api/utils.py

from datetime import datetime
import json
import uuid
import asyncio
import random
import string
from typing import Any, Dict, Optional

import httpx
from fastapi import HTTPException
from api.config import (
    MODEL_ALIASES,
    get_headers_api_chat,
    get_headers_chat,
    BASE_URL,
    AGENT_MODE,
    TRENDING_AGENT_MODE,
    MODEL_PREFIXES,
    API_ENDPOINT,
    generate_id,
    MODELS
)
from api.models import ChatRequest
from api.logger import setup_logger
from api.image import ImageResponse, to_data_uri  # Assuming image utilities are here

logger = setup_logger(__name__)

# Helper function to create chat completion data
def create_chat_completion_data(
    content: str, model: str, timestamp: int, finish_reason: Optional[str] = None
) -> Dict[str, Any]:
    return {
        "id": f"chatcmpl-{uuid.uuid4()}",
        "object": "chat.completion.chunk",
        "created": timestamp,
        "model": model,
        "choices": [
            {
                "index": 0,
                "delta": {"content": content, "role": "assistant"},
                "finish_reason": finish_reason,
            }
        ],
        "usage": None,
    }

# Function to convert message to dictionary format, ensuring base64 data and optional model prefix
def message_to_dict(message, model_prefix: Optional[str] = None):
    content = message.content if isinstance(message.content, str) else message.content[0]["text"]
    if model_prefix:
        content = f"{model_prefix} {content}"
    if isinstance(message.content, list) and len(message.content) == 2 and "image_url" in message.content[1]:
        # Ensure base64 images are always included for all models
        return {
            "role": message.role,
            "content": content,
            "data": {
                "imageBase64": message.content[1]["image_url"]["url"],
                "fileText": "",
                "title": "snapshot",
            },
        }
    return {"role": message.role, "content": content}

# Function to strip model prefix from content if present
def strip_model_prefix(content: str, model_prefix: Optional[str] = None) -> str:
    """Remove the model prefix from the response content if present."""
    if model_prefix and content.startswith(model_prefix):
        logger.debug(f"Stripping prefix '{model_prefix}' from content.")
        return content[len(model_prefix):].strip()
    return content

# Function to get the correct referer URL for logging
def get_referer_url(chat_id: str, model: str) -> str:
    """Generate the referer URL based on specific models listed in MODELS."""
    return f"{BASE_URL}/chat/{chat_id}?model={model}"

# Process streaming response with headers from config.py
async def process_streaming_response(request: ChatRequest, web_search: bool = False):
    chat_id = generate_id()
    referer_url = get_referer_url(chat_id, request.model)
    logger.info(f"Generated Chat ID: {chat_id} - Model: {request.model} - URL: {referer_url}")

    # Resolve model aliases
    resolved_model = MODEL_ALIASES.get(request.model, request.model)

    agent_mode = AGENT_MODE.get(resolved_model, {})
    trending_agent_mode = TRENDING_AGENT_MODE.get(resolved_model, {})
    model_prefix = MODEL_PREFIXES.get(resolved_model, "")

    headers_api_chat = get_headers_api_chat()

    if resolved_model == 'o1-preview':
        delay_seconds = random.randint(1, 60)
        logger.info(f"Introducing a delay of {delay_seconds} seconds for model 'o1-preview' (Chat ID: {chat_id})")
        await asyncio.sleep(delay_seconds)

    json_data = {
        "agentMode": agent_mode,
        "clickedAnswer2": False,
        "clickedAnswer3": False,
        "clickedForceWebSearch": False,
        "codeModelMode": True,
        "githubToken": None,
        "id": chat_id,
        "isChromeExt": False,
        "isMicMode": False,
        "maxTokens": request.max_tokens,
        "messages": [message_to_dict(msg, model_prefix=model_prefix) for msg in request.messages],
        "mobileClient": False,
        "playgroundTemperature": request.temperature,
        "playgroundTopP": request.top_p,
        "previewToken": None,
        "trendingAgentMode": trending_agent_mode,
        "userId": None,
        "userSelectedModel": resolved_model if resolved_model in MODELS else None,
        "userSystemPrompt": None,
        "validated": "69783381-2ce4-4dbd-ac78-35e9063feabc",
        "visitFromDelta": False,
        "webSearchMode": web_search,  # Include web search mode
    }

    async with httpx.AsyncClient() as client:
        try:
            async with client.stream(
                "POST",
                API_ENDPOINT,
                headers=headers_api_chat,
                json=json_data,
                timeout=100,
            ) as response:
                response.raise_for_status()
                async for line in response.aiter_lines():
                    timestamp = int(datetime.now().timestamp())
                    if line:
                        content = line
                        if content.startswith("$@$v=undefined-rv1$@$"):
                            content = content[21:]
                        cleaned_content = strip_model_prefix(content, model_prefix)
                        yield f"data: {json.dumps(create_chat_completion_data(cleaned_content, resolved_model, timestamp))}\n\n"

                yield f"data: {json.dumps(create_chat_completion_data('', resolved_model, timestamp, 'stop'))}\n\n"
                yield "data: [DONE]\n\n"
        except httpx.HTTPStatusError as e:
            logger.error(f"HTTP error occurred for Chat ID {chat_id}: {e}")
            raise HTTPException(status_code=e.response.status_code, detail=str(e))
        except httpx.RequestError as e:
            logger.error(f"Error occurred during request for Chat ID {chat_id}: {e}")
            raise HTTPException(status_code=500, detail=str(e))

# Process non-streaming response with headers from config.py
async def process_non_streaming_response(request: ChatRequest, web_search: bool = False):
    chat_id = generate_id()
    referer_url = get_referer_url(chat_id, request.model)
    logger.info(f"Generated Chat ID: {chat_id} - Model: {request.model} - URL: {referer_url}")

    # Resolve model aliases
    resolved_model = MODEL_ALIASES.get(request.model, request.model)

    agent_mode = AGENT_MODE.get(resolved_model, {})
    trending_agent_mode = TRENDING_AGENT_MODE.get(resolved_model, {})
    model_prefix = MODEL_PREFIXES.get(resolved_model, "")

    headers_api_chat = get_headers_api_chat()

    if resolved_model == 'o1-preview':
        delay_seconds = random.randint(20, 60)
        logger.info(f"Introducing a delay of {delay_seconds} seconds for model 'o1-preview' (Chat ID: {chat_id})")
        await asyncio.sleep(delay_seconds)

    json_data = {
        "agentMode": agent_mode,
        "clickedAnswer2": False,
        "clickedAnswer3": False,
        "clickedForceWebSearch": False,
        "codeModelMode": True,
        "githubToken": None,
        "id": chat_id,
        "isChromeExt": False,
        "isMicMode": False,
        "maxTokens": request.max_tokens,
        "messages": [message_to_dict(msg, model_prefix=model_prefix) for msg in request.messages],
        "mobileClient": False,
        "playgroundTemperature": request.temperature,
        "playgroundTopP": request.top_p,
        "previewToken": None,
        "trendingAgentMode": trending_agent_mode,
        "userId": None,
        "userSelectedModel": resolved_model if resolved_model in MODELS else None,
        "userSystemPrompt": None,
        "validated": "69783381-2ce4-4dbd-ac78-35e9063feabc",
        "visitFromDelta": False,
        "webSearchMode": web_search,  # Include web search mode
    }

    full_response = ""
    async with httpx.AsyncClient() as client:
        try:
            async with client.stream(
                method="POST",
                url=API_ENDPOINT,
                headers=headers_api_chat,
                json=json_data,
                timeout=100,
            ) as response:
                response.raise_for_status()
                async for chunk in response.aiter_text():
                    full_response += chunk
        except httpx.HTTPStatusError as e:
            logger.error(f"HTTP error occurred for Chat ID {chat_id}: {e}")
            raise HTTPException(status_code=e.response.status_code, detail=str(e))
        except httpx.RequestError as e:
            logger.error(f"Error occurred during request for Chat ID {chat_id}: {e}")
            raise HTTPException(status_code=500, detail=str(e))

    if full_response.startswith("$@$v=undefined-rv1$@$"):
        full_response = full_response[21:]

    cleaned_full_response = strip_model_prefix(full_response, model_prefix)

    return {
        "id": f"chatcmpl-{uuid.uuid4()}",
        "object": "chat.completion",
        "created": int(datetime.now().timestamp()),
        "model": resolved_model,
        "choices": [
            {
                "index": 0,
                "message": {"role": "assistant", "content": cleaned_full_response},
                "finish_reason": "stop",
            }
        ],
        "usage": None,
    }