Spaces:
Sleeping
Sleeping
Update main.py
Browse files
main.py
CHANGED
@@ -1,271 +1,46 @@
|
|
1 |
-
from fastapi import FastAPI, Request, HTTPException
|
2 |
-
from fastapi.responses import JSONResponse, StreamingResponse
|
3 |
-
from fastapi.middleware.cors import CORSMiddleware
|
4 |
-
from pydantic import BaseModel
|
5 |
-
from typing import List, Dict, Any, Union
|
6 |
import os
|
7 |
-
import
|
8 |
-
import
|
9 |
-
import json
|
10 |
from dotenv import load_dotenv
|
|
|
11 |
|
12 |
load_dotenv()
|
|
|
|
|
13 |
|
14 |
-
|
15 |
-
|
16 |
-
MAX_RETRY_COUNT = int(os.getenv("MAX_RETRY_COUNT", "3"))
|
17 |
-
RETRY_DELAY = int(os.getenv("RETRY_DELAY", "5000"))
|
18 |
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
"Accept-Language": "en-US,en;q=0.9",
|
23 |
-
"Origin": "https://duckduckgo.com/",
|
24 |
-
"Referer": "https://duckduckgo.com/",
|
25 |
-
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/130.0.0.0 Safari/537.36",
|
26 |
-
}
|
27 |
|
28 |
-
app =
|
|
|
|
|
|
|
29 |
|
30 |
-
|
31 |
-
app.add_middleware(
|
32 |
-
CORSMiddleware,
|
33 |
-
allow_origins=["*"],
|
34 |
-
allow_methods=["*"],
|
35 |
-
allow_headers=["*"],
|
36 |
-
)
|
37 |
|
38 |
-
#
|
39 |
-
|
40 |
-
role: str
|
41 |
-
content: Union[str, List[Dict[str, Any]]]
|
42 |
|
43 |
-
|
44 |
-
|
45 |
-
messages: List[Message]
|
46 |
-
stream: bool = False
|
47 |
|
48 |
-
|
49 |
-
|
50 |
-
async def add_process_time(request: Request, call_next):
|
51 |
-
start_time = time.time()
|
52 |
-
response = await call_next(request)
|
53 |
-
process_time = time.time() - start_time
|
54 |
-
print(f"{request.method} {response.status_code} {request.url.path} {process_time*1000:.2f} ms")
|
55 |
-
return response
|
56 |
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
async def ping():
|
63 |
-
return {"message": "pong"}
|
64 |
-
|
65 |
-
@app.get(f"{API_PREFIX}v1/models")
|
66 |
-
async def get_models():
|
67 |
-
return {
|
68 |
-
"object": "list",
|
69 |
-
"data": [
|
70 |
-
{"id": "gpt-4o-mini", "object": "model", "owned_by": "ddg"},
|
71 |
-
{"id": "claude-3-haiku", "object": "model", "owned_by": "ddg"},
|
72 |
-
{"id": "llama-3.1-70b", "object": "model", "owned_by": "ddg"},
|
73 |
-
{"id": "mixtral-8x7b", "object": "model", "owned_by": "ddg"},
|
74 |
-
{"id": "o3-mini", "object": "model", "owned_by": "ddg"},
|
75 |
],
|
76 |
-
|
77 |
-
|
78 |
-
@app.post(f"{API_PREFIX}v1/chat/completions")
|
79 |
-
async def chat_completions(request: ChatCompletionRequest):
|
80 |
-
try:
|
81 |
-
model = convert_model(request.model)
|
82 |
-
content = messages_to_text(request.messages)
|
83 |
-
return await create_completion(model, content, request.stream)
|
84 |
-
except Exception as e:
|
85 |
-
raise HTTPException(status_code=500, detail=str(e))
|
86 |
-
|
87 |
-
def convert_model(input_model: str) -> str:
|
88 |
-
"""Convert public model names to DuckDuckGo internal model names"""
|
89 |
-
model_mapping = {
|
90 |
-
"claude-3-haiku": "claude-3-haiku-20240307",
|
91 |
-
"llama-3.1-70b": "meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
|
92 |
-
"mixtral-8x7b": "mistralai/Mixtral-8x7B-Instruct-v0.1",
|
93 |
-
"o3-mini": "o3-mini"
|
94 |
-
}
|
95 |
-
return model_mapping.get(input_model.lower(), "gpt-4o-mini")
|
96 |
-
|
97 |
-
def messages_to_text(messages: List[Message]) -> str:
|
98 |
-
"""Convert message array to text format expected by DuckDuckGo API"""
|
99 |
-
result = ""
|
100 |
-
for message in messages:
|
101 |
-
role = "user" if message.role == "system" else message.role
|
102 |
-
|
103 |
-
if role in ["user", "assistant"]:
|
104 |
-
# Handle both string content and structured content
|
105 |
-
if isinstance(message.content, list):
|
106 |
-
content_str = "".join([item.get("text", "") for item in message.content if item.get("text", "")])
|
107 |
-
else:
|
108 |
-
content_str = message.content
|
109 |
-
|
110 |
-
result += f"{role}:{content_str};\r\n"
|
111 |
|
112 |
-
return
|
113 |
-
|
114 |
-
async def request_token() -> str:
|
115 |
-
"""Get auth token from DuckDuckGo"""
|
116 |
-
try:
|
117 |
-
async with httpx.AsyncClient() as client:
|
118 |
-
response = await client.get(
|
119 |
-
"https://duckduckgo.com/duckchat/v1/status",
|
120 |
-
headers={**FAKE_HEADERS, "x-vqd-accept": "1"}
|
121 |
-
)
|
122 |
-
return response.headers.get("x-vqd-4", "")
|
123 |
-
except Exception as e:
|
124 |
-
print(f"Token request error: {e}")
|
125 |
-
return ""
|
126 |
-
|
127 |
-
async def create_completion(model: str, content: str, return_stream: bool, retry_count: int = 0):
|
128 |
-
"""Create a chat completion via DuckDuckGo API"""
|
129 |
-
token = await request_token()
|
130 |
-
|
131 |
-
try:
|
132 |
-
async with httpx.AsyncClient() as client:
|
133 |
-
response = await client.post(
|
134 |
-
"https://duckduckgo.com/duckchat/v1/chat",
|
135 |
-
headers={
|
136 |
-
**FAKE_HEADERS,
|
137 |
-
"Accept": "text/event-stream",
|
138 |
-
"Content-Type": "application/json",
|
139 |
-
"x-vqd-4": token,
|
140 |
-
},
|
141 |
-
json={
|
142 |
-
"model": model,
|
143 |
-
"messages": [{"role": "user", "content": content}]
|
144 |
-
},
|
145 |
-
stream=True
|
146 |
-
)
|
147 |
-
|
148 |
-
if response.status_code != 200:
|
149 |
-
raise HTTPException(status_code=response.status_code, detail="API request failed")
|
150 |
-
|
151 |
-
return await process_stream(model, response, return_stream)
|
152 |
-
except Exception as e:
|
153 |
-
if retry_count < MAX_RETRY_COUNT:
|
154 |
-
print(f"Retrying... attempt {retry_count + 1}")
|
155 |
-
await asyncio.sleep(RETRY_DELAY / 1000)
|
156 |
-
return await create_completion(model, content, return_stream, retry_count + 1)
|
157 |
-
raise HTTPException(status_code=500, detail=str(e))
|
158 |
-
|
159 |
-
async def process_stream(model: str, response, return_stream: bool):
|
160 |
-
"""Process streaming response from DuckDuckGo"""
|
161 |
-
buffer = ""
|
162 |
-
full_text = ""
|
163 |
-
|
164 |
-
async def generate_stream():
|
165 |
-
nonlocal buffer, full_text
|
166 |
-
|
167 |
-
# Process chunks as they arrive
|
168 |
-
async for chunk in response.aiter_bytes():
|
169 |
-
chunk_str = chunk.decode('utf-8').strip()
|
170 |
-
|
171 |
-
# Handle buffer from previous chunk if needed
|
172 |
-
if buffer:
|
173 |
-
chunk_str = buffer + chunk_str
|
174 |
-
buffer = ""
|
175 |
-
|
176 |
-
# Handle incomplete chunks
|
177 |
-
if not chunk_str.endswith('"}') and "[DONE]" not in chunk_str:
|
178 |
-
buffer = chunk_str
|
179 |
-
continue
|
180 |
-
|
181 |
-
# Process each line in the chunk
|
182 |
-
for line in chunk_str.split('\n'):
|
183 |
-
if len(line) < 6:
|
184 |
-
continue
|
185 |
-
|
186 |
-
# Remove prefix (data: )
|
187 |
-
line = line[6:] if line.startswith("data: ") else line
|
188 |
-
|
189 |
-
# Handle completion signal
|
190 |
-
if line == "[DONE]":
|
191 |
-
if return_stream:
|
192 |
-
yield f"data: {json.dumps(create_stop_chunk(model))}\n\n"
|
193 |
-
return
|
194 |
-
|
195 |
-
# Parse and handle message content
|
196 |
-
try:
|
197 |
-
data = json.loads(line)
|
198 |
-
if data.get("action") == "success" and "message" in data:
|
199 |
-
message = data["message"]
|
200 |
-
full_text += message
|
201 |
-
|
202 |
-
if return_stream:
|
203 |
-
yield f"data: {json.dumps(create_chunk(message, model))}\n\n"
|
204 |
-
except json.JSONDecodeError:
|
205 |
-
continue
|
206 |
-
|
207 |
-
# Return appropriate response based on streaming preference
|
208 |
-
if return_stream:
|
209 |
-
return StreamingResponse(generate_stream(), media_type="text/event-stream")
|
210 |
-
else:
|
211 |
-
# For non-streaming, consume the generator and return complete response
|
212 |
-
async for _ in generate_stream():
|
213 |
-
pass # Just collecting text in full_text
|
214 |
-
|
215 |
-
return JSONResponse(content=create_complete_response(full_text, model))
|
216 |
-
|
217 |
-
def create_chunk(text: str, model: str) -> dict:
|
218 |
-
"""Create a streaming chunk response"""
|
219 |
-
return {
|
220 |
-
"id": "chatcmpl-123",
|
221 |
-
"object": "chat.completion.chunk",
|
222 |
-
"created": int(time.time()),
|
223 |
-
"model": model,
|
224 |
-
"choices": [
|
225 |
-
{
|
226 |
-
"index": 0,
|
227 |
-
"delta": {"content": text},
|
228 |
-
"finish_reason": None,
|
229 |
-
},
|
230 |
-
],
|
231 |
-
}
|
232 |
-
|
233 |
-
def create_stop_chunk(model: str) -> dict:
|
234 |
-
"""Create a final streaming chunk with stop reason"""
|
235 |
-
return {
|
236 |
-
"id": "chatcmpl-123",
|
237 |
-
"object": "chat.completion.chunk",
|
238 |
-
"created": int(time.time()),
|
239 |
-
"model": model,
|
240 |
-
"choices": [
|
241 |
-
{
|
242 |
-
"index": 0,
|
243 |
-
"delta": {},
|
244 |
-
"finish_reason": "stop",
|
245 |
-
},
|
246 |
-
],
|
247 |
-
}
|
248 |
-
|
249 |
-
def create_complete_response(text: str, model: str) -> dict:
|
250 |
-
"""Create a complete non-streaming response"""
|
251 |
-
return {
|
252 |
-
"id": "chatcmpl-123",
|
253 |
-
"object": "chat.completion",
|
254 |
-
"created": int(time.time()),
|
255 |
-
"model": model,
|
256 |
-
"usage": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0},
|
257 |
-
"choices": [
|
258 |
-
{
|
259 |
-
"message": {"content": text, "role": "assistant"},
|
260 |
-
"index": 0,
|
261 |
-
"finish_reason": "stop",
|
262 |
-
},
|
263 |
-
],
|
264 |
-
}
|
265 |
-
|
266 |
-
# Only needed for retry delays
|
267 |
-
import asyncio
|
268 |
|
269 |
-
if __name__ ==
|
270 |
-
|
271 |
-
uvicorn.run("app:app", host="0.0.0.0", port=7860, reload=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import os
|
2 |
+
from g4f.client import Client
|
3 |
+
from flask import Flask, jsonify, request
|
|
|
4 |
from dotenv import load_dotenv
|
5 |
+
from flask_cors import CORS
|
6 |
|
7 |
load_dotenv()
|
8 |
+
app = Flask(__name__)
|
9 |
+
client = Client()
|
10 |
|
11 |
+
port = 7860
|
12 |
+
CORS(app)
|
|
|
|
|
13 |
|
14 |
+
@app.route('/', methods=['GET'])
|
15 |
+
def getMain():
|
16 |
+
return 'g4f-proxy is running'
|
|
|
|
|
|
|
|
|
|
|
17 |
|
18 |
+
@app.route('/', methods=['POST'])
|
19 |
+
def postG4F():
|
20 |
+
if not request.is_json:
|
21 |
+
return jsonify({"error": "Request must be JSON"}), 400
|
22 |
|
23 |
+
data = request.get_json()
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
|
25 |
+
# Required attributes
|
26 |
+
required_fields = ["system", "user"]
|
|
|
|
|
27 |
|
28 |
+
# Check for missing fields
|
29 |
+
missing_fields = [field for field in required_fields if field not in data]
|
|
|
|
|
30 |
|
31 |
+
if missing_fields:
|
32 |
+
return jsonify({"error": f"Missing fields: {', '.join(missing_fields)}"}), 400
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
+
response = client.chat.completions.create(
|
35 |
+
model="gpt-4o-mini",
|
36 |
+
messages=[
|
37 |
+
{"role": "system", "content": data['system']},
|
38 |
+
{"role": "user", "content": data['user']}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
],
|
40 |
+
web_search=False
|
41 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
|
43 |
+
return response.choices[0].message.content
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
|
45 |
+
if __name__ == '__main__':
|
46 |
+
app.run(debug=True, port=7860)
|
|