test24 / api /utils.py
Niansuh's picture
Update api/utils.py
41e74b3 verified
raw
history blame
9.33 kB
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_MAPPING,
get_headers_api_chat,
get_headers_chat,
BASE_URL,
AGENT_MODE,
TRENDING_AGENT_MODE,
MODEL_PREFIXES,
MODEL_REFERERS
)
from api.models import ChatRequest # Ensure ChatRequest is correctly imported
from api.logger import (
log_generated_chat_id,
log_model_delay,
log_http_error,
log_request_error,
log_strip_prefix
)
# Helper function to create a random alphanumeric chat ID
def generate_chat_id(length: int = 7) -> str:
characters = string.ascii_letters + string.digits
return ''.join(random.choices(characters, k=length))
# 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 with optional model prefix
def message_to_dict(message, model_prefix: Optional[str] = None):
if isinstance(message.content, str):
content = message.content
if model_prefix:
content = f"{model_prefix} {content}"
return {"role": message.role, "content": content}
elif isinstance(message.content, list) and len(message.content) == 2:
content = message.content[0]["text"]
if model_prefix:
content = f"{model_prefix} {content}"
return {
"role": message.role,
"content": content,
"data": {
"imageBase64": message.content[1]["image_url"]["url"],
"fileText": "",
"title": "snapshot",
},
}
else:
return {"role": message.role, "content": message.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):
log_strip_prefix(model_prefix)
return content[len(model_prefix):].strip()
return content
# Process streaming response with headers from config.py
async def process_streaming_response(request: ChatRequest):
# Determine if a Chat ID and referer URL should be generated
if request.model in MODEL_REFERERS:
chat_id = generate_chat_id()
chat_url = f"/chat/{chat_id}?model={request.model}"
log_generated_chat_id(chat_id, chat_url)
referer_path = MODEL_REFERERS[request.model]
referer_url = f"{BASE_URL}{referer_path}"
else:
chat_id = None
referer_url = BASE_URL # Use base URL for models not in MODEL_REFERERS
agent_mode = AGENT_MODE.get(request.model, {})
trending_agent_mode = TRENDING_AGENT_MODE.get(request.model, {})
model_prefix = MODEL_PREFIXES.get(request.model, "")
headers_api_chat = get_headers_api_chat(referer_url)
if request.model == 'o1-preview':
delay_seconds = random.randint(20, 60)
log_model_delay(delay_seconds, chat_id, request.model if chat_id else "unknown")
await asyncio.sleep(delay_seconds)
json_data = {
"agentMode": agent_mode,
"clickedAnswer2": False,
"clickedAnswer3": False,
"clickedForceWebSearch": False,
"codeModelMode": True,
"githubToken": None,
"id": chat_id if chat_id else "unknown",
"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": MODEL_MAPPING.get(request.model, request.model),
"userSystemPrompt": None,
"validated": "69783381-2ce4-4dbd-ac78-35e9063feabc",
"visitFromDelta": False,
}
async with httpx.AsyncClient() as client:
try:
async with client.stream(
"POST",
f"{BASE_URL}/api/chat",
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, request.model, timestamp))}\n\n"
yield f"data: {json.dumps(create_chat_completion_data('', request.model, timestamp, 'stop'))}\n\n"
yield "data: [DONE]\n\n"
except httpx.HTTPStatusError as e:
log_http_error(chat_id if chat_id else "unknown", e)
raise HTTPException(status_code=e.response.status_code, detail=str(e))
except httpx.RequestError as e:
log_request_error(chat_id if chat_id else "unknown", 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):
if request.model in MODEL_REFERERS:
chat_id = generate_chat_id()
chat_url = f"/chat/{chat_id}?model={request.model}"
log_generated_chat_id(chat_id, chat_url)
referer_path = MODEL_REFERERS[request.model]
referer_url = f"{BASE_URL}{referer_path}"
else:
chat_id = None
referer_url = BASE_URL
agent_mode = AGENT_MODE.get(request.model, {})
trending_agent_mode = TRENDING_AGENT_MODE.get(request.model, {})
model_prefix = MODEL_PREFIXES.get(request.model, "")
headers_api_chat = get_headers_api_chat(referer_url)
headers_chat = get_headers_chat(chat_url, next_action=str(uuid.uuid4()), next_router_state_tree=json.dumps([""])) if chat_id else None
if request.model == 'o1-preview':
delay_seconds = random.randint(20, 60)
log_model_delay(delay_seconds, chat_id if chat_id else "unknown", request.model)
await asyncio.sleep(delay_seconds)
json_data = {
"agentMode": agent_mode,
"clickedAnswer2": False,
"clickedAnswer3": False,
"clickedForceWebSearch": False,
"codeModelMode": True,
"githubToken": None,
"id": chat_id if chat_id else "unknown",
"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": MODEL_MAPPING.get(request.model, request.model),
"userSystemPrompt": None,
"validated": "69783381-2ce4-4dbd-ac78-35e9063feabc",
"visitFromDelta": False,
}
full_response = ""
async with httpx.AsyncClient() as client:
try:
async with client.stream(
method="POST", url=f"{BASE_URL}/api/chat", headers=headers_api_chat, json=json_data
) as response:
response.raise_for_status()
async for chunk in response.aiter_text():
full_response += chunk
except httpx.HTTPStatusError as e:
log_http_error(chat_id if chat_id else "unknown", e)
raise HTTPException(status_code=e.response.status_code, detail=str(e))
except httpx.RequestError as e:
log_request_error(chat_id if chat_id else "unknown", 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": request.model,
"choices": [
{
"index": 0,
"message": {"role": "assistant", "content": cleaned_full_response},
"finish_reason": "stop",
}
],
"usage": None,
}