webscout 8.3__py3-none-any.whl → 8.3.2__py3-none-any.whl

webscout/Provider/OPENAI/api.py

@@ -1,1035 +0,0 @@
-"""
-Webscout OpenAI-Compatible API Server
-
-A FastAPI-based server that provides OpenAI-compatible endpoints for various LLM providers.
-Supports streaming and non-streaming chat completions with comprehensive error handling,
-authentication, and provider management.
-"""
-
-from __future__ import annotations
-
-import json
-import logging
-import os
-import secrets
-import sys
-import time
-import uuid
-import inspect
-import re
-import codecs
-from typing import List, Dict, Optional, Union, Any, Generator, Callable
-import types
-
-import uvicorn
-from fastapi import FastAPI, Response, Request, Body
-from fastapi.middleware.cors import CORSMiddleware
-from fastapi.responses import StreamingResponse, RedirectResponse, JSONResponse
-from fastapi.openapi.utils import get_openapi
-from fastapi.routing import APIRoute
-from fastapi.exceptions import RequestValidationError
-from fastapi.security import APIKeyHeader
-from starlette.exceptions import HTTPException as StarletteHTTPException
-
-def clean_text(text):
-    """Clean text by removing null bytes and control characters except newlines and tabs."""
-    if not isinstance(text, str):
-        return text
-    
-    # Remove null bytes
-    text = text.replace('\x00', '')
-    
-    # Keep newlines, tabs, and other printable characters, remove other control chars
-    # This regex matches control characters except \n, \r, \t
-    return re.sub(r'[\x01-\x08\x0b\x0c\x0e-\x1f\x7f]', '', text)
-from starlette.status import (
-    HTTP_422_UNPROCESSABLE_ENTITY,
-    HTTP_404_NOT_FOUND,
-    HTTP_401_UNAUTHORIZED,
-    HTTP_403_FORBIDDEN,
-    HTTP_500_INTERNAL_SERVER_ERROR,
-)
-
-from webscout.Provider.OPENAI.pydantic_imports import BaseModel, Field
-from typing import Literal
-
-# Import provider classes from the OPENAI directory
-from webscout.Provider.OPENAI import *
-from webscout.Provider.OPENAI.utils import (
-    ChatCompletion, Choice, ChatCompletionMessage, CompletionUsage
-)
-
-
-# Configuration constants
-DEFAULT_PORT = 8000
-DEFAULT_HOST = "0.0.0.0"
-API_VERSION = "v1"
-
-# Setup logging
-logging.basicConfig(
-    level=logging.INFO,
-    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
-    handlers=[
-        logging.StreamHandler(sys.stdout),
-    ]
-)
-logger = logging.getLogger("webscout.api")
-
-
-class ServerConfig:
-    """Centralized configuration management for the API server."""
-
-    def __init__(self):
-        self.api_key: Optional[str] = None
-        self.provider_map: Dict[str, Any] = {}
-        self.default_provider: str = "ChatGPT"
-        self.base_url: Optional[str] = None
-        self.host: str = DEFAULT_HOST
-        self.port: int = DEFAULT_PORT
-        self.debug: bool = False
-        self.cors_origins: List[str] = ["*"]
-        self.max_request_size: int = 10 * 1024 * 1024  # 10MB
-        self.request_timeout: int = 300  # 5 minutes
-
-    def update(self, **kwargs) -> None:
-        """Update configuration with provided values."""
-        for key, value in kwargs.items():
-            if hasattr(self, key) and value is not None:
-                setattr(self, key, value)
-                logger.info(f"Config updated: {key} = {value}")
-
-    def validate(self) -> None:
-        """Validate configuration settings."""
-        if self.port < 1 or self.port > 65535:
-            raise ValueError(f"Invalid port number: {self.port}")
-
-        if self.default_provider not in self.provider_map and self.provider_map:
-            available_providers = list(set(v.__name__ for v in self.provider_map.values()))
-            logger.warning(f"Default provider '{self.default_provider}' not found. Available: {available_providers}")
-
-
-# Global configuration instance
-config = ServerConfig()
-
-
-# Define Pydantic models for multimodal content parts, aligning with OpenAI's API
-class TextPart(BaseModel):
-    """Text content part for multimodal messages."""
-    type: Literal["text"]
-    text: str
-
-
-class ImageURL(BaseModel):
-    """Image URL configuration for multimodal messages."""
-    url: str  # Can be http(s) or data URI
-    detail: Optional[Literal["auto", "low", "high"]] = Field(
-        "auto",
-        description="Specifies the detail level of the image."
-    )
-
-
-class ImagePart(BaseModel):
-    """Image content part for multimodal messages."""
-    type: Literal["image_url"]
-    image_url: ImageURL
-
-
-MessageContentParts = Union[TextPart, ImagePart]
-
-
-class Message(BaseModel):
-    """Chat message model compatible with OpenAI API."""
-    role: Literal["system", "user", "assistant", "function", "tool"]
-    content: Optional[Union[str, List[MessageContentParts]]] = Field(
-        None,
-        description="The content of the message. Can be a string, a list of content parts (for multimodal), or null."
-    )
-    name: Optional[str] = None
-    # Future: Add tool_calls and tool_call_id for function calling support
-    # tool_calls: Optional[List[ToolCall]] = None
-    # tool_call_id: Optional[str] = None
-
-class ChatCompletionRequest(BaseModel):
-    model: str = Field(..., description="ID of the model to use. See the model endpoint for the available models.")
-    messages: List[Message] = Field(..., description="A list of messages comprising the conversation so far.")
-    temperature: Optional[float] = Field(None, description="What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.")
-    top_p: Optional[float] = Field(None, description="An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered.")
-    n: Optional[int] = Field(1, description="How many chat completion choices to generate for each input message.")
-    stream: Optional[bool] = Field(False, description="If set, partial message deltas will be sent, like in ChatGPT. Tokens will be sent as data-only server-sent events as they become available, with the stream terminated by a data: [DONE] message.")
-    max_tokens: Optional[int] = Field(None, description="The maximum number of tokens to generate in the chat completion.")
-    presence_penalty: Optional[float] = Field(None, description="Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.")
-    frequency_penalty: Optional[float] = Field(None, description="Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.")
-    logit_bias: Optional[Dict[str, float]] = Field(None, description="Modify the likelihood of specified tokens appearing in the completion.")
-    user: Optional[str] = Field(None, description="A unique identifier representing your end-user, which can help the API to monitor and detect abuse.")
-    stop: Optional[Union[str, List[str]]] = Field(None, description="Up to 4 sequences where the API will stop generating further tokens.")
-
-    class Config:
-        extra = "ignore"  # Ignore extra fields that aren't in the model
-        schema_extra = {
-            "example": {
-                "model": "ChatGPT/gpt-4o",
-                "messages": [
-                    {"role": "system", "content": "You are a helpful assistant."},
-                    {"role": "user", "content": "Hello, how are you?"}
-                ],
-                "temperature": 0.7,
-                "max_tokens": 150,
-                "stream": False
-            }
-        }
-
-class ModelInfo(BaseModel):
-    """Model information for the models endpoint."""
-    id: str
-    object: str = "model"
-    created: int
-    owned_by: str
-
-
-class ModelListResponse(BaseModel):
-    """Response model for the models list endpoint."""
-    object: str = "list"
-    data: List[ModelInfo]
-
-
-class ErrorDetail(BaseModel):
-    """Error detail structure compatible with OpenAI API."""
-    message: str
-    type: str = "server_error"
-    param: Optional[str] = None
-    code: Optional[str] = None
-
-
-class ErrorResponse(BaseModel):
-    """Error response structure compatible with OpenAI API."""
-    error: ErrorDetail
-
-
-class APIError(Exception):
-    """Custom exception for API errors."""
-
-    def __init__(self, message: str, status_code: int = HTTP_500_INTERNAL_SERVER_ERROR,
-                 error_type: str = "server_error", param: Optional[str] = None,
-                 code: Optional[str] = None):
-        self.message = message
-        self.status_code = status_code
-        self.error_type = error_type
-        self.param = param
-        self.code = code
-        super().__init__(message)
-
-    def to_response(self) -> JSONResponse:
-        """Convert to FastAPI JSONResponse."""
-        error_detail = ErrorDetail(
-            message=self.message,
-            type=self.error_type,
-            param=self.param,
-            code=self.code
-        )
-        error_response = ErrorResponse(error=error_detail)
-        return JSONResponse(
-            status_code=self.status_code,
-            content=error_response.model_dump(exclude_none=True)
-        )
-
-
-class AppConfig:
-    """Legacy configuration class for backward compatibility."""
-    api_key: Optional[str] = None
-    provider_map = {}
-    default_provider = "ChatGPT"
-    base_url: Optional[str] = None
-
-    @classmethod
-    def set_config(cls, **data):
-        """Set configuration values."""
-        for key, value in data.items():
-            setattr(cls, key, value)
-        # Sync with new config system
-        config.update(**data)
-
-# Custom route class to handle dynamic base URLs
-# Note: The /docs 404 issue is likely related to server execution (Werkzeug logs vs. Uvicorn script).
-# This DynamicBaseRoute, when AppConfig.base_url is None, should act as a passthrough and not break /docs.
-# If AppConfig.base_url is set, this route class has limitations in correctly handling prefixed routes
-# without more complex path manipulation or using FastAPI's APIRouter prefixing/mounting features.
-class DynamicBaseRoute(APIRoute):
-    def get_route_handler(self) -> Callable:
-        original_route_handler = super().get_route_handler()
-        async def custom_route_handler(request: Request) -> Response:
-            if AppConfig.base_url:
-                if not request.url.path.startswith(AppConfig.base_url):
-                    # This logic might need refinement if base_url is used.
-                    # For API routes not matching the prefix, a 404 might be appropriate.
-                    # Docs routes (/docs, /openapi.json) are usually at the root.
-                    # The current 'pass' allows root docs even if base_url is set for APIs.
-                    pass
-            return await original_route_handler(request)
-        return custom_route_handler
-
-def create_app():
-    app = FastAPI(
-        title="Webscout OpenAI API",
-        description="OpenAI API compatible interface for various LLM providers",
-        version="0.1.0",
-        docs_url="/docs",
-        redoc_url="/redoc",
-        openapi_url="/openapi.json",
-    )
-    app.router.route_class = DynamicBaseRoute
-    app.add_middleware(
-        CORSMiddleware,
-        allow_origins=["*"],
-        allow_credentials=True,
-        allow_methods=["*"],
-        allow_headers=["*"],
-    )
-    api = Api(app)
-    api.register_authorization()
-    api.register_validation_exception_handler()
-    api.register_routes()
-    initialize_provider_map()
-
-    def custom_openapi():
-        if app.openapi_schema:
-            return app.openapi_schema
-
-        openapi_schema = get_openapi(
-            title=app.title,
-            version=app.version,
-            description=app.description,
-            routes=app.routes,
-        )
-
-        if "components" not in openapi_schema: openapi_schema["components"] = {}
-        if "schemas" not in openapi_schema["components"]: openapi_schema["components"]["schemas"] = {}
-
-        # Use Pydantic's schema generation for accuracy
-        # Assuming Pydantic v1 .schema() or v2 .model_json_schema() based on pydantic_imports
-        # For broader compatibility, trying .schema() first.
-        # If using Pydantic v2 primarily, .model_json_schema() is preferred.
-        schema_method_name = "model_json_schema" if hasattr(BaseModel, "model_json_schema") else "schema"
-
-        # Add/update schemas derived from Pydantic models to ensure they are correctly defined
-        pydantic_models_to_register = {
-            "TextPart": TextPart,
-            "ImageURL": ImageURL,
-            "ImagePart": ImagePart,
-            "Message": Message,
-            "ChatCompletionRequest": ChatCompletionRequest,
-        }
-
-        for name, model_cls in pydantic_models_to_register.items():
-            if schema_method_name == "model_json_schema":
-                schema_data = model_cls.model_json_schema(ref_template="#/components/schemas/{model}")
-            else:
-                schema_data = model_cls.schema()
-            # Pydantic might add a "title" to the schema, which is often not desired for component schemas
-            if "title" in schema_data:
-                del schema_data["title"]
-            openapi_schema["components"]["schemas"][name] = schema_data
-
-        app.openapi_schema = openapi_schema
-        return app.openapi_schema
-
-    app.openapi = custom_openapi
-    return app
-
-def create_app_debug():
-    return create_app()
-
-def initialize_provider_map() -> None:
-    """Initialize the provider map by discovering available providers."""
-    logger.info("Initializing provider map...")
-
-    try:
-        from webscout.Provider.OPENAI.base import OpenAICompatibleProvider
-        module = sys.modules["webscout.Provider.OPENAI"]
-
-        provider_count = 0
-        model_count = 0
-
-        for name, obj in inspect.getmembers(module):
-            if (
-                inspect.isclass(obj)
-                and issubclass(obj, OpenAICompatibleProvider)
-                and obj.__name__ != "OpenAICompatibleProvider"
-            ):
-                provider_name = obj.__name__
-                AppConfig.provider_map[provider_name] = obj
-                config.provider_map[provider_name] = obj
-                provider_count += 1
-
-                # Register available models for this provider
-                if hasattr(obj, "AVAILABLE_MODELS") and isinstance(
-                    obj.AVAILABLE_MODELS, (list, tuple, set)
-                ):
-                    for model in obj.AVAILABLE_MODELS:
-                        if model and isinstance(model, str):
-                            model_key = f"{provider_name}/{model}"
-                            AppConfig.provider_map[model_key] = obj
-                            config.provider_map[model_key] = obj
-                            model_count += 1
-
-        # Fallback to ChatGPT if no providers found
-        if not AppConfig.provider_map:
-            logger.warning("No providers found, using ChatGPT fallback")
-            try:
-                from webscout.Provider.OPENAI.chatgpt import ChatGPT
-                fallback_models = ["gpt-4", "gpt-4o", "gpt-4o-mini", "gpt-3.5-turbo"]
-
-                AppConfig.provider_map["ChatGPT"] = ChatGPT
-                config.provider_map["ChatGPT"] = ChatGPT
-
-                for model in fallback_models:
-                    model_key = f"ChatGPT/{model}"
-                    AppConfig.provider_map[model_key] = ChatGPT
-                    config.provider_map[model_key] = ChatGPT
-
-                AppConfig.default_provider = "ChatGPT"
-                config.default_provider = "ChatGPT"
-                provider_count = 1
-                model_count = len(fallback_models)
-            except ImportError as e:
-                logger.error(f"Failed to import ChatGPT fallback: {e}")
-                raise APIError("No providers available", HTTP_500_INTERNAL_SERVER_ERROR)
-
-        logger.info(f"Initialized {provider_count} providers with {model_count} models")
-
-    except Exception as e:
-        logger.error(f"Failed to initialize provider map: {e}")
-        raise APIError(f"Provider initialization failed: {e}", HTTP_500_INTERNAL_SERVER_ERROR)
-
-class Api:
-    def __init__(self, app: FastAPI) -> None:
-        self.app = app
-        self.get_api_key = APIKeyHeader(name="authorization", auto_error=False)
-
-    def register_authorization(self):
-        @self.app.middleware("http")
-        async def authorization(request: Request, call_next):
-            if AppConfig.api_key is not None:
-                auth_header = await self.get_api_key(request)
-                path = request.url.path
-                if path.startswith("/v1"): # Only protect /v1 routes
-                    # Also allow access to /docs, /openapi.json etc. if AppConfig.base_url is not set or path is not under it
-                    # This logic should be fine as it only protects /v1 paths
-                    if auth_header is None:
-                        return ErrorResponse.from_message("API key required", HTTP_401_UNAUTHORIZED)
-                    if auth_header.startswith("Bearer "):
-                        auth_header = auth_header[7:]
-                    if AppConfig.api_key is None or not secrets.compare_digest(AppConfig.api_key, auth_header): # AppConfig.api_key check is redundant after outer if
-                        return ErrorResponse.from_message("Invalid API key", HTTP_403_FORBIDDEN)
-            return await call_next(request)
-
-    def register_validation_exception_handler(self):
-        """Register comprehensive exception handlers."""
-
-        @self.app.exception_handler(APIError)
-        async def api_error_handler(request: Request, exc: APIError):
-            """Handle custom API errors."""
-            logger.error(f"API Error: {exc.message} (Status: {exc.status_code})")
-            return exc.to_response()
-
-        @self.app.exception_handler(RequestValidationError)
-        async def validation_exception_handler(request: Request, exc: RequestValidationError):
-            errors = exc.errors()
-            error_messages = []
-            body = await request.body()
-            is_empty_body = not body or body.strip() in (b"", b"null", b"{}")
-            for error in errors:
-                loc = error.get("loc", [])
-                # Ensure loc_str is user-friendly
-                loc_str_parts = []
-                for item in loc:
-                    if item == "body": # Skip "body" part if it's the first element of a longer path
-                        if len(loc) > 1: continue
-                    loc_str_parts.append(str(item))
-                loc_str = " -> ".join(loc_str_parts)
-
-                msg = error.get("msg", "Validation error")
-
-                # Check if this error is for the 'content' field specifically due to multimodal input
-                if len(loc) >=3 and loc[0] == 'body' and loc[1] == 'messages' and loc[-1] == 'content':
-                     # Check if the error type suggests a string was expected but a list (or vice-versa) was given for content
-                    if "Input should be a valid string" in msg and error.get("input_type") == "list":
-                         error_messages.append({
-                            "loc": loc,
-                            "message": f"Invalid message content: {msg}. Ensure content matches the expected format (string or list of content parts). Path: {loc_str}",
-                            "type": error.get("type", "validation_error")
-                        })
-                         continue # Skip default message formatting for this specific case
-                    elif "Input should be a valid list" in msg and error.get("input_type") == "string":
-                         error_messages.append({
-                            "loc": loc,
-                            "message": f"Invalid message content: {msg}. Ensure content matches the expected format (string or list of content parts). Path: {loc_str}",
-                            "type": error.get("type", "validation_error")
-                        })
-                         continue
-
-                if "body" in loc:
-                    if len(loc) > 1 and loc[1] == "messages":
-                        error_messages.append({
-                            "loc": loc,
-                            "message": "The 'messages' field is required and must be a non-empty array of message objects. " + f"Error: {msg} at {loc_str}",
-                            "type": error.get("type", "validation_error")
-                        })
-                    elif len(loc) > 1 and loc[1] == "model":
-                        error_messages.append({
-                            "loc": loc,
-                            "message": "The 'model' field is required and must be a string. " + f"Error: {msg} at {loc_str}",
-                            "type": error.get("type", "validation_error")
-                        })
-                    else:
-                        error_messages.append({
-                            "loc": loc,
-                            "message": f"{msg} at {loc_str}",
-                            "type": error.get("type", "validation_error")
-                        })
-                else:
-                    error_messages.append({
-                        "loc": loc,
-                        "message": f"{msg} at {loc_str}",
-                        "type": error.get("type", "validation_error")
-                    })
-            if request.url.path == "/v1/chat/completions":
-                example = ChatCompletionRequest.Config.schema_extra["example"]
-                if is_empty_body:
-                    return JSONResponse(
-                        status_code=HTTP_422_UNPROCESSABLE_ENTITY,
-                        content={
-                            "error": {
-                                "message": "Request body is required and must include 'model' and 'messages'.",
-                                "type": "invalid_request_error",
-                                "param": None,
-                                "code": "body_missing"
-                            },
-                            "example": example
-                        }
-                    )
-                return JSONResponse(
-                    status_code=HTTP_422_UNPROCESSABLE_ENTITY,
-                    content={"detail": error_messages, "example": example}
-                )
-            return JSONResponse(
-                status_code=HTTP_422_UNPROCESSABLE_ENTITY,
-                content={"detail": error_messages}
-            )
-        @self.app.exception_handler(StarletteHTTPException)
-        async def http_exception_handler(request: Request, exc: StarletteHTTPException):
-            return JSONResponse(
-                status_code=exc.status_code,
-                content={"detail": exc.detail}
-            )
-        @self.app.exception_handler(Exception)
-        async def general_exception_handler(request: Request, exc: Exception):
-            return JSONResponse(
-                status_code=HTTP_500_INTERNAL_SERVER_ERROR,
-                content={"detail": f"Internal server error: {str(exc)}"}
-            )
-
-    def register_routes(self):
-        @self.app.get("/", include_in_schema=False)
-        async def root():
-            # Note: If /docs is 404ing, check if server is Uvicorn (expected) or Werkzeug (from logs).
-            # Werkzeug logs suggest possible execution of a Flask app or WSGI misconfiguration.
-            # This api.py file is intended for Uvicorn.
-            return RedirectResponse(url="/docs")
-
-        @self.app.get("/v1/models", response_model=ModelListResponse)
-        async def list_models():
-            models = []
-            for model_name, provider_class in AppConfig.provider_map.items():
-                if "/" not in model_name:
-                    continue  # Skip provider names
-                if any(m["id"] == model_name for m in models):
-                    continue
-                models.append({
-                    "id": model_name,
-                    "object": "model",
-                    "created": int(time.time()),
-                    "owned_by": provider_class.__name__
-                })
-            return {
-                "object": "list",
-                "data": models
-            }
-
-        @self.app.post(
-            "/v1/chat/completions",
-            response_model_exclude_none=True,
-            response_model_exclude_unset=True,
-            openapi_extra={ # This ensures the example is shown in docs
-                "requestBody": {
-                    "content": {
-                        "application/json": {
-                            "schema": {
-                                "$ref": "#/components/schemas/ChatCompletionRequest" # Relies on custom_openapi
-                            },
-                            "example": ChatCompletionRequest.Config.schema_extra["example"]
-                        }
-                    }
-                }
-            }
-        )
-        async def chat_completions(
-            chat_request: ChatCompletionRequest = Body(...)
-        ):
-            """Handle chat completion requests with comprehensive error handling."""
-            start_time = time.time()
-            request_id = f"chatcmpl-{uuid.uuid4()}"
-
-            try:
-                logger.info(f"Processing chat completion request {request_id} for model: {chat_request.model}")
-
-                # Resolve provider and model
-                provider_class, model_name = resolve_provider_and_model(chat_request.model)
-
-                # Initialize provider with error handling
-                try:
-                    provider = provider_class()
-                    logger.debug(f"Initialized provider: {provider_class.__name__}")
-                except Exception as e:
-                    logger.error(f"Failed to initialize provider {provider_class.__name__}: {e}")
-                    raise APIError(
-                        f"Failed to initialize provider {provider_class.__name__}: {e}",
-                        HTTP_500_INTERNAL_SERVER_ERROR,
-                        "provider_error"
-                    )
-
-                # Process and validate messages
-                processed_messages = process_messages(chat_request.messages)
-
-                # Prepare parameters for provider
-                params = prepare_provider_params(chat_request, model_name, processed_messages)
-
-                # Handle streaming vs non-streaming
-                if chat_request.stream:
-                    return await handle_streaming_response(provider, params, request_id)
-                else:
-                    return await handle_non_streaming_response(provider, params, request_id, start_time)
-
-            except APIError:
-                # Re-raise API errors as-is
-                raise
-            except Exception as e:
-                logger.error(f"Unexpected error in chat completion {request_id}: {e}")
-                raise APIError(
-                    f"Internal server error: {str(e)}",
-                    HTTP_500_INTERNAL_SERVER_ERROR,
-                    "internal_error"
-                )
-
-
-def resolve_provider_and_model(model_identifier: str) -> tuple[Any, str]:
-    """Resolve provider class and model name from model identifier."""
-    provider_class = None
-    model_name = None
-
-    # Check for explicit provider/model syntax
-    if model_identifier in AppConfig.provider_map and "/" in model_identifier:
-        provider_class = AppConfig.provider_map[model_identifier]
-        _, model_name = model_identifier.split("/", 1)
-    elif "/" in model_identifier:
-        provider_name, model_name = model_identifier.split("/", 1)
-        provider_class = AppConfig.provider_map.get(provider_name)
-    else:
-        provider_class = AppConfig.provider_map.get(AppConfig.default_provider)
-        model_name = model_identifier
-
-    if not provider_class:
-        available_providers = list(set(v.__name__ for v in AppConfig.provider_map.values()))
-        raise APIError(
-            f"Provider for model '{model_identifier}' not found. Available providers: {available_providers}",
-            HTTP_404_NOT_FOUND,
-            "model_not_found",
-            param="model"
-        )
-
-    # Validate model availability
-    if hasattr(provider_class, "AVAILABLE_MODELS") and model_name is not None:
-        available = getattr(provider_class, "AVAILABLE_MODELS", None)
-        # If it's a property, get from instance
-        if isinstance(available, property):
-            try:
-                available = getattr(provider_class(), "AVAILABLE_MODELS", [])
-            except Exception:
-                available = []
-        # If still not iterable, fallback to empty list
-        if not isinstance(available, (list, tuple, set)):
-            available = list(available) if hasattr(available, "__iter__") and not isinstance(available, str) else []
-        if available and model_name not in available:
-            raise APIError(
-                f"Model '{model_name}' not supported by provider '{provider_class.__name__}'. Available models: {available}",
-                HTTP_404_NOT_FOUND,
-                "model_not_found",
-                param="model"
-            )
-
-    return provider_class, model_name
-
-
-def process_messages(messages: List[Message]) -> List[Dict[str, Any]]:
-    """Process and validate chat messages."""
-    processed_messages = []
-
-    for i, msg_in in enumerate(messages):
-        try:
-            message_dict_out = {"role": msg_in.role}
-
-            if msg_in.content is None:
-                message_dict_out["content"] = None
-            elif isinstance(msg_in.content, str):
-                message_dict_out["content"] = msg_in.content
-            else:  # List[MessageContentParts]
-                message_dict_out["content"] = [
-                    part.model_dump(exclude_none=True) for part in msg_in.content
-                ]
-
-            if msg_in.name:
-                message_dict_out["name"] = msg_in.name
-
-            processed_messages.append(message_dict_out)
-
-        except Exception as e:
-            raise APIError(
-                f"Invalid message at index {i}: {str(e)}",
-                HTTP_422_UNPROCESSABLE_ENTITY,
-                "invalid_request_error",
-                param=f"messages[{i}]"
-            )
-
-    return processed_messages
-
-
-def prepare_provider_params(chat_request: ChatCompletionRequest, model_name: str,
-                          processed_messages: List[Dict[str, Any]]) -> Dict[str, Any]:
-    """Prepare parameters for the provider."""
-    params = {
-        "model": model_name,
-        "messages": processed_messages,
-        "stream": chat_request.stream,
-    }
-
-    # Add optional parameters if present
-    optional_params = [
-        "temperature", "max_tokens", "top_p", "presence_penalty",
-        "frequency_penalty", "stop", "user"
-    ]
-
-    for param in optional_params:
-        value = getattr(chat_request, param, None)
-        if value is not None:
-            params[param] = value
-
-    return params
-
-
-async def handle_streaming_response(provider: Any, params: Dict[str, Any], request_id: str) -> StreamingResponse:
-    """Handle streaming chat completion response."""
-    async def streaming():
-        try:
-            logger.debug(f"Starting streaming response for request {request_id}")
-            completion_stream = provider.chat.completions.create(**params)
-
-            # Check if it's iterable (generator, iterator, or other iterable types)
-            if hasattr(completion_stream, '__iter__') and not isinstance(completion_stream, (str, bytes, dict)):
-                try:
-                    for chunk in completion_stream:
-                        # Standardize chunk format before sending
-                        if hasattr(chunk, 'model_dump'):  # Pydantic v2
-                            chunk_data = chunk.model_dump(exclude_none=True)
-                        elif hasattr(chunk, 'dict'):  # Pydantic v1
-                            chunk_data = chunk.dict(exclude_none=True)
-                        elif isinstance(chunk, dict):
-                            chunk_data = chunk
-                        else:  # Fallback for unknown chunk types
-                            chunk_data = chunk
-                        
-                        # Clean text content in the chunk to remove control characters
-                        if isinstance(chunk_data, dict) and 'choices' in chunk_data:
-                            for choice in chunk_data.get('choices', []):
-                                if isinstance(choice, dict):
-                                    # Handle delta for streaming
-                                    if 'delta' in choice and isinstance(choice['delta'], dict) and 'content' in choice['delta']:
-                                        choice['delta']['content'] = clean_text(choice['delta']['content'])
-                                    # Handle message for non-streaming
-                                    elif 'message' in choice and isinstance(choice['message'], dict) and 'content' in choice['message']:
-                                        choice['message']['content'] = clean_text(choice['message']['content'])
-                        
-                        yield f"data: {json.dumps(chunk_data, ensure_ascii=False)}\n\n"
-                except TypeError as te:
-                    logger.error(f"Error iterating over completion_stream: {te}")
-                    # Fall back to treating as non-generator response
-                    if hasattr(completion_stream, 'model_dump'):
-                        response_data = completion_stream.model_dump(exclude_none=True)
-                    elif hasattr(completion_stream, 'dict'):
-                        response_data = completion_stream.dict(exclude_none=True)
-                    else:
-                        response_data = completion_stream
-                    
-                    # Clean text content in the response
-                    if isinstance(response_data, dict) and 'choices' in response_data:
-                        for choice in response_data.get('choices', []):
-                            if isinstance(choice, dict):
-                                if 'delta' in choice and isinstance(choice['delta'], dict) and 'content' in choice['delta']:
-                                    choice['delta']['content'] = clean_text(choice['delta']['content'])
-                                elif 'message' in choice and isinstance(choice['message'], dict) and 'content' in choice['message']:
-                                    choice['message']['content'] = clean_text(choice['message']['content'])
-                    
-                    yield f"data: {json.dumps(response_data, ensure_ascii=False)}\n\n"
-            else:  # Non-generator response
-                if hasattr(completion_stream, 'model_dump'):
-                    response_data = completion_stream.model_dump(exclude_none=True)
-                elif hasattr(completion_stream, 'dict'):
-                    response_data = completion_stream.dict(exclude_none=True)
-                else:
-                    response_data = completion_stream
-                
-                # Clean text content in the response
-                if isinstance(response_data, dict) and 'choices' in response_data:
-                    for choice in response_data.get('choices', []):
-                        if isinstance(choice, dict):
-                            if 'delta' in choice and isinstance(choice['delta'], dict) and 'content' in choice['delta']:
-                                choice['delta']['content'] = clean_text(choice['delta']['content'])
-                            elif 'message' in choice and isinstance(choice['message'], dict) and 'content' in choice['message']:
-                                choice['message']['content'] = clean_text(choice['message']['content'])
-                
-                yield f"data: {json.dumps(response_data, ensure_ascii=False)}\n\n"
-
-        except Exception as e:
-            logger.error(f"Error in streaming response for request {request_id}: {e}")
-            error_message = clean_text(str(e))
-            error_data = {
-                "error": {
-                    "message": error_message,
-                    "type": "server_error",
-                    "code": "streaming_error"
-                }
-            }
-            yield f"data: {json.dumps(error_data, ensure_ascii=False)}\n\n"
-        finally:
-            yield "data: [DONE]\n\n"
-    return StreamingResponse(streaming(), media_type="text/event-stream")
-
-
-async def handle_non_streaming_response(provider: Any, params: Dict[str, Any],
-                                      request_id: str, start_time: float) -> Dict[str, Any]:
-    """Handle non-streaming chat completion response."""
-    try:
-        logger.debug(f"Starting non-streaming response for request {request_id}")
-        completion = provider.chat.completions.create(**params)
-
-        if completion is None:
-            # Return a valid OpenAI-compatible error response
-            return ChatCompletion(
-                id=request_id,
-                created=int(time.time()),
-                model=params.get("model", "unknown"),
-                choices=[Choice(
-                    index=0,
-                    message=ChatCompletionMessage(role="assistant", content="No response generated."),
-                    finish_reason="error"
-                )],
-                usage=CompletionUsage(prompt_tokens=0, completion_tokens=0, total_tokens=0)
-            ).model_dump(exclude_none=True)
-
-        # Standardize response format
-        if hasattr(completion, "model_dump"):  # Pydantic v2
-            response_data = completion.model_dump(exclude_none=True)
-        elif hasattr(completion, "dict"):  # Pydantic v1
-            response_data = completion.dict(exclude_none=True)
-        elif isinstance(completion, dict):
-            response_data = completion
-        else:
-            raise APIError(
-                "Invalid response format from provider",
-                HTTP_500_INTERNAL_SERVER_ERROR,
-                "provider_error"
-            )
-        
-        # Clean text content in the response to remove control characters
-        if isinstance(response_data, dict) and 'choices' in response_data:
-            for choice in response_data.get('choices', []):
-                if isinstance(choice, dict) and 'message' in choice:
-                    if isinstance(choice['message'], dict) and 'content' in choice['message']:
-                        choice['message']['content'] = clean_text(choice['message']['content'])
-
-        elapsed = time.time() - start_time
-        logger.info(f"Completed non-streaming request {request_id} in {elapsed:.2f}s")
-
-        return response_data
-
-    except Exception as e:
-        logger.error(f"Error in non-streaming response for request {request_id}: {e}")
-        error_message = clean_text(str(e))
-        raise APIError(
-            f"Provider error: {error_message}",
-            HTTP_500_INTERNAL_SERVER_ERROR,
-            "provider_error"
-        )
-
-def format_exception(e: Union[Exception, str]) -> str:
-    if isinstance(e, str):
-        message = e
-    else:
-        message = f"{e.__class__.__name__}: {str(e)}" # Keep it concise
-    return json.dumps({
-        "error": {
-            "message": message,
-            "type": "server_error", # Or more specific if possible
-            "param": None,
-            "code": "internal_server_error" # Or more specific
-        }
-    })
-
-def start_server(
-    port: int = DEFAULT_PORT,
-    host: str = DEFAULT_HOST,
-    api_key: str = None,
-    default_provider: str = None,
-    base_url: str = None,
-    workers: int = 1,
-    log_level: str = 'info',
-    debug: bool = False
-):
-    """Start the API server with the given configuration."""
-    run_api(
-        host=host,
-        port=port,
-        api_key=api_key,
-        default_provider=default_provider,
-        base_url=base_url,
-        workers=workers,
-        log_level=log_level,
-        debug=debug,
-    )
-
-def run_api(
-    host: str = '0.0.0.0',
-    port: int = None,
-    api_key: str = None,
-    default_provider: str = None,
-    base_url: str = None,
-    debug: bool = False,
-    workers: int = 1,
-    log_level: str = 'info',
-    show_available_providers: bool = True,
-) -> None:
-    print("Starting Webscout OpenAI API server...")
-    if port is None:
-        port = DEFAULT_PORT
-    AppConfig.set_config(
-        api_key=api_key,
-        default_provider=default_provider or AppConfig.default_provider,
-        base_url=base_url
-    )
-    # initialize_provider_map() # This is called inside create_app now.
-                              # Call here if create_app doesn't exist yet or for early info.
-                              # For showing providers, it needs to be called before printing.
-    if show_available_providers: # Initialize map if needed for display before app creation
-        if not AppConfig.provider_map: # Avoid re-initializing if already done by app creation logic path
-            initialize_provider_map()
-
-        print("\n=== Webscout OpenAI API Server ===")
-        print(f"Server URL: http://{host if host != '0.0.0.0' else 'localhost'}:{port}")
-        if AppConfig.base_url:
-            print(f"Base Path: {AppConfig.base_url}")
-            api_endpoint_base = f"http://{host if host != '0.0.0.0' else 'localhost'}:{port}{AppConfig.base_url}"
-        else:
-            api_endpoint_base = f"http://{host if host != '0.0.0.0' else 'localhost'}:{port}"
-
-        print(f"API Endpoint: {api_endpoint_base}/v1/chat/completions")
-        print(f"Docs URL: {api_endpoint_base}/docs") # Adjusted for potential base_url in display
-        print(f"API Authentication: {'Enabled' if api_key else 'Disabled'}")
-        print(f"Default Provider: {AppConfig.default_provider}")
-        print(f"Workers: {workers}")
-        print(f"Log Level: {log_level}")
-        print(f"Debug Mode: {'Enabled' if debug else 'Disabled'}")
-
-        providers = list(set(v.__name__ for v in AppConfig.provider_map.values()))
-        print(f"\n--- Available Providers ({len(providers)}) ---")
-        for i, provider_name in enumerate(sorted(providers), 1):
-            print(f"{i}. {provider_name}")
-
-        provider_class_names = set(v.__name__ for v in AppConfig.provider_map.values())
-        models = sorted([model for model in AppConfig.provider_map.keys() if model not in provider_class_names])
-        if models:
-            print(f"\n--- Available Models ({len(models)}) ---")
-            for i, model_name in enumerate(models, 1):
-                print(f"{i}. {model_name} (via {AppConfig.provider_map[model_name].__name__})")
-        else:
-            print("\nNo specific models registered. Use provider names as models.")
-
-        print("\nUse Ctrl+C to stop the server.")
-        print("=" * 40 + "\n")
-
-    uvicorn_app_str = "webscout.Provider.OPENAI.api:create_app_debug" if debug else "webscout.Provider.OPENAI.api:create_app"
-
-    # Configure uvicorn settings
-    uvicorn_config = {
-        "app": uvicorn_app_str,
-        "host": host,
-        "port": int(port),
-        "factory": True,
-        "reload": debug,  # Enable reload only in debug mode for stability
-        "log_level": log_level.lower() if log_level else ("debug" if debug else "info"),
-    }
-
-    # Add workers only if not in debug mode (reload and workers are incompatible)
-    if not debug and workers > 1:
-        uvicorn_config["workers"] = workers
-        print(f"Starting with {workers} workers...")
-    elif debug:
-        print("Debug mode enabled - using single worker with reload...")
-
-    # Note: Logs show "werkzeug". If /docs 404s persist, ensure Uvicorn is the actual server running.
-    # The script uses uvicorn.run, so "werkzeug" logs are unexpected for this file.
-    uvicorn.run(**uvicorn_config)
-
-if __name__ == "__main__":
-    import argparse
-
-    # Read environment variables with fallbacks
-    default_port = int(os.getenv('WEBSCOUT_PORT', os.getenv('PORT', DEFAULT_PORT)))
-    default_host = os.getenv('WEBSCOUT_HOST', DEFAULT_HOST)
-    default_workers = int(os.getenv('WEBSCOUT_WORKERS', '1'))
-    default_log_level = os.getenv('WEBSCOUT_LOG_LEVEL', 'info')
-    default_api_key = os.getenv('WEBSCOUT_API_KEY', os.getenv('API_KEY'))
-    default_provider = os.getenv('WEBSCOUT_DEFAULT_PROVIDER', os.getenv('DEFAULT_PROVIDER'))
-    default_base_url = os.getenv('WEBSCOUT_BASE_URL', os.getenv('BASE_URL'))
-    default_debug = os.getenv('WEBSCOUT_DEBUG', os.getenv('DEBUG', 'false')).lower() == 'true'
-
-    parser = argparse.ArgumentParser(description='Start Webscout OpenAI-compatible API server')
-    parser.add_argument('--port', type=int, default=default_port, help=f'Port to run the server on (default: {default_port})')
-    parser.add_argument('--host', type=str, default=default_host, help=f'Host to bind the server to (default: {default_host})')
-    parser.add_argument('--workers', type=int, default=default_workers, help=f'Number of worker processes (default: {default_workers})')
-    parser.add_argument('--log-level', type=str, default=default_log_level, choices=['debug', 'info', 'warning', 'error', 'critical'], help=f'Log level (default: {default_log_level})')
-    parser.add_argument('--api-key', type=str, default=default_api_key, help='API key for authentication (optional)')
-    parser.add_argument('--default-provider', type=str, default=default_provider, help='Default provider to use (optional)')
-    parser.add_argument('--base-url', type=str, default=default_base_url, help='Base URL for the API (optional, e.g., /api/v1)')
-    parser.add_argument('--debug', action='store_true', default=default_debug, help='Run in debug mode')
-    args = parser.parse_args()
-
-    # Print configuration summary
-    print(f"Configuration:")
-    print(f"  Host: {args.host}")
-    print(f"  Port: {args.port}")
-    print(f"  Workers: {args.workers}")
-    print(f"  Log Level: {args.log_level}")
-    print(f"  Debug Mode: {args.debug}")
-    print(f"  API Key: {'Set' if args.api_key else 'Not set'}")
-    print(f"  Default Provider: {args.default_provider or 'Not set'}")
-    print(f"  Base URL: {args.base_url or 'Not set'}")
-    print()
-
-    run_api(
-        host=args.host,
-        port=args.port,
-        workers=args.workers,
-        log_level=args.log_level,
-        api_key=args.api_key,
-        default_provider=args.default_provider,
-        base_url=args.base_url,
-        debug=args.debug
-    )