include:
-
-#### **OpenAI**
-
-* `gpt_4o`
-* `gpt_4o_mini`
-* `gpt_4_1`
-* `gpt_4_1_mini`
-* `gpt_4_1_nano`
-* `o4_mini`
-* `o4_mini_high`
-* `o3`
-
-#### **Anthropic**
-
-* `claude_3_7_sonnet`
-* `claude_3_7_sonnet_thinking`
-* `claude_3_5_haiku`
-
-#### **Google**
-
-* `gemini_2_0_flash`
-* `gemini_2_0_flash_free`
-* `gemini_2_5_flash`
-* `gemini_2_5_flash_thinking`
-* `gemini_2_5_pro`
-
-#### **Deepseek**
-
-* `deepseek_v3_free`
-* `deepseek_v3`
-* `deepseek_r1_free`
-* `deepseek_r1`
-
-#### **xAI**
-
-* `grok_3_mini`
-* `grok_3`
-
-#### **Microsoft**
-
-* `mai_ds_r1_free`
-
-#### **Others**
-
-* `llama_4_maverick_free`
-* `llama_4_scout`
-* `mistral_small_3_1_24B_free`
-
-All of them are instances of `LLMModel`, which includes cost and model name settings.
-
-### Using Custom Models
-
-You can define and use your own custom model if it's available on OpenRouter.
-
-```python
-from OpenRouterProvider.Chatbot_manager import Chat_message, Chatbot_manager
-from OpenRouterProvider.LLMs import LLMModel
-
-# Declare a custom model
-my_model = LLMModel(
-    name="my-org/my-custom-model",  # Model name for OpenRouter
-    input_cost=0.5,                 # Optional: cost per 1M input tokens
-    output_cost=2.0                 # Optional: cost per 1M output tokens
-)
-
-# Use the custom model
-ai = Chatbot_manager(system_prompt="Please answer in English.")
-query = Chat_message(text="Explain black holes simply.")
-response = ai.invoke(model=my_model, query=query)
-print(response.text)
-```
-
-You only need to know the model name as used on OpenRouter. `input_cost` and `output_cost` are optional and currently, they are not used in this library. Please wait the future update.