webscout 8.2.7__py3-none-any.whl → 8.2.8__py3-none-any.whl

webscout/Provider/OPENAI/base.py

@@ -0,0 +1,249 @@
+from abc import ABC, abstractmethod
+from typing import List, Dict, Optional, Union, Generator, Any, TypedDict, Callable
+import json
+import logging
+from dataclasses import dataclass
+
+logger = logging.getLogger(__name__)
+
+
+# Import the utils for response structures
+from webscout.Provider.OPENAI.utils import ChatCompletion, ChatCompletionChunk, Choice, ChatCompletionMessage, ToolCall, ToolFunction
+
+# Define tool-related structures
+class ToolDefinition(TypedDict):
+    """Definition of a tool that can be called by the AI"""
+    type: str
+    function: Dict[str, Any]
+
+class FunctionParameters(TypedDict):
+    """Parameters for a function"""
+    type: str
+    properties: Dict[str, Dict[str, Any]]
+    required: List[str]
+
+class FunctionDefinition(TypedDict):
+    """Definition of a function that can be called by the AI"""
+    name: str
+    description: str
+    parameters: FunctionParameters
+
+@dataclass
+class Tool:
+    """Tool class that can be passed to the provider"""
+    name: str
+    description: str
+    parameters: Dict[str, Dict[str, Any]]
+    required_params: List[str] = None
+    implementation: Optional[Callable] = None
+    
+    def to_dict(self) -> ToolDefinition:
+        """Convert to OpenAI tool definition format"""
+        function_def = {
+            "name": self.name,
+            "description": self.description,
+            "parameters": {
+                "type": "object",
+                "properties": self.parameters,
+                "required": self.required_params or list(self.parameters.keys())
+            }
+        }
+        
+        return {
+            "type": "function",
+            "function": function_def
+        }
+    
+    def execute(self, arguments: Dict[str, Any]) -> Any:
+        """Execute the tool with the given arguments"""
+        if not self.implementation:
+            return f"Tool '{self.name}' does not have an implementation."
+            
+        try:
+            return self.implementation(**arguments)
+        except Exception as e:
+            logger.error(f"Error executing tool '{self.name}': {str(e)}")
+            return f"Error executing tool '{self.name}': {str(e)}"
+
+class BaseCompletions(ABC):
+    @abstractmethod
+    def create(
+        self,
+        *,
+        model: str,
+        messages: List[Dict[str, Any]],  # Changed to Any to support complex message structures
+        max_tokens: Optional[int] = None,
+        stream: bool = False,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        tools: Optional[List[Union[Tool, Dict[str, Any]]]] = None,  # Support for tool definitions
+        tool_choice: Optional[Union[str, Dict[str, Any]]] = None,  # Support for tool_choice parameter
+        **kwargs: Any
+    ) -> Union[ChatCompletion, Generator[ChatCompletionChunk, None, None]]:
+        """
+        Abstract method to create chat completions with tool support.
+        
+        Args:
+            model: The model to use for completion
+            messages: List of message dictionaries
+            max_tokens: Maximum number of tokens to generate
+            stream: Whether to stream the response
+            temperature: Sampling temperature
+            top_p: Nucleus sampling parameter
+            tools: List of tool definitions available for the model to use
+            tool_choice: Control over which tool the model should use
+            **kwargs: Additional model-specific parameters
+            
+        Returns:
+            Either a completion object or a generator of completion chunks if streaming
+        """
+        raise NotImplementedError
+    
+    def format_tool_calls(self, tools: List[Union[Tool, Dict[str, Any]]]) -> List[Dict[str, Any]]:
+        """Convert tools to the format expected by the provider"""
+        formatted_tools = []
+        
+        for tool in tools:
+            if isinstance(tool, Tool):
+                formatted_tools.append(tool.to_dict())
+            elif isinstance(tool, dict):
+                # Assume already formatted correctly
+                formatted_tools.append(tool)
+            else:
+                logger.warning(f"Skipping invalid tool type: {type(tool)}")
+        
+        return formatted_tools
+    
+    def process_tool_calls(self, tool_calls: List[Dict[str, Any]], available_tools: Dict[str, Tool]) -> List[Dict[str, Any]]:
+        """
+        Process tool calls and execute the relevant tools.
+        
+        Args:
+            tool_calls: List of tool calls from the model
+            available_tools: Dictionary mapping tool names to their implementations
+            
+        Returns:
+            List of results from executing the tools
+        """
+        results = []
+        
+        for call in tool_calls:
+            try:
+                function_call = call.get("function", {})
+                tool_name = function_call.get("name")
+                arguments_str = function_call.get("arguments", "{}")
+                
+                # Parse arguments 
+                try:
+                    if isinstance(arguments_str, str):
+                        arguments = json.loads(arguments_str)
+                    else:
+                        arguments = arguments_str
+                except json.JSONDecodeError:
+                    results.append({
+                        "tool_call_id": call.get("id"),
+                        "result": f"Error: Could not parse arguments JSON: {arguments_str}"
+                    })
+                    continue
+                
+                # Execute the tool if available
+                if tool_name in available_tools:
+                    tool_result = available_tools[tool_name].execute(arguments)
+                    results.append({
+                        "tool_call_id": call.get("id"),
+                        "result": str(tool_result)
+                    })
+                else:
+                    results.append({
+                        "tool_call_id": call.get("id"),
+                        "result": f"Error: Tool '{tool_name}' not found."
+                    })
+            except Exception as e:
+                logger.error(f"Error processing tool call: {str(e)}")
+                results.append({
+                    "tool_call_id": call.get("id", "unknown"),
+                    "result": f"Error processing tool call: {str(e)}"
+                })
+        
+        return results
+
+
+class BaseChat(ABC):
+    completions: BaseCompletions
+
+
+class OpenAICompatibleProvider(ABC):
+    """
+    Abstract Base Class for providers mimicking the OpenAI Python client structure.
+    Requires a nested 'chat.completions' structure with tool support.
+    """
+    chat: BaseChat
+    available_tools: Dict[str, Tool] = {}  # Dictionary of available tools
+    supports_tools: bool = False  # Whether the provider supports tools
+    supports_tool_choice: bool = False  # Whether the provider supports tool_choice
+
+    @abstractmethod
+    def __init__(self, api_key: Optional[str] = None, tools: Optional[List[Tool]] = None, **kwargs: Any):
+        """
+        Initialize the provider, potentially with an API key and tools.
+        
+        Args:
+            api_key: Optional API key for the provider
+            tools: Optional list of tools to make available to the provider
+            **kwargs: Additional provider-specific parameters
+        """
+        self.available_tools = {}
+        if tools:
+            self.register_tools(tools)
+        raise NotImplementedError
+
+    @property
+    @abstractmethod
+    def models(self):
+        """
+        Property that returns an object with a .list() method returning available models.
+        Subclasses must implement this property.
+        """
+        pass
+    
+    def register_tools(self, tools: List[Tool]) -> None:
+        """
+        Register tools with the provider.
+        
+        Args:
+            tools: List of Tool objects to register
+        """
+        for tool in tools:
+            self.available_tools[tool.name] = tool
+            
+    def get_tool_by_name(self, name: str) -> Optional[Tool]:
+        """Get a tool by name"""
+        return self.available_tools.get(name)
+    
+    def format_tool_response(self, messages: List[Dict[str, Any]], tool_results: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
+        """
+        Format tool results as messages to be sent back to the provider.
+        
+        Args:
+            messages: The original messages
+            tool_results: Results from executing tools
+            
+        Returns:
+            Updated message list with tool results
+        """
+        updated_messages = messages.copy()
+        
+        # Find the assistant message with tool calls
+        for i, msg in enumerate(reversed(updated_messages)):
+            if msg.get("role") == "assistant" and "tool_calls" in msg:
+                # For each tool result, add a tool message
+                for result in tool_results:
+                    tool_message = {
+                        "role": "tool",
+                        "tool_call_id": result["tool_call_id"],
+                        "content": result["result"]
+                    }
+                    updated_messages.append(tool_message)
+                break
+        
+        return updated_messages

webscout/Provider/OPENAI/c4ai.py

@@ -0,0 +1,373 @@
+import time
+import uuid
+import requests
+import json
+import re
+from typing import List, Dict, Optional, Union, Generator, Any
+
+# Import base classes and utility structures
+from .base import OpenAICompatibleProvider, BaseChat, BaseCompletions
+from .utils import (
+    ChatCompletionChunk, ChatCompletion, Choice, ChoiceDelta,
+    ChatCompletionMessage, CompletionUsage,
+    get_system_prompt, get_last_user_message, format_prompt # Import format_prompt
+)
+
+# Attempt to import LitAgent, fallback if not available
+try:
+    from webscout.litagent import LitAgent
+except ImportError:
+    # Define a dummy LitAgent if webscout is not installed or accessible
+    class LitAgent:
+        def generate_fingerprint(self, browser: str = "chrome") -> Dict[str, Any]:
+            # Return minimal default headers if LitAgent is unavailable
+            print("Warning: LitAgent not found. Using default minimal headers.")
+            return {
+                "accept": "*/*",
+                "accept_language": "en-US,en;q=0.9",
+                "platform": "Windows",
+                "sec_ch_ua": '"Not/A)Brand";v="99", "Google Chrome";v="127", "Chromium";v="127"',
+                "user_agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36",
+                "browser_type": browser,
+            }
+        def random(self) -> str:
+             # Return a default user agent if LitAgent is unavailable
+             print("Warning: LitAgent not found. Using default user agent.")
+             return "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36"
+
+
+class Completions(BaseCompletions):
+    def __init__(self, client: 'C4AI'):
+        self._client = client
+
+    def create(
+        self,
+        *,
+        model: str,
+        messages: List[Dict[str, str]],
+        max_tokens: Optional[int] = 2000,
+        stream: bool = False,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        **kwargs: Any
+    ) -> Union[ChatCompletion, Generator[ChatCompletionChunk, None, None]]:
+        """
+        Creates a model response for the given chat conversation.
+        Mimics openai.chat.completions.create
+        """
+        # Extract system prompt using utility function
+        system_prompt = get_system_prompt(messages)
+        if not system_prompt:
+            system_prompt = "You are a helpful assistant."
+
+        # Format the conversation history using format_prompt
+        # Note: C4AI API might expect only the *last* user message here.
+        # Sending the full history might cause issues.
+        # We exclude the system prompt from format_prompt as it's sent separately.
+        # We also set do_continue=True as C4AI adds its own assistant prompt implicitly.
+        conversation_prompt = format_prompt(messages, include_system=False, do_continue=True)
+
+        if not conversation_prompt:
+             # Fallback to last user message if formatted prompt is empty
+             last_user_message = get_last_user_message(messages)
+             if not last_user_message:
+                 raise ValueError("No user message found or formatted prompt is empty.")
+             conversation_prompt = last_user_message
+
+        # Create or get conversation ID
+        if model not in self._client._conversation_data:
+            conversation_id = self._client.create_conversation(model, system_prompt)
+            if not conversation_id:
+                raise IOError(f"Failed to create conversation with model {model}")
+        else:
+            conversation_id = self._client._conversation_data[model]["conversationId"]
+            self._client._conversation_data[model]["messageId"] = self._client.fetch_message_id(conversation_id)
+
+        request_id = f"chatcmpl-{uuid.uuid4()}"
+        created_time = int(time.time())
+
+        # Pass the formatted conversation prompt
+        if stream:
+            return self._create_stream(request_id, created_time, model, conversation_id, conversation_prompt, system_prompt)
+        else:
+            return self._create_non_stream(request_id, created_time, model, conversation_id, conversation_prompt, system_prompt)
+
+    def _create_stream(
+        self, request_id: str, created_time: int, model: str, conversation_id: str, prompt: str, system_prompt: str
+    ) -> Generator[ChatCompletionChunk, None, None]:
+        try:
+            message_id = self._client._conversation_data[model]["messageId"]
+            url = f"{self._client.url}/api/chat/message"
+            payload = {
+                "conversationId": conversation_id,
+                "messageId": message_id,
+                "model": model,
+                "prompt": prompt, # Use the formatted conversation history as prompt
+                "preprompt": system_prompt,
+                "temperature": 0.7,
+                "top_p": 1,
+                "top_k": 50,
+                "max_tokens": self._client.max_tokens_to_sample,
+                "stop": [],
+                "stream": True
+            }
+
+            response = self._client.session.post(
+                url,
+                headers=self._client.headers,
+                json=payload,
+                stream=True,
+                timeout=self._client.timeout
+            )
+            response.raise_for_status()
+
+            full_text = ""
+            for line in response.iter_lines():
+                if line:
+                    line = line.decode('utf-8')
+                    if line.startswith('data: '):
+                        data = line[6:]
+                        if data == "[DONE]":
+                            break
+
+                        try:
+                            json_data = json.loads(data)
+                            delta_text = json_data.get('text', '')
+                            new_content = delta_text[len(full_text):]
+                            full_text = delta_text
+                            delta = ChoiceDelta(content=new_content)
+                            choice = Choice(index=0, delta=delta, finish_reason=None)
+                            chunk = ChatCompletionChunk(
+                                id=request_id,
+                                choices=[choice],
+                                created=created_time,
+                                model=model
+                            )
+                            yield chunk
+                        except json.JSONDecodeError:
+                            continue
+
+            delta = ChoiceDelta(content=None)
+            choice = Choice(index=0, delta=delta, finish_reason="stop")
+            chunk = ChatCompletionChunk(
+                id=request_id,
+                choices=[choice],
+                created=created_time,
+                model=model
+            )
+            yield chunk
+
+        except Exception as e:
+            print(f"Error during C4AI stream request: {e}")
+            raise IOError(f"C4AI request failed: {e}") from e
+
+    def _create_non_stream(
+        self, request_id: str, created_time: int, model: str, conversation_id: str, prompt: str, system_prompt: str
+    ) -> ChatCompletion:
+        try:
+            message_id = self._client._conversation_data[model]["messageId"]
+            url = f"{self._client.url}/api/chat/message"
+            payload = {
+                "conversationId": conversation_id,
+                "messageId": message_id,
+                "model": model,
+                "prompt": prompt, # Use the formatted conversation history as prompt
+                "preprompt": system_prompt,
+                "temperature": 0.7,
+                "top_p": 1,
+                "top_k": 50,
+                "max_tokens": self._client.max_tokens_to_sample,
+                "stop": [],
+                "stream": False
+            }
+
+            response = self._client.session.post(
+                url,
+                headers=self._client.headers,
+                json=payload,
+                timeout=self._client.timeout
+            )
+            response.raise_for_status()
+
+            data = response.json()
+            response_text = data.get('text', '')
+            message = ChatCompletionMessage(role="assistant", content=response_text)
+            choice = Choice(index=0, message=message, finish_reason="stop")
+            # Estimate tokens based on the formatted prompt
+            prompt_tokens = len(prompt) // 4
+            completion_tokens = len(response_text) // 4
+            usage = CompletionUsage(
+                prompt_tokens=prompt_tokens,
+                completion_tokens=completion_tokens,
+                total_tokens=prompt_tokens + completion_tokens
+            )
+            completion = ChatCompletion(
+                id=request_id,
+                choices=[choice],
+                created=created_time,
+                model=model,
+                usage=usage
+            )
+            return completion
+
+        except Exception as e:
+            print(f"Error during C4AI non-stream request: {e}")
+            raise IOError(f"C4AI request failed: {e}") from e
+
+class Chat(BaseChat):
+    def __init__(self, client: 'C4AI'):
+        self.completions = Completions(client)
+
+class C4AI(OpenAICompatibleProvider):
+    """
+    OpenAI-compatible client for C4AI API.
+
+    Usage:
+        client = C4AI()
+        response = client.chat.completions.create(
+            model="command-a-03-2025",
+            messages=[{"role": "user", "content": "Hello!"}]
+        )
+    """
+
+    AVAILABLE_MODELS = [
+        'command-a-03-2025',
+        'command-r-plus-08-2024',
+        'command-r-08-2024',
+        'command-r-plus',
+        'command-r',
+        'command-r7b-12-2024',
+        'command-r7b-arabic-02-2025'
+    ]
+
+    def __init__(
+        self,
+        timeout: Optional[int] = None,
+        browser: str = "chrome"
+    ):
+        """
+        Initialize the C4AI client.
+
+        Args:
+            timeout: Request timeout in seconds.
+            browser: Browser name for LitAgent to generate User-Agent.
+        """
+        self.timeout = timeout
+        self.url = "https://cohereforai-c4ai-command.hf.space"
+        self.session = requests.Session()
+        self.max_tokens_to_sample = 2000
+
+        agent = LitAgent()
+        fingerprint = agent.generate_fingerprint(browser)
+
+        self.headers = {
+            "Content-Type": "application/json",
+            "User-Agent": fingerprint["user_agent"],
+            "Accept": "*/*",
+            "Accept-Encoding": "gzip, deflate, br, zstd",
+            "Accept-Language": fingerprint["accept_language"],
+            "Origin": "https://cohereforai-c4ai-command.hf.space",
+            "Referer": "https://cohereforai-c4ai-command.hf.space/",
+            "Sec-Ch-Ua": fingerprint["sec_ch_ua"] or "\"Chromium\";v=\"120\"",
+            "Sec-Ch-Ua-Mobile": "?0",
+            "Sec-Ch-Ua-Platform": f"\"{fingerprint['platform']}\"",
+            "Sec-Fetch-Dest": "empty",
+            "Sec-Fetch-Mode": "cors",
+            "Sec-Fetch-Site": "same-origin",
+            "DNT": "1",
+            "Priority": "u=1, i"
+        }
+
+        self._conversation_data = {}
+        self.chat = Chat(self)
+        self.update_available_models()
+
+    def update_available_models(self):
+        """Update the list of available models from the server."""
+        try:
+            response = requests.get("https://cohereforai-c4ai-command.hf.space/")
+            text = response.text
+            models_match = re.search(r'models:(\[.+?\]),oldModels:', text)
+
+            if not models_match:
+                return
+
+            models_text = models_match.group(1)
+            models_text = re.sub(r',parameters:{[^}]+?}', '', models_text)
+            models_text = models_text.replace('void 0', 'null')
+
+            def add_quotation_mark(match):
+                return f'{match.group(1)}"{match.group(2)}":'
+
+            models_text = re.sub(r'([{,])([A-Za-z0-9_]+?):', add_quotation_mark, models_text)
+
+            models_data = json.loads(models_text)
+            self.AVAILABLE_MODELS = [model["id"] for model in models_data]
+        except Exception:
+            pass
+
+    def create_conversation(self, model: str, system_prompt: str):
+        """Create a new conversation with the specified model."""
+        url = f"{self.url}/api/conversation"
+        payload = {
+            "model": model,
+            "preprompt": system_prompt,
+        }
+
+        try:
+            response = self.session.post(
+                url,
+                headers=self.headers,
+                json=payload,
+                timeout=self.timeout
+            )
+            response.raise_for_status()
+
+            data = response.json()
+            conversation_id = data.get("conversationId")
+
+            if conversation_id:
+                self._conversation_data[model] = {
+                    "conversationId": conversation_id,
+                    "messageId": self.fetch_message_id(conversation_id)
+                }
+                return conversation_id
+
+            return None
+
+        except Exception as e:
+            print(f"Error creating conversation: {e}")
+            return None
+
+    def fetch_message_id(self, conversation_id: str):
+        """Fetch the latest message ID for a conversation."""
+        url = f"{self.url}/api/conversation/{conversation_id}"
+
+        try:
+            response = self.session.get(
+                url,
+                headers=self.headers,
+                timeout=self.timeout
+            )
+            response.raise_for_status()
+
+            json_data = response.json()
+
+            if json_data.get("nodes", []) and json_data["nodes"][-1].get("type") == "error":
+                return str(uuid.uuid4())
+
+            data = json_data["nodes"][1]["data"]
+            keys = data[data[0]["messages"]]
+            message_keys = data[keys[-1]]
+            message_id = data[message_keys["id"]]
+
+            return message_id
+
+        except Exception:
+            return str(uuid.uuid4())
+    @property
+    def models(self):
+        class _ModelList:
+            def list(inner_self):
+                return type(self).AVAILABLE_MODELS
+        return _ModelList()