webscout 7.9__py3-none-any.whl → 8.1__py3-none-any.whl

webscout/Provider/OPENAI/utils.py

@@ -0,0 +1,211 @@
+from typing import List, Dict, Optional, Any, Union
+from dataclasses import dataclass, asdict, is_dataclass
+from enum import Enum
+
+# --- OpenAI Response Structure Mimics ---
+# Moved here for reusability across different OpenAI-compatible providers
+
+class ToolCallType(str, Enum):
+    """Type of tool call."""
+    FUNCTION = "function"
+
+@dataclass
+class BaseModel:
+    """Base class for all models."""
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert the model to a dictionary."""
+        def _convert(obj: Any) -> Any:
+            if is_dataclass(obj):
+                return {k: _convert(v) for k, v in asdict(obj).items() if v is not None}
+            elif isinstance(obj, list):
+                return [_convert(item) for item in obj]
+            elif isinstance(obj, dict):
+                return {k: _convert(v) for k, v in obj.items() if v is not None}
+            elif isinstance(obj, Enum):
+                return obj.value
+            return obj
+        return _convert(self)
+
+    def __getitem__(self, key):
+        """Support dictionary-style access."""
+        if hasattr(self, key):
+            return getattr(self, key)
+        raise KeyError(f"{key} not found in {self.__class__.__name__}")
+
+    def get(self, key, default=None):
+        """Dictionary-style get method with default value."""
+        try:
+            return self[key]
+        except KeyError:
+            return default
+
+    def __contains__(self, key):
+        """Support 'in' operator."""
+        return hasattr(self, key)
+
+@dataclass
+class FunctionCall(BaseModel):
+    """Function call specification."""
+    name: str
+    arguments: str
+
+@dataclass
+class ToolFunction(BaseModel):
+    """Function specification in a tool."""
+    name: str
+    arguments: str
+
+@dataclass
+class ToolCall(BaseModel):
+    """Tool call specification."""
+    id: str
+    type: str
+    function: ToolFunction
+
+@dataclass
+class CompletionUsage(BaseModel):
+    """Token usage information."""
+    prompt_tokens: int
+    completion_tokens: int
+    total_tokens: int
+    prompt_tokens_details: Optional[Dict[str, Any]] = None
+
+@dataclass
+class ChoiceDelta(BaseModel):
+    """Delta content in streaming response."""
+    content: Optional[str] = None
+    function_call: Optional[FunctionCall] = None
+    role: Optional[str] = None
+    tool_calls: Optional[List[ToolCall]] = None
+
+@dataclass
+class ChatCompletionMessage(BaseModel):
+    """Chat message in completion response."""
+    role: str
+    content: Optional[str] = None
+    function_call: Optional[FunctionCall] = None
+    tool_calls: Optional[List[ToolCall]] = None
+
+@dataclass
+class Choice(BaseModel):
+    """Choice in completion response."""
+    index: int
+    message: Optional[ChatCompletionMessage] = None
+    delta: Optional[ChoiceDelta] = None
+    finish_reason: Optional[str] = None
+    logprobs: Optional[Dict[str, Any]] = None
+
+@dataclass
+class ChatCompletion(BaseModel):
+    """Chat completion response."""
+    id: str
+    created: int
+    model: str
+    choices: List[Choice]
+    object: str = "chat.completion"
+    system_fingerprint: Optional[str] = None
+    usage: Optional[CompletionUsage] = None
+
+@dataclass
+class ChatCompletionChunk(BaseModel):
+    """Streaming chat completion response chunk."""
+    id: str
+    created: int
+    model: str
+    choices: List[Choice]
+    object: str = "chat.completion.chunk"
+    system_fingerprint: Optional[str] = None
+
+
+# --- Helper Functions ---
+
+def format_prompt(messages: List[Dict[str, Any]], add_special_tokens: bool = False,
+                do_continue: bool = False, include_system: bool = True) -> str:
+    """
+    Format a series of messages into a single string, optionally adding special tokens.
+
+    Args:
+        messages: A list of message dictionaries, each containing 'role' and 'content'.
+        add_special_tokens: Whether to add special formatting tokens.
+        do_continue: If True, don't add the final "Assistant:" prompt.
+        include_system: Whether to include system messages in the formatted output.
+
+    Returns:
+        A formatted string containing all messages.
+    """
+    # Helper function to convert content to string
+    def to_string(value) -> str:
+        if isinstance(value, str):
+            return value
+        elif isinstance(value, dict):
+            if "text" in value:
+                return value.get("text", "")
+            return ""
+        elif isinstance(value, list):
+            return "".join([to_string(v) for v in value])
+        return str(value)
+
+    # If there's only one message and no special tokens needed, just return its content
+    if not add_special_tokens and len(messages) <= 1:
+        return to_string(messages[0]["content"])
+
+    # Filter and process messages
+    processed_messages = [
+        (message["role"], to_string(message["content"]))
+        for message in messages
+        if include_system or message.get("role") != "system"
+    ]
+
+    # Format each message as "Role: Content"
+    formatted = "\n".join([
+        f'{role.capitalize()}: {content}'
+        for role, content in processed_messages
+        if content.strip()
+    ])
+
+    # Add final prompt for assistant if needed
+    if do_continue:
+        return formatted
+
+    return f"{formatted}\nAssistant:"
+
+
+def get_system_prompt(messages: List[Dict[str, Any]]) -> str:
+    """
+    Extract and concatenate all system messages.
+
+    Args:
+        messages: A list of message dictionaries.
+
+    Returns:
+        A string containing all system messages concatenated with newlines.
+    """
+    return "\n".join([m["content"] for m in messages if m["role"] == "system"])
+
+
+def get_last_user_message(messages: List[Dict[str, Any]]) -> str:
+    """
+    Get the content of the last user message in the conversation.
+
+    Args:
+        messages: A list of message dictionaries.
+
+    Returns:
+        The content of the last user message as a string.
+    """
+    for message in reversed(messages):
+        if message["role"] == "user":
+            if isinstance(message["content"], str):
+                return message["content"]
+            # Handle complex content structures
+            if isinstance(message["content"], dict) and "text" in message["content"]:
+                return message["content"]["text"]
+            if isinstance(message["content"], list):
+                text_parts = []
+                for part in message["content"]:
+                    if isinstance(part, dict) and part.get("type") == "text":
+                        text_parts.append(part.get("text", ""))
+                    elif isinstance(part, str):
+                        text_parts.append(part)
+                return "".join(text_parts)
+    return ""

webscout/Provider/OPENAI/venice.py

@@ -0,0 +1,428 @@
+import time
+import uuid
+import requests
+import json
+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
+)
+
+# 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 "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36"
+
+# --- Venice Client ---
+
+class Completions(BaseCompletions):
+    def __init__(self, client: 'Venice'):
+        self._client = client
+
+    def create(
+        self,
+        *,
+        model: str,
+        messages: List[Dict[str, str]],
+        max_tokens: Optional[int] = 2049,
+        stream: bool = False,
+        temperature: Optional[float] = 0.8,
+        top_p: Optional[float] = 0.9,
+        **kwargs: Any
+    ) -> Union[ChatCompletion, Generator[ChatCompletionChunk, None, None]]:
+        """
+        Creates a model response for the given chat conversation.
+        Mimics openai.chat.completions.create
+        """
+        # Extract system message if present for systemPrompt parameter
+        system_prompt = self._client.system_prompt
+        for msg in messages:
+            if msg["role"] == "system":
+                system_prompt = msg["content"]
+                break
+
+        # Prepare the payload for Venice API
+        payload = {
+            "requestId": str(uuid.uuid4())[:7],
+            "modelId": self._client.convert_model_name(model),
+            "prompt": messages,
+            "systemPrompt": system_prompt,
+            "conversationType": "text",
+            "temperature": temperature if temperature is not None else self._client.temperature,
+            "webEnabled": True,
+            "topP": top_p if top_p is not None else self._client.top_p,
+            "includeVeniceSystemPrompt": False,
+            "isCharacter": False,
+            "clientProcessingTime": 2000
+        }
+
+        # Add optional parameters if provided
+        if max_tokens is not None and max_tokens > 0:
+            payload["max_tokens"] = max_tokens
+
+        # Add any additional parameters
+        for key, value in kwargs.items():
+            if key not in payload:
+                payload[key] = value
+
+        request_id = f"chatcmpl-{uuid.uuid4()}"
+        created_time = int(time.time())
+
+        if stream:
+            return self._create_stream(request_id, created_time, model, payload)
+        else:
+            return self._create_non_stream(request_id, created_time, model, payload)
+
+    def _create_stream(
+        self, request_id: str, created_time: int, model: str, payload: Dict[str, Any]
+    ) -> Generator[ChatCompletionChunk, None, None]:
+        try:
+            response = self._client.session.post(
+                self._client.api_endpoint,
+                json=payload,
+                stream=True,
+                timeout=self._client.timeout
+            )
+
+            # Handle non-200 responses
+            if response.status_code != 200:
+                raise IOError(
+                    f"Failed to generate response - ({response.status_code}, {response.reason}) - {response.text}"
+                )
+
+            # Track token usage across chunks
+            prompt_tokens = 0
+            completion_tokens = 0
+            total_tokens = 0
+
+            # Estimate prompt tokens based on message length
+            prompt_tokens = 0
+            for msg in payload.get("prompt", []):
+                prompt_tokens += len(msg.get("content", "").split())
+            prompt_tokens += len(payload.get("systemPrompt", "").split())
+
+            for line in response.iter_lines():
+                if not line:
+                    continue
+
+                try:
+                    # Decode bytes to string
+                    line_data = line.decode('utf-8').strip()
+                    if '"kind":"content"' in line_data:
+                        data = json.loads(line_data)
+                        if 'content' in data:
+                            content = data['content']
+
+                            # Format the content (replace escaped newlines)
+                            content = self._client.format_text(content)
+
+                            # Update token counts
+                            completion_tokens += 1
+                            total_tokens = prompt_tokens + completion_tokens
+
+                            # Create the delta object
+                            delta = ChoiceDelta(
+                                content=content,
+                                role="assistant",
+                                tool_calls=None
+                            )
+
+                            # Create the choice object
+                            choice = Choice(
+                                index=0,
+                                delta=delta,
+                                finish_reason=None,
+                                logprobs=None
+                            )
+
+                            # Create the chunk object
+                            chunk = ChatCompletionChunk(
+                                id=request_id,
+                                choices=[choice],
+                                created=created_time,
+                                model=model,
+                                system_fingerprint=None
+                            )
+
+                            # Convert to dict for proper formatting
+                            chunk_dict = chunk.to_dict()
+
+                            # Add usage information to match OpenAI format
+                            usage_dict = {
+                                "prompt_tokens": prompt_tokens,
+                                "completion_tokens": completion_tokens,
+                                "total_tokens": total_tokens,
+                                "estimated_cost": None
+                            }
+
+                            chunk_dict["usage"] = usage_dict
+
+                            # Return the chunk object for internal processing
+                            yield chunk
+                except json.JSONDecodeError:
+                    continue
+                except UnicodeDecodeError:
+                    continue
+
+            # Final chunk with finish_reason="stop"
+            delta = ChoiceDelta(
+                content=None,
+                role=None,
+                tool_calls=None
+            )
+
+            choice = Choice(
+                index=0,
+                delta=delta,
+                finish_reason="stop",
+                logprobs=None
+            )
+
+            chunk = ChatCompletionChunk(
+                id=request_id,
+                choices=[choice],
+                created=created_time,
+                model=model,
+                system_fingerprint=None
+            )
+
+            chunk_dict = chunk.to_dict()
+            chunk_dict["usage"] = {
+                "prompt_tokens": prompt_tokens,
+                "completion_tokens": completion_tokens,
+                "total_tokens": total_tokens,
+                "estimated_cost": None
+            }
+
+            yield chunk
+
+        except Exception as e:
+            print(f"Error during Venice stream request: {e}")
+            raise IOError(f"Venice request failed: {e}") from e
+
+    def _create_non_stream(
+        self, request_id: str, created_time: int, model: str, payload: Dict[str, Any]
+    ) -> ChatCompletion:
+        try:
+            # For non-streaming, we still use streaming internally to collect the full response
+            response = self._client.session.post(
+                self._client.api_endpoint,
+                json=payload,
+                stream=True,
+                timeout=self._client.timeout
+            )
+
+            # Handle non-200 responses
+            if response.status_code != 200:
+                raise IOError(
+                    f"Failed to generate response - ({response.status_code}, {response.reason}) - {response.text}"
+                )
+
+            # Collect the full response
+            full_text = ""
+            for line in response.iter_lines():
+                if not line:
+                    continue
+
+                try:
+                    # Decode bytes to string
+                    line_data = line.decode('utf-8').strip()
+                    if '"kind":"content"' in line_data:
+                        data = json.loads(line_data)
+                        if 'content' in data:
+                            content = data['content']
+                            full_text += content
+                except json.JSONDecodeError:
+                    continue
+                except UnicodeDecodeError:
+                    continue
+
+            # Format the text (replace escaped newlines)
+            full_text = self._client.format_text(full_text)
+
+            # Estimate token counts
+            prompt_tokens = 0
+            for msg in payload.get("prompt", []):
+                prompt_tokens += len(msg.get("content", "").split())
+            prompt_tokens += len(payload.get("systemPrompt", "").split())
+            completion_tokens = len(full_text.split())
+            total_tokens = prompt_tokens + completion_tokens
+
+            # Create the message object
+            message = ChatCompletionMessage(
+                role="assistant",
+                content=full_text
+            )
+
+            # Create the choice object
+            choice = Choice(
+                index=0,
+                message=message,
+                finish_reason="stop"
+            )
+
+            # Create the usage object
+            usage = CompletionUsage(
+                prompt_tokens=prompt_tokens,
+                completion_tokens=completion_tokens,
+                total_tokens=total_tokens
+            )
+
+            # Create the completion object
+            completion = ChatCompletion(
+                id=request_id,
+                choices=[choice],
+                created=created_time,
+                model=model,
+                usage=usage,
+            )
+
+            return completion
+
+        except Exception as e:
+            print(f"Error during Venice non-stream request: {e}")
+            raise IOError(f"Venice request failed: {e}") from e
+
+class Chat(BaseChat):
+    def __init__(self, client: 'Venice'):
+        self.completions = Completions(client)
+
+class Venice(OpenAICompatibleProvider):
+    """
+    OpenAI-compatible client for Venice AI API.
+
+    Usage:
+        client = Venice()
+        response = client.chat.completions.create(
+            model="mistral-31-24b",
+            messages=[{"role": "user", "content": "Hello!"}]
+        )
+    """
+
+    AVAILABLE_MODELS = [
+        "mistral-31-24b",
+        "llama-3.2-3b-akash",
+        "qwen2dot5-coder-32b",
+        "deepseek-coder-v2-lite",
+    ]
+
+    # No model mapping needed as we use the model names directly
+
+    def __init__(
+        self,
+        timeout: Optional[int] = None,
+        browser: str = "chrome"
+    ):
+        """
+        Initialize the Venice client.
+
+        Args:
+            timeout: Request timeout in seconds (None for no timeout)
+            browser: Browser to emulate in user agent
+        """
+        self.timeout = timeout
+        self.temperature = 0.8  # Default temperature
+        self.top_p = 0.9  # Default top_p
+        self.system_prompt = "You are a helpful AI assistant."  # Default system prompt
+        self.api_endpoint = "https://venice.ai/api/inference/chat"
+        self.session = requests.Session()
+
+        # Initialize LitAgent for user agent generation
+        agent = LitAgent()
+        self.fingerprint = agent.generate_fingerprint(browser)
+
+        # Headers for the request
+        self.headers = {
+            "User-Agent": self.fingerprint["user_agent"],
+            "accept": self.fingerprint["accept"],
+            "accept-language": self.fingerprint["accept_language"],
+            "content-type": "application/json",
+            "origin": "https://venice.ai",
+            "referer": "https://venice.ai/chat/",
+            "sec-ch-ua": self.fingerprint["sec_ch_ua"] or '"Google Chrome";v="133", "Chromium";v="133", "Not?A_Brand";v="24"',
+            "sec-ch-ua-mobile": "?0",
+            "sec-ch-ua-platform": f'"{self.fingerprint["platform"]}"',
+            "sec-fetch-dest": "empty",
+            "sec-fetch-mode": "cors",
+            "sec-fetch-site": "same-origin"
+        }
+
+        self.session.headers.update(self.headers)
+
+        # Initialize the chat interface
+        self.chat = Chat(self)
+
+    def format_text(self, text: str) -> str:
+        """
+        Format text by replacing escaped newlines with actual newlines.
+
+        Args:
+            text: Text to format
+
+        Returns:
+            Formatted text
+        """
+        # Use a more comprehensive approach to handle all escape sequences
+        try:
+            # First handle double backslashes to avoid issues
+            text = text.replace('\\\\', '\\')
+
+            # Handle common escape sequences
+            text = text.replace('\\n', '\n')
+            text = text.replace('\\r', '\r')
+            text = text.replace('\\t', '\t')
+            text = text.replace('\\"', '"')
+            text = text.replace("\\'", "'")
+
+            # Handle any remaining escape sequences using JSON decoding
+            # This is a fallback in case there are other escape sequences
+            try:
+                # Add quotes to make it a valid JSON string
+                json_str = f'"{text}"'
+                # Use json module to decode all escape sequences
+                decoded = json.loads(json_str)
+                return decoded
+            except json.JSONDecodeError:
+                # If JSON decoding fails, return the text with the replacements we've already done
+                return text
+        except Exception as e:
+            # If any error occurs, return the original text
+            print(f"Warning: Error formatting text: {e}")
+            return text
+
+    def convert_model_name(self, model: str) -> str:
+        """
+        Convert model names to ones supported by Venice.
+
+        Args:
+            model: Model name to convert
+
+        Returns:
+            Venice model name
+        """
+        # If the model is already a valid Venice model, return it
+        if model in self.AVAILABLE_MODELS:
+            return model
+
+        # Default to the most capable model
+        print(f"Warning: Unknown model '{model}'. Using 'mistral-31-24b' instead.")
+        return "mistral-31-24b"