webscout 8.0__py3-none-any.whl → 8.2__py3-none-any.whl

webscout/Provider/OPENAI/uncovrAI.py

@@ -0,0 +1,455 @@
+import time
+import uuid
+import re
+import json
+import cloudscraper
+from typing import List, Dict, Optional, Union, Generator, Any
+
+from webscout.litagent import LitAgent
+from .base import BaseChat, BaseCompletions, OpenAICompatibleProvider
+from .utils import (
+    ChatCompletion,
+    ChatCompletionChunk,
+    Choice,
+    ChatCompletionMessage,
+    ChoiceDelta,
+    CompletionUsage,
+    format_prompt,
+    get_system_prompt,
+    get_last_user_message
+)
+
+# ANSI escape codes for formatting
+BOLD = "\033[1m"
+RED = "\033[91m"
+RESET = "\033[0m"
+
+class Completions(BaseCompletions):
+    def __init__(self, client: 'UncovrAI'):
+        self._client = client
+
+    def create(
+        self,
+        *,
+        model: str,
+        messages: List[Dict[str, str]],
+        max_tokens: Optional[int] = None,
+        stream: bool = False,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        **kwargs: Any
+    ) -> Union[ChatCompletion, Generator[ChatCompletionChunk, None, None]]:
+        """
+        Create a chat completion using the UncovrAI API.
+
+        Args:
+            model: The model to use for completion
+            messages: A list of messages in the conversation
+            max_tokens: Maximum number of tokens to generate
+            stream: Whether to stream the response
+            temperature: Controls randomness (mapped to UncovrAI's temperature)
+            top_p: Controls diversity (not directly used by UncovrAI)
+            **kwargs: Additional parameters
+
+        Returns:
+            A ChatCompletion object or a generator of ChatCompletionChunk objects
+        """
+        # Validate model
+        if model not in self._client.AVAILABLE_MODELS:
+            raise ValueError(f"Invalid model: {model}. Choose from: {self._client.AVAILABLE_MODELS}")
+
+        # Map temperature to UncovrAI's scale (0-100)
+        # Default to 32 (medium) if not provided
+        uncovr_temperature = 32
+        if temperature is not None:
+            # Map from 0-1 scale to 0-100 scale
+            uncovr_temperature = int(temperature * 100)
+            # Ensure it's within bounds
+            uncovr_temperature = max(0, min(100, uncovr_temperature))
+
+        # Map creativity from kwargs or use default
+        creativity = kwargs.get("creativity", "medium")
+
+        # Get focus and tools from kwargs or use defaults
+        selected_focus = kwargs.get("selected_focus", ["web"])
+        selected_tools = kwargs.get("selected_tools", ["quick-cards"])
+
+        # Generate request ID and timestamp
+        request_id = str(uuid.uuid4())
+        created_time = int(time.time())
+
+        # Format the conversation using utility functions
+        conversation_prompt = format_prompt(messages, add_special_tokens=False, do_continue=True)
+
+        # Prepare the request payload
+        payload = {
+            "content": conversation_prompt,
+            "chatId": self._client.chat_id,
+            "userMessageId": str(uuid.uuid4()),
+            "ai_config": {
+                "selectedFocus": selected_focus,
+                "selectedTools": selected_tools,
+                "agentId": "chat",
+                "modelId": model,
+                "temperature": uncovr_temperature,
+                "creativity": creativity
+            }
+        }
+
+        # Handle streaming response
+        if stream:
+            return self._handle_streaming_response(
+                payload=payload,
+                model=model,
+                request_id=request_id,
+                created_time=created_time
+            )
+
+        # Handle non-streaming response
+        return self._handle_non_streaming_response(
+            payload=payload,
+            model=model,
+            request_id=request_id,
+            created_time=created_time
+        )
+
+    def _handle_streaming_response(
+        self,
+        *,
+        payload: Dict[str, Any],
+        model: str,
+        request_id: str,
+        created_time: int
+    ) -> Generator[ChatCompletionChunk, None, None]:
+        """Handle streaming response from UncovrAI API."""
+        try:
+            with self._client.session.post(
+                self._client.url,
+                json=payload,
+                stream=True,
+                timeout=self._client.timeout
+            ) as response:
+                if response.status_code != 200:
+                    # If we get a non-200 response, try refreshing our identity once
+                    if response.status_code in [403, 429]:
+                        self._client.refresh_identity()
+                        # Retry with new identity
+                        with self._client.session.post(
+                            self._client.url,
+                            json=payload,
+                            stream=True,
+                            timeout=self._client.timeout
+                        ) as retry_response:
+                            if not retry_response.ok:
+                                raise IOError(
+                                    f"Failed to generate response after identity refresh - "
+                                    f"({retry_response.status_code}, {retry_response.reason}) - "
+                                    f"{retry_response.text}"
+                                )
+                            response = retry_response
+                    else:
+                        raise IOError(f"Request failed with status code {response.status_code}")
+
+                # Process the streaming response
+                streaming_text = ""
+                for line in response.iter_lines():
+                    if line:
+                        try:
+                            line = line.decode('utf-8')
+
+                            # Use regex to match content messages
+                            content_match = re.match(r'^0:\s*"?(.*?)"?$', line)
+                            if content_match:  # Content message
+                                content = content_match.group(1)
+                                # Format the content to handle escape sequences
+                                content = self._client.format_text(content)
+                                streaming_text += content
+
+                                # Create a chunk for this part of the response
+                                delta = ChoiceDelta(content=content)
+                                choice = Choice(
+                                    index=0,
+                                    delta=delta,
+                                    finish_reason=None
+                                )
+                                chunk = ChatCompletionChunk(
+                                    id=request_id,
+                                    choices=[choice],
+                                    created=created_time,
+                                    model=model
+                                )
+
+                                yield chunk
+
+                            # Check for error messages
+                            error_match = re.match(r'^2:\[{"type":"error","error":"(.*?)"}]$', line)
+                            if error_match:
+                                error_msg = error_match.group(1)
+                                raise IOError(f"API Error: {error_msg}")
+
+                        except (json.JSONDecodeError, UnicodeDecodeError):
+                            continue
+
+                # Yield a final chunk with finish_reason="stop"
+                delta = ChoiceDelta()
+                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"{RED}Error during UncovrAI streaming request: {e}{RESET}")
+            raise IOError(f"UncovrAI streaming request failed: {e}") from e
+
+    def _handle_non_streaming_response(
+        self,
+        *,
+        payload: Dict[str, Any],
+        model: str,
+        request_id: str,
+        created_time: int
+    ) -> ChatCompletion:
+        """Handle non-streaming response from UncovrAI API."""
+        try:
+            response = self._client.session.post(
+                self._client.url,
+                json=payload,
+                timeout=self._client.timeout
+            )
+
+            if response.status_code != 200:
+                if response.status_code in [403, 429]:
+                    self._client.refresh_identity()
+                    response = self._client.session.post(
+                        self._client.url,
+                        json=payload,
+                        timeout=self._client.timeout
+                    )
+                    if not response.ok:
+                        raise IOError(
+                            f"Failed to generate response after identity refresh - "
+                            f"({response.status_code}, {response.reason}) - "
+                            f"{response.text}"
+                        )
+                else:
+                    raise IOError(f"Request failed with status code {response.status_code}")
+
+            full_response = ""
+            for line in response.iter_lines():
+                if line:
+                    try:
+                        line = line.decode('utf-8')
+                        content_match = re.match(r'^0:\s*"?(.*?)"?$', line)
+                        if content_match:
+                            content = content_match.group(1)
+                            full_response += content
+
+                        # Check for error messages
+                        error_match = re.match(r'^2:\[{"type":"error","error":"(.*?)"}]$', line)
+                        if error_match:
+                            error_msg = error_match.group(1)
+                            raise IOError(f"API Error: {error_msg}")
+
+                    except (json.JSONDecodeError, UnicodeDecodeError):
+                        continue
+
+            # Format the full response to handle escape sequences
+            full_response = self._client.format_text(full_response)
+
+            # Create message, choice, and usage objects
+            message = ChatCompletionMessage(
+                role="assistant",
+                content=full_response
+            )
+
+            choice = Choice(
+                index=0,
+                message=message,
+                finish_reason="stop"
+            )
+
+            # Estimate token usage (this is approximate)
+            prompt_tokens = len(payload["content"]) // 4
+            completion_tokens = len(full_response) // 4
+            total_tokens = prompt_tokens + completion_tokens
+
+            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"{RED}Error during UncovrAI non-stream request: {e}{RESET}")
+            raise IOError(f"UncovrAI request failed: {e}") from e
+
+class Chat(BaseChat):
+    def __init__(self, client: 'UncovrAI'):
+        self.completions = Completions(client)
+
+class UncovrAI(OpenAICompatibleProvider):
+    """
+    OpenAI-compatible client for Uncovr AI API.
+
+    Usage:
+        client = UncovrAI()
+        response = client.chat.completions.create(
+            model="default",
+            messages=[{"role": "user", "content": "Hello!"}]
+        )
+        print(response.choices[0].message.content)
+    """
+
+    AVAILABLE_MODELS = [
+        "default",
+        "gpt-4o-mini",
+        "gemini-2-flash",
+        "gemini-2-flash-lite",
+        "groq-llama-3-1-8b",
+        "o3-mini",
+        "deepseek-r1-distill-qwen-32b",
+        # The following models are not available in the free plan:
+        # "claude-3-7-sonnet",
+        # "gpt-4o",
+        # "claude-3-5-sonnet-v2",
+        # "deepseek-r1-distill-llama-70b",
+        # "gemini-2-flash-lite-preview",
+        # "qwen-qwq-32b"
+    ]
+
+    def __init__(
+        self,
+        timeout: int = 30,
+        browser: str = "chrome",
+        chat_id: Optional[str] = None,
+        user_id: Optional[str] = None,
+        proxies: dict = {}
+    ):
+        """
+        Initialize the UncovrAI client.
+
+        Args:
+            timeout: Request timeout in seconds
+            browser: Browser name for LitAgent to generate fingerprint
+            chat_id: Optional chat ID (will generate one if not provided)
+            user_id: Optional user ID (will generate one if not provided)
+            proxies: Optional proxy configuration
+        """
+        self.url = "https://uncovr.app/api/workflows/chat"
+        self.timeout = timeout
+
+        # Initialize LitAgent for user agent generation
+        self.agent = LitAgent()
+
+        # Use fingerprinting to create a consistent browser identity
+        self.fingerprint = self.agent.generate_fingerprint(browser)
+
+        # Use the fingerprint for headers
+        self.headers = {
+            "Accept": self.fingerprint["accept"],
+            "Accept-Encoding": "gzip, deflate, br, zstd",
+            "Accept-Language": self.fingerprint["accept_language"],
+            "Content-Type": "application/json",
+            "Origin": "https://uncovr.app",
+            "Referer": "https://uncovr.app/",
+            "Sec-CH-UA": self.fingerprint["sec_ch_ua"] or '"Not)A;Brand";v="99", "Microsoft Edge";v="127", "Chromium";v="127"',
+            "Sec-CH-UA-Mobile": "?0",
+            "Sec-CH-UA-Platform": f'"{self.fingerprint["platform"]}"',
+            "User-Agent": self.fingerprint["user_agent"],
+            "Sec-Fetch-Dest": "empty",
+            "Sec-Fetch-Mode": "cors",
+            "Sec-Fetch-Site": "same-origin"
+        }
+
+        # Use cloudscraper to bypass Cloudflare protection
+        self.session = cloudscraper.create_scraper()
+        self.session.headers.update(self.headers)
+        self.session.proxies.update(proxies)
+
+        # Set chat and user IDs
+        self.chat_id = chat_id or str(uuid.uuid4())
+        self.user_id = user_id or f"user_{str(uuid.uuid4())[:8].upper()}"
+
+        # Initialize chat interface
+        self.chat = Chat(self)
+
+    def refresh_identity(self, browser: str = None):
+        """
+        Refreshes the browser identity fingerprint.
+
+        Args:
+            browser: Specific browser to use for the new fingerprint
+        """
+        browser = browser or self.fingerprint.get("browser_type", "chrome")
+        self.fingerprint = self.agent.generate_fingerprint(browser)
+
+        # Update headers with new fingerprint
+        self.headers.update({
+            "Accept": self.fingerprint["accept"],
+            "Accept-Language": self.fingerprint["accept_language"],
+            "Sec-CH-UA": self.fingerprint["sec_ch_ua"] or self.headers["Sec-CH-UA"],
+            "Sec-CH-UA-Platform": f'"{self.fingerprint["platform"]}"',
+            "User-Agent": self.fingerprint["user_agent"],
+        })
+
+        # Update session headers
+        for header, value in self.headers.items():
+            self.session.headers[header] = value
+
+        return self.fingerprint
+
+    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
+            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"{RED}Warning: Error formatting text: {e}{RESET}")
+            return text

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 ""