Update app.py
Browse files
app.py
CHANGED
|
@@ -1,68 +1,42 @@
|
|
| 1 |
import random
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
import asyncio
|
| 3 |
import json
|
| 4 |
import time
|
| 5 |
import uuid
|
| 6 |
import logging
|
| 7 |
|
| 8 |
-
from typing import List, Optional
|
| 9 |
-
|
| 10 |
import g4f
|
| 11 |
from g4f.client import Client
|
| 12 |
|
| 13 |
-
from fastapi import FastAPI, Request, HTTPException
|
| 14 |
-
from fastapi.middleware.cors import CORSMiddleware
|
| 15 |
-
from fastapi.responses import StreamingResponse, JSONResponse
|
| 16 |
-
from pydantic import BaseModel
|
| 17 |
-
|
| 18 |
-
# =====================================================
|
| 19 |
-
# SAFE PROVIDERS
|
| 20 |
-
# =====================================================
|
| 21 |
-
|
| 22 |
-
SAFE_PROVIDER_NAMES = [
|
| 23 |
-
"DeepInfra",
|
| 24 |
-
"DDG",
|
| 25 |
-
"Cloudflare",
|
| 26 |
-
]
|
| 27 |
|
| 28 |
# =====================================================
|
| 29 |
# LOGGING
|
| 30 |
# =====================================================
|
| 31 |
|
| 32 |
-
logging.basicConfig(
|
| 33 |
-
level=logging.INFO,
|
| 34 |
-
format="%(asctime)s [%(levelname)s] %(message)s"
|
| 35 |
-
)
|
| 36 |
-
|
| 37 |
logger = logging.getLogger(__name__)
|
| 38 |
|
|
|
|
| 39 |
# =====================================================
|
| 40 |
# CONFIG
|
| 41 |
# =====================================================
|
| 42 |
|
| 43 |
API_KEY = "sk-your-secret-key"
|
| 44 |
|
| 45 |
-
#
|
| 46 |
-
|
| 47 |
-
# =====================================================
|
| 48 |
-
|
| 49 |
-
AVAILABLE_PROVIDERS = []
|
| 50 |
-
|
| 51 |
-
for provider_name in SAFE_PROVIDER_NAMES:
|
| 52 |
-
|
| 53 |
-
try:
|
| 54 |
-
|
| 55 |
-
provider = getattr(g4f.Provider, provider_name)
|
| 56 |
-
|
| 57 |
-
AVAILABLE_PROVIDERS.append(provider)
|
| 58 |
|
| 59 |
-
|
|
|
|
| 60 |
|
| 61 |
-
except Exception:
|
| 62 |
-
|
| 63 |
-
logger.warning(
|
| 64 |
-
f"Provider not available: {provider_name}"
|
| 65 |
-
)
|
| 66 |
|
| 67 |
# =====================================================
|
| 68 |
# FASTAPI
|
|
@@ -70,9 +44,10 @@ for provider_name in SAFE_PROVIDER_NAMES:
|
|
| 70 |
|
| 71 |
app = FastAPI(
|
| 72 |
title="Universal AI Gateway",
|
| 73 |
-
version="
|
| 74 |
)
|
| 75 |
|
|
|
|
| 76 |
# =====================================================
|
| 77 |
# CORS
|
| 78 |
# =====================================================
|
|
@@ -85,6 +60,7 @@ app.add_middleware(
|
|
| 85 |
allow_headers=["*"],
|
| 86 |
)
|
| 87 |
|
|
|
|
| 88 |
# =====================================================
|
| 89 |
# MODELS
|
| 90 |
# =====================================================
|
|
@@ -101,6 +77,7 @@ class ChatRequest(BaseModel):
|
|
| 101 |
temperature: Optional[float] = 0.7
|
| 102 |
max_tokens: Optional[int] = 4096
|
| 103 |
|
|
|
|
| 104 |
# =====================================================
|
| 105 |
# AUTH
|
| 106 |
# =====================================================
|
|
@@ -109,6 +86,7 @@ def verify_api_key(req: Request):
|
|
| 109 |
|
| 110 |
auth = req.headers.get("Authorization")
|
| 111 |
|
|
|
|
| 112 |
if not auth:
|
| 113 |
return True
|
| 114 |
|
|
@@ -128,6 +106,7 @@ def verify_api_key(req: Request):
|
|
| 128 |
|
| 129 |
return True
|
| 130 |
|
|
|
|
| 131 |
# =====================================================
|
| 132 |
# ROOT
|
| 133 |
# =====================================================
|
|
@@ -138,13 +117,10 @@ async def root():
|
|
| 138 |
return {
|
| 139 |
"status": "online",
|
| 140 |
"service": "Universal AI Gateway",
|
| 141 |
-
"version": "
|
| 142 |
-
"providers": [
|
| 143 |
-
p.__name__
|
| 144 |
-
for p in AVAILABLE_PROVIDERS
|
| 145 |
-
]
|
| 146 |
}
|
| 147 |
|
|
|
|
| 148 |
# =====================================================
|
| 149 |
# MODELS
|
| 150 |
# =====================================================
|
|
@@ -185,6 +161,7 @@ async def get_models():
|
|
| 185 |
|
| 186 |
logger.error(f"Models error: {e}")
|
| 187 |
|
|
|
|
| 188 |
if not models_data:
|
| 189 |
|
| 190 |
for model in fallback_models:
|
|
@@ -201,22 +178,12 @@ async def get_models():
|
|
| 201 |
"data": models_data
|
| 202 |
}
|
| 203 |
|
| 204 |
-
# =====================================================
|
| 205 |
-
# PROVIDER PICKER
|
| 206 |
-
# =====================================================
|
| 207 |
-
|
| 208 |
-
def pick_provider():
|
| 209 |
-
|
| 210 |
-
if not AVAILABLE_PROVIDERS:
|
| 211 |
-
return None
|
| 212 |
-
|
| 213 |
-
return random.choice(AVAILABLE_PROVIDERS)
|
| 214 |
|
| 215 |
# =====================================================
|
| 216 |
-
#
|
| 217 |
# =====================================================
|
| 218 |
|
| 219 |
-
async def
|
| 220 |
model,
|
| 221 |
messages,
|
| 222 |
stream=False
|
|
@@ -224,33 +191,25 @@ async def create_completion(
|
|
| 224 |
|
| 225 |
last_error = None
|
| 226 |
|
| 227 |
-
|
| 228 |
-
|
| 229 |
-
random.shuffle(providers)
|
| 230 |
-
|
| 231 |
-
for provider in providers:
|
| 232 |
|
| 233 |
try:
|
| 234 |
|
| 235 |
logger.info(
|
| 236 |
-
f"
|
| 237 |
)
|
| 238 |
|
| 239 |
client = Client()
|
| 240 |
|
|
|
|
| 241 |
response = await asyncio.wait_for(
|
| 242 |
asyncio.to_thread(
|
| 243 |
client.chat.completions.create,
|
| 244 |
model=model,
|
| 245 |
messages=messages,
|
| 246 |
-
provider=provider,
|
| 247 |
stream=stream
|
| 248 |
),
|
| 249 |
-
timeout=
|
| 250 |
-
)
|
| 251 |
-
|
| 252 |
-
logger.info(
|
| 253 |
-
f"Success provider: {provider.__name__}"
|
| 254 |
)
|
| 255 |
|
| 256 |
return response
|
|
@@ -260,14 +219,13 @@ async def create_completion(
|
|
| 260 |
last_error = e
|
| 261 |
|
| 262 |
logger.warning(
|
| 263 |
-
f"
|
| 264 |
)
|
| 265 |
|
| 266 |
await asyncio.sleep(1)
|
| 267 |
|
| 268 |
-
raise Exception(
|
| 269 |
-
|
| 270 |
-
)
|
| 271 |
|
| 272 |
# =====================================================
|
| 273 |
# CHAT COMPLETIONS
|
|
@@ -303,7 +261,7 @@ async def chat_completions(
|
|
| 303 |
|
| 304 |
try:
|
| 305 |
|
| 306 |
-
response = await
|
| 307 |
model=body.model,
|
| 308 |
messages=messages,
|
| 309 |
stream=True
|
|
@@ -311,6 +269,8 @@ async def chat_completions(
|
|
| 311 |
|
| 312 |
chunk_id = f"chatcmpl-{uuid.uuid4().hex}"
|
| 313 |
|
|
|
|
|
|
|
| 314 |
for chunk in response:
|
| 315 |
|
| 316 |
try:
|
|
@@ -327,6 +287,8 @@ async def chat_completions(
|
|
| 327 |
|
| 328 |
if content:
|
| 329 |
|
|
|
|
|
|
|
| 330 |
payload = {
|
| 331 |
"id": chunk_id,
|
| 332 |
"object": "chat.completion.chunk",
|
|
@@ -356,6 +318,21 @@ async def chat_completions(
|
|
| 356 |
f"Chunk error: {chunk_error}"
|
| 357 |
)
|
| 358 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 359 |
final_payload = {
|
| 360 |
"id": chunk_id,
|
| 361 |
"object": "chat.completion.chunk",
|
|
@@ -377,6 +354,22 @@ async def chat_completions(
|
|
| 377 |
|
| 378 |
yield "data: [DONE]\n\n"
|
| 379 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 380 |
except Exception as e:
|
| 381 |
|
| 382 |
logger.error(f"Streaming error: {e}")
|
|
@@ -409,7 +402,7 @@ async def chat_completions(
|
|
| 409 |
|
| 410 |
try:
|
| 411 |
|
| 412 |
-
response = await
|
| 413 |
model=body.model,
|
| 414 |
messages=messages,
|
| 415 |
stream=False
|
|
@@ -449,6 +442,15 @@ async def chat_completions(
|
|
| 449 |
}
|
| 450 |
})
|
| 451 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 452 |
except Exception as e:
|
| 453 |
|
| 454 |
logger.error(f"Chat error: {e}")
|
|
@@ -458,6 +460,7 @@ async def chat_completions(
|
|
| 458 |
detail=str(e)
|
| 459 |
)
|
| 460 |
|
|
|
|
| 461 |
# =====================================================
|
| 462 |
# RUN
|
| 463 |
# =====================================================
|
|
@@ -466,8 +469,6 @@ if __name__ == "__main__":
|
|
| 466 |
|
| 467 |
import uvicorn
|
| 468 |
|
| 469 |
-
logger.info("Starting Universal AI Gateway")
|
| 470 |
-
|
| 471 |
uvicorn.run(
|
| 472 |
app,
|
| 473 |
host="0.0.0.0",
|
|
|
|
| 1 |
import random
|
| 2 |
+
import httpx
|
| 3 |
+
|
| 4 |
+
from fastapi import FastAPI, Request, HTTPException
|
| 5 |
+
from fastapi.middleware.cors import CORSMiddleware
|
| 6 |
+
from fastapi.responses import StreamingResponse, JSONResponse
|
| 7 |
+
from pydantic import BaseModel
|
| 8 |
+
from typing import List, Optional
|
| 9 |
+
|
| 10 |
import asyncio
|
| 11 |
import json
|
| 12 |
import time
|
| 13 |
import uuid
|
| 14 |
import logging
|
| 15 |
|
|
|
|
|
|
|
| 16 |
import g4f
|
| 17 |
from g4f.client import Client
|
| 18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
# =====================================================
|
| 21 |
# LOGGING
|
| 22 |
# =====================================================
|
| 23 |
|
| 24 |
+
logging.basicConfig(level=logging.INFO)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
logger = logging.getLogger(__name__)
|
| 26 |
|
| 27 |
+
|
| 28 |
# =====================================================
|
| 29 |
# CONFIG
|
| 30 |
# =====================================================
|
| 31 |
|
| 32 |
API_KEY = "sk-your-secret-key"
|
| 33 |
|
| 34 |
+
# Timeout لمنع التعليق
|
| 35 |
+
REQUEST_TIMEOUT = 40
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
|
| 37 |
+
# عدد المحاولات
|
| 38 |
+
MAX_RETRIES = 3
|
| 39 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
# =====================================================
|
| 42 |
# FASTAPI
|
|
|
|
| 44 |
|
| 45 |
app = FastAPI(
|
| 46 |
title="Universal AI Gateway",
|
| 47 |
+
version="4.1.0"
|
| 48 |
)
|
| 49 |
|
| 50 |
+
|
| 51 |
# =====================================================
|
| 52 |
# CORS
|
| 53 |
# =====================================================
|
|
|
|
| 60 |
allow_headers=["*"],
|
| 61 |
)
|
| 62 |
|
| 63 |
+
|
| 64 |
# =====================================================
|
| 65 |
# MODELS
|
| 66 |
# =====================================================
|
|
|
|
| 77 |
temperature: Optional[float] = 0.7
|
| 78 |
max_tokens: Optional[int] = 4096
|
| 79 |
|
| 80 |
+
|
| 81 |
# =====================================================
|
| 82 |
# AUTH
|
| 83 |
# =====================================================
|
|
|
|
| 86 |
|
| 87 |
auth = req.headers.get("Authorization")
|
| 88 |
|
| 89 |
+
# السماح للاختبار
|
| 90 |
if not auth:
|
| 91 |
return True
|
| 92 |
|
|
|
|
| 106 |
|
| 107 |
return True
|
| 108 |
|
| 109 |
+
|
| 110 |
# =====================================================
|
| 111 |
# ROOT
|
| 112 |
# =====================================================
|
|
|
|
| 117 |
return {
|
| 118 |
"status": "online",
|
| 119 |
"service": "Universal AI Gateway",
|
| 120 |
+
"version": "4.1.0"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 121 |
}
|
| 122 |
|
| 123 |
+
|
| 124 |
# =====================================================
|
| 125 |
# MODELS
|
| 126 |
# =====================================================
|
|
|
|
| 161 |
|
| 162 |
logger.error(f"Models error: {e}")
|
| 163 |
|
| 164 |
+
# fallback
|
| 165 |
if not models_data:
|
| 166 |
|
| 167 |
for model in fallback_models:
|
|
|
|
| 178 |
"data": models_data
|
| 179 |
}
|
| 180 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 181 |
|
| 182 |
# =====================================================
|
| 183 |
+
# SAFE COMPLETION
|
| 184 |
# =====================================================
|
| 185 |
|
| 186 |
+
async def safe_completion(
|
| 187 |
model,
|
| 188 |
messages,
|
| 189 |
stream=False
|
|
|
|
| 191 |
|
| 192 |
last_error = None
|
| 193 |
|
| 194 |
+
for attempt in range(MAX_RETRIES):
|
|
|
|
|
|
|
|
|
|
|
|
|
| 195 |
|
| 196 |
try:
|
| 197 |
|
| 198 |
logger.info(
|
| 199 |
+
f"Attempt {attempt + 1} | model={model}"
|
| 200 |
)
|
| 201 |
|
| 202 |
client = Client()
|
| 203 |
|
| 204 |
+
# timeout لمنع التعليق
|
| 205 |
response = await asyncio.wait_for(
|
| 206 |
asyncio.to_thread(
|
| 207 |
client.chat.completions.create,
|
| 208 |
model=model,
|
| 209 |
messages=messages,
|
|
|
|
| 210 |
stream=stream
|
| 211 |
),
|
| 212 |
+
timeout=REQUEST_TIMEOUT
|
|
|
|
|
|
|
|
|
|
|
|
|
| 213 |
)
|
| 214 |
|
| 215 |
return response
|
|
|
|
| 219 |
last_error = e
|
| 220 |
|
| 221 |
logger.warning(
|
| 222 |
+
f"Attempt failed {attempt + 1}: {e}"
|
| 223 |
)
|
| 224 |
|
| 225 |
await asyncio.sleep(1)
|
| 226 |
|
| 227 |
+
raise Exception(last_error)
|
| 228 |
+
|
|
|
|
| 229 |
|
| 230 |
# =====================================================
|
| 231 |
# CHAT COMPLETIONS
|
|
|
|
| 261 |
|
| 262 |
try:
|
| 263 |
|
| 264 |
+
response = await safe_completion(
|
| 265 |
model=body.model,
|
| 266 |
messages=messages,
|
| 267 |
stream=True
|
|
|
|
| 269 |
|
| 270 |
chunk_id = f"chatcmpl-{uuid.uuid4().hex}"
|
| 271 |
|
| 272 |
+
chunk_count = 0
|
| 273 |
+
|
| 274 |
for chunk in response:
|
| 275 |
|
| 276 |
try:
|
|
|
|
| 287 |
|
| 288 |
if content:
|
| 289 |
|
| 290 |
+
chunk_count += 1
|
| 291 |
+
|
| 292 |
payload = {
|
| 293 |
"id": chunk_id,
|
| 294 |
"object": "chat.completion.chunk",
|
|
|
|
| 318 |
f"Chunk error: {chunk_error}"
|
| 319 |
)
|
| 320 |
|
| 321 |
+
# إذا لم يصل أي chunk
|
| 322 |
+
if chunk_count == 0:
|
| 323 |
+
|
| 324 |
+
error_payload = {
|
| 325 |
+
"error": {
|
| 326 |
+
"message": "Provider returned empty stream",
|
| 327 |
+
"type": "empty_stream"
|
| 328 |
+
}
|
| 329 |
+
}
|
| 330 |
+
|
| 331 |
+
yield (
|
| 332 |
+
f"data: "
|
| 333 |
+
f"{json.dumps(error_payload)}\n\n"
|
| 334 |
+
)
|
| 335 |
+
|
| 336 |
final_payload = {
|
| 337 |
"id": chunk_id,
|
| 338 |
"object": "chat.completion.chunk",
|
|
|
|
| 354 |
|
| 355 |
yield "data: [DONE]\n\n"
|
| 356 |
|
| 357 |
+
except asyncio.TimeoutError:
|
| 358 |
+
|
| 359 |
+
logger.error("Streaming timeout")
|
| 360 |
+
|
| 361 |
+
error_payload = {
|
| 362 |
+
"error": {
|
| 363 |
+
"message": "Request timeout",
|
| 364 |
+
"type": "timeout"
|
| 365 |
+
}
|
| 366 |
+
}
|
| 367 |
+
|
| 368 |
+
yield (
|
| 369 |
+
f"data: "
|
| 370 |
+
f"{json.dumps(error_payload)}\n\n"
|
| 371 |
+
)
|
| 372 |
+
|
| 373 |
except Exception as e:
|
| 374 |
|
| 375 |
logger.error(f"Streaming error: {e}")
|
|
|
|
| 402 |
|
| 403 |
try:
|
| 404 |
|
| 405 |
+
response = await safe_completion(
|
| 406 |
model=body.model,
|
| 407 |
messages=messages,
|
| 408 |
stream=False
|
|
|
|
| 442 |
}
|
| 443 |
})
|
| 444 |
|
| 445 |
+
except asyncio.TimeoutError:
|
| 446 |
+
|
| 447 |
+
logger.error("Request timeout")
|
| 448 |
+
|
| 449 |
+
raise HTTPException(
|
| 450 |
+
status_code=408,
|
| 451 |
+
detail="Request timeout"
|
| 452 |
+
)
|
| 453 |
+
|
| 454 |
except Exception as e:
|
| 455 |
|
| 456 |
logger.error(f"Chat error: {e}")
|
|
|
|
| 460 |
detail=str(e)
|
| 461 |
)
|
| 462 |
|
| 463 |
+
|
| 464 |
# =====================================================
|
| 465 |
# RUN
|
| 466 |
# =====================================================
|
|
|
|
| 469 |
|
| 470 |
import uvicorn
|
| 471 |
|
|
|
|
|
|
|
| 472 |
uvicorn.run(
|
| 473 |
app,
|
| 474 |
host="0.0.0.0",
|