181 lines
6.6 KiB
Python
Executable File
181 lines
6.6 KiB
Python
Executable File
#!/usr/bin/env python3
|
|
"""OpenAI-compatible key-pool proxy template.
|
|
|
|
Use this only with legitimate upstream API keys supplied by the user, e.g.
|
|
OPENAI_KEYS=sk-proj-xxx,sk-proj-yyy. Do not adapt it to ChatGPT web
|
|
session tokens, Auth0 access tokens, cookies, or bulk signup/token pooling.
|
|
|
|
Run:
|
|
OPENAI_KEYS=sk-proj-xxx PROXY_AUTH_TOKEN=local-token python openai_key_pool_proxy.py
|
|
|
|
Test:
|
|
curl http://127.0.0.1:8091/health
|
|
curl http://127.0.0.1:8091/v1/models -H 'Authorization: Bearer local-token'
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import json
|
|
import os
|
|
import threading
|
|
import time
|
|
import urllib.error
|
|
import urllib.request
|
|
from dataclasses import dataclass, field
|
|
from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
|
|
from typing import Any
|
|
|
|
UPSTREAM_BASE = os.getenv("UPSTREAM_BASE", "https://api.openai.com").rstrip("/")
|
|
PROXY_AUTH_TOKEN = os.getenv("PROXY_AUTH_TOKEN", "local-token")
|
|
|
|
|
|
def redact(key: str) -> str:
|
|
return key[:7] + "..." + key[-4:] if len(key) > 12 else key[:4] + "..."
|
|
|
|
|
|
@dataclass
|
|
class KeyState:
|
|
key: str
|
|
name: str
|
|
enabled: bool = True
|
|
last_status: int | None = None
|
|
last_error: str | None = None
|
|
ok_count: int = 0
|
|
fail_count: int = 0
|
|
last_used: float | None = None
|
|
disabled_reason: str | None = None
|
|
|
|
def public(self) -> dict[str, Any]:
|
|
return {
|
|
"name": self.name,
|
|
"key": redact(self.key),
|
|
"enabled": self.enabled,
|
|
"last_status": self.last_status,
|
|
"last_error": self.last_error,
|
|
"ok_count": self.ok_count,
|
|
"fail_count": self.fail_count,
|
|
"last_used": self.last_used,
|
|
"disabled_reason": self.disabled_reason,
|
|
}
|
|
|
|
|
|
@dataclass
|
|
class Pool:
|
|
keys: list[KeyState]
|
|
cursor: int = 0
|
|
lock: threading.Lock = field(default_factory=threading.Lock)
|
|
|
|
def choose(self) -> KeyState | None:
|
|
with self.lock:
|
|
live = [k for k in self.keys if k.enabled]
|
|
if not live:
|
|
return None
|
|
item = live[self.cursor % len(live)]
|
|
self.cursor += 1
|
|
item.last_used = time.time()
|
|
return item
|
|
|
|
def mark(self, item: KeyState, status: int, body: bytes) -> None:
|
|
text = body[:1200].decode("utf-8", errors="ignore")
|
|
with self.lock:
|
|
item.last_status = status
|
|
item.last_error = None if 200 <= status < 300 else text
|
|
if 200 <= status < 300:
|
|
item.ok_count += 1
|
|
return
|
|
item.fail_count += 1
|
|
if status in (401, 403):
|
|
item.enabled = False
|
|
item.disabled_reason = f"HTTP {status}"
|
|
elif status == 429 and "billing_not_active" in text:
|
|
item.enabled = False
|
|
item.disabled_reason = "billing_not_active"
|
|
|
|
def public(self) -> dict[str, Any]:
|
|
with self.lock:
|
|
return {"total": len(self.keys), "enabled": sum(k.enabled for k in self.keys), "keys": [k.public() for k in self.keys]}
|
|
|
|
|
|
def load_keys() -> Pool:
|
|
raw = os.getenv("OPENAI_KEYS", "").replace("\n", ",")
|
|
seen: set[str] = set()
|
|
items: list[KeyState] = []
|
|
for idx, key in enumerate([x.strip() for x in raw.split(",") if x.strip()], 1):
|
|
if key in seen:
|
|
continue
|
|
seen.add(key)
|
|
items.append(KeyState(key=key, name=f"key{idx}"))
|
|
return Pool(items)
|
|
|
|
|
|
POOL = load_keys()
|
|
|
|
|
|
class Handler(BaseHTTPRequestHandler):
|
|
def _json(self, status: int, obj: Any) -> None:
|
|
body = json.dumps(obj, ensure_ascii=False, indent=2).encode()
|
|
self.send_response(status)
|
|
self.send_header("Content-Type", "application/json")
|
|
self.send_header("Content-Length", str(len(body)))
|
|
self.end_headers()
|
|
self.wfile.write(body)
|
|
|
|
def _authorized(self) -> bool:
|
|
return not PROXY_AUTH_TOKEN or self.headers.get("Authorization") == f"Bearer {PROXY_AUTH_TOKEN}"
|
|
|
|
def _body(self) -> bytes:
|
|
return self.rfile.read(int(self.headers.get("Content-Length") or 0))
|
|
|
|
def do_GET(self) -> None: # noqa: N802
|
|
if self.path in ("/", "/health"):
|
|
return self._json(200, {"ok": True, "upstream": UPSTREAM_BASE, "pool": POOL.public()})
|
|
return self._proxy()
|
|
|
|
def do_POST(self) -> None: # noqa: N802
|
|
return self._proxy()
|
|
|
|
def _proxy(self) -> None:
|
|
if not self.path.startswith("/v1/"):
|
|
return self._json(404, {"error": "only /v1/* is proxied"})
|
|
if not self._authorized():
|
|
return self._json(401, {"error": "proxy auth required"})
|
|
body = self._body()
|
|
attempts = max(1, min(3, len([k for k in POOL.keys if k.enabled]) or 1))
|
|
last_status, last_body, last_headers = 503, b'{"error":"no usable key"}', {"Content-Type": "application/json"}
|
|
for _ in range(attempts):
|
|
item = POOL.choose()
|
|
if not item:
|
|
break
|
|
status, resp_body, resp_headers = call_upstream(self.command, self.path, self.headers, body, item.key)
|
|
POOL.mark(item, status, resp_body)
|
|
last_status, last_body, last_headers = status, resp_body, resp_headers
|
|
if status < 400 or status in (400, 404, 422):
|
|
break
|
|
if status not in (401, 403, 429, 500, 502, 503, 504):
|
|
break
|
|
self.send_response(last_status)
|
|
self.send_header("Content-Type", last_headers.get("Content-Type", "application/json"))
|
|
self.send_header("Content-Length", str(len(last_body)))
|
|
self.end_headers()
|
|
self.wfile.write(last_body)
|
|
|
|
|
|
def call_upstream(method: str, path: str, incoming_headers: Any, body: bytes, upstream_key: str) -> tuple[int, bytes, dict[str, str]]:
|
|
headers = {k: v for k, v in incoming_headers.items() if k.lower() not in {"host", "authorization", "content-length", "connection"}}
|
|
headers["Authorization"] = f"Bearer {upstream_key}"
|
|
if body and "Content-Type" not in headers:
|
|
headers["Content-Type"] = "application/json"
|
|
req = urllib.request.Request(UPSTREAM_BASE + path, data=body if method != "GET" else None, headers=headers, method=method)
|
|
try:
|
|
with urllib.request.urlopen(req, timeout=120) as resp:
|
|
return resp.status, resp.read(), dict(resp.headers.items())
|
|
except urllib.error.HTTPError as e:
|
|
return e.code, e.read(), dict(e.headers.items())
|
|
except Exception as e:
|
|
return 502, json.dumps({"error": str(e)}, ensure_ascii=False).encode(), {"Content-Type": "application/json"}
|
|
|
|
|
|
if __name__ == "__main__":
|
|
print(f"proxy=http://127.0.0.1:8091 upstream={UPSTREAM_BASE} keys={len(POOL.keys)}")
|
|
ThreadingHTTPServer(("127.0.0.1", 8091), Handler).serve_forever()
|