chore: remove noisy healthcheck messages

.gitignore

@@ -2,3 +2,4 @@
 __pycache__/
 .ruff_cache/
 .venv/
+.pytest_cache/

Dockerfile

@@ -28,6 +28,9 @@ VOLUME ["/data"]
 
 EXPOSE 80
 
+HEALTHCHECK --start-period=5s --start-interval=1s CMD \
+    test "$(curl -fsS "http://127.0.0.1:$PORT/health")" = "ok"
+
 STOPSIGNAL SIGINT
 
 CMD ["/entrypoint.sh"]

README.md

@@ -16,7 +16,7 @@ Response shape:
     "used_percent": 0,
     "remaining_percent": 100,
     "exhausted": false,
-    "needs_refresh": false
+    "needs_prepare": false
   },
   "usage": {
     "primary_window": {
@@ -52,12 +52,15 @@ Behavior:
 4. When usage reaches `CHATGPT_PREPARE_THRESHOLD`, service prepares `next_account`.
 5. When usage reaches `CHATGPT_SWITCH_THRESHOLD`, service switches active account to `next_account`.
 
+## Disposable Email Provider
+
+- Default provider is `mail.tm` API (`MailTmProvider`) and does not use browser automation.
+- Flow: fetch domains -> create account with random address/password -> get JWT token -> poll messages.
+
 ## Startup Behavior
 
-On startup, service:
-
-1. Ensures active token exists and is usable.
-2. Ensures `next_account` is prepared for ChatGPT.
+On startup, service ensures active token exists and is usable.
+Standby preparation runs through provider lifecycle hooks/background trigger when needed.
 
 ## Data Files
 
@@ -70,6 +73,14 @@ On startup, service:
 PYTHONPATH=./src python src/server.py
 ```
 
+## Unit Tests
+
+The project has unit tests only (no integration/network tests).
+
+```bash
+pytest -q
+```
+
 ## Docker Notes
 
 - Dockerfile sets `DATA_DIR=/data`.

pyproject.toml

@@ -8,5 +8,10 @@ dependencies = [
     "pkce==1.0.3",
 ]
 
+[project.optional-dependencies]
+dev = [
+    "pytest>=8.0.0",
+]
+
 [tool.uv]
 package = false

scripts/run_token_refresh_flow.py

@@ -0,0 +1,55 @@
+import argparse
+import asyncio
+import json
+import logging
+
+import browser
+import server
+
+
+class _FakeRequest:
+    def __init__(self, provider: str):
+        self.match_info = {"provider": provider}
+        self.method = "GET"
+        self.path_qs = f"/{provider}/token"
+
+
+def _enable_headed_browser() -> bool:
+    if "--no-startup-window" in browser.CHROME_FLAGS:
+        browser.CHROME_FLAGS.remove("--no-startup-window")
+        return True
+    return False
+
+
+async def _run(provider: str) -> int:
+    patched = _enable_headed_browser()
+    logging.info("Headed mode patch applied: %s", patched)
+
+    request = _FakeRequest(provider)
+    response = await server.token_handler(request)
+    payload = json.loads(response.body.decode("utf-8"))
+
+    logging.info("Response status: %s", response.status)
+    logging.info("Response body: %s", json.dumps(payload, indent=2))
+    return 0 if response.status == 200 else 1
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(
+        description=(
+            "Run the same token refresh/issue flow as server /{provider}/token "
+            "in headed browser mode (non-headless)."
+        )
+    )
+    parser.add_argument("--provider", default="chatgpt")
+    args = parser.parse_args()
+
+    logging.basicConfig(
+        level=logging.INFO,
+        format="%(asctime)s %(levelname)s %(name)s: %(message)s",
+    )
+    return asyncio.run(_run(args.provider))
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

src/browser.py

@@ -1,6 +1,7 @@
 import asyncio
 import json
 import logging
+import socket
 import shutil
 import subprocess
 import tempfile
@@ -44,7 +45,12 @@ CHROME_FLAGS = [
     "--disable-search-engine-choice-screen",
 ]
 
-DEFAULT_CDP_PORT = 9222
+
+def _allocate_free_port() -> int:
+    with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+        s.bind(("127.0.0.1", 0))
+        s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+        return int(s.getsockname()[1])
 
 
 def _fetch_ws_endpoint(port: int) -> str | None:
@@ -79,10 +85,9 @@ class ManagedBrowser:
             shutil.rmtree(self.profile_dir, ignore_errors=True)
 
 
-async def launch(
-    playwright: Playwright, cdp_port: int = DEFAULT_CDP_PORT
-) -> ManagedBrowser:
+async def launch(playwright: Playwright) -> ManagedBrowser:
     chrome_path = playwright.chromium.executable_path
+    cdp_port = _allocate_free_port()
     profile_dir = Path(tempfile.mkdtemp(prefix="megapt_profile-", dir="/tmp"))
 
     args = [

src/email_providers/__init__.py

@@ -1,5 +1,11 @@
 from .base import BaseProvider
+from .mail_tm import MailTmProvider
 from .ten_minute_mail import TenMinuteMailProvider
 from .temp_mail_org import TempMailOrgProvider
 
-__all__ = ["BaseProvider", "TenMinuteMailProvider", "TempMailOrgProvider"]
+__all__ = [
+    "BaseProvider",
+    "MailTmProvider",
+    "TenMinuteMailProvider",
+    "TempMailOrgProvider",
+]

src/email_providers/base.py

@@ -12,5 +12,5 @@ class BaseProvider(ABC):
         pass
 
     @abstractmethod
-    async def get_latest_message(self, email: str) -> str | None:
+    async def get_latest_message(self) -> str | None:
         pass

src/email_providers/mail_tm.py

@@ -0,0 +1,227 @@
+import asyncio
+import logging
+import os
+import secrets
+import string
+from typing import Any
+
+import aiohttp
+from playwright.async_api import BrowserContext
+
+from .base import BaseProvider
+from utils.randoms import generate_password
+
+logger = logging.getLogger(__name__)
+
+_API_BASE = os.environ.get("MAIL_TM_API_BASE", "https://api.mail.tm")
+_TIMEOUT_SECONDS = 20
+_FIRST_NAMES = [
+    "james",
+    "john",
+    "robert",
+    "michael",
+    "david",
+    "william",
+    "joseph",
+    "thomas",
+    "daniel",
+    "mark",
+    "paul",
+    "kevin",
+]
+_LAST_NAMES = [
+    "smith",
+    "johnson",
+    "williams",
+    "brown",
+    "jones",
+    "miller",
+    "davis",
+    "wilson",
+    "anderson",
+    "taylor",
+    "martin",
+    "thompson",
+]
+
+
+def _generate_local_part() -> str:
+    first = secrets.choice(_FIRST_NAMES)
+    last = secrets.choice(_LAST_NAMES)
+    digits = "".join(secrets.choice(string.digits) for _ in range(8))
+    return f"{first}{last}{digits}"
+
+
+class MailTmProvider(BaseProvider):
+    def __init__(self, browser_session: BrowserContext):
+        super().__init__(browser_session)
+        self._address: str | None = None
+        self._password: str | None = None
+        self._token: str | None = None
+
+    async def _request(
+        self,
+        method: str,
+        path: str,
+        *,
+        token: str | None = None,
+        json_body: dict[str, Any] | None = None,
+    ) -> tuple[int, dict[str, Any] | list[Any] | None]:
+        url = f"{_API_BASE.rstrip('/')}{path}"
+        headers: dict[str, str] = {}
+        if token:
+            headers["Authorization"] = f"Bearer {token}"
+
+        timeout = aiohttp.ClientTimeout(total=_TIMEOUT_SECONDS)
+        try:
+            async with aiohttp.ClientSession(timeout=timeout) as session:
+                async with session.request(
+                    method,
+                    url,
+                    headers=headers,
+                    json=json_body,
+                ) as resp:
+                    status = resp.status
+                    try:
+                        payload = await resp.json()
+                    except aiohttp.ContentTypeError:
+                        payload = None
+                    return status, payload
+        except aiohttp.ClientError as e:
+            logger.warning("[mail.tm] request failed %s %s: %s", method, path, e)
+            return 0, None
+
+    async def _get_domains(self) -> list[str]:
+        status, payload = await self._request("GET", "/domains")
+        if status != 200 or not isinstance(payload, dict):
+            raise RuntimeError("mail.tm domains request failed")
+
+        members = payload.get("hydra:member")
+        if not isinstance(members, list):
+            raise RuntimeError("mail.tm domains response has unexpected format")
+
+        domains: list[str] = []
+        for item in members:
+            if not isinstance(item, dict):
+                continue
+            domain = item.get("domain")
+            is_active = bool(item.get("isActive", True))
+            if isinstance(domain, str) and domain and is_active:
+                domains.append(domain)
+
+        if not domains:
+            raise RuntimeError("mail.tm returned no active domains")
+        return domains
+
+    async def _create_account(self, address: str, password: str) -> bool:
+        status, _ = await self._request(
+            "POST",
+            "/accounts",
+            json_body={"address": address, "password": password},
+        )
+        if status in (200, 201):
+            return True
+        return False
+
+    async def _create_token(self, address: str, password: str) -> str | None:
+        status, payload = await self._request(
+            "POST",
+            "/token",
+            json_body={"address": address, "password": password},
+        )
+        if status != 200 or not isinstance(payload, dict):
+            return None
+        token = payload.get("token")
+        if isinstance(token, str) and token:
+            return token
+        return None
+
+    async def get_new_email(self) -> str:
+        domains = await self._get_domains()
+
+        for _ in range(8):
+            domain = secrets.choice(domains)
+            address = f"{_generate_local_part()}@{domain}"
+            password = generate_password(length=24)
+
+            created = await self._create_account(address, password)
+            if not created:
+                continue
+
+            token = await self._create_token(address, password)
+            if not token:
+                continue
+
+            self._address = address
+            self._password = password
+            self._token = token
+            logger.info("[mail.tm] New mailbox acquired: %s", address)
+            return address
+
+        raise RuntimeError("mail.tm could not create account")
+
+    async def _list_messages(self) -> list[dict[str, Any]]:
+        if not self._token:
+            return []
+        status, payload = await self._request(
+            "GET",
+            "/messages",
+            token=self._token,
+        )
+        if status == 401 and self._address and self._password:
+            token = await self._create_token(self._address, self._password)
+            if token:
+                self._token = token
+                status, payload = await self._request(
+                    "GET",
+                    "/messages",
+                    token=self._token,
+                )
+
+        if status != 200 or not isinstance(payload, dict):
+            return []
+
+        members = payload.get("hydra:member")
+        if not isinstance(members, list):
+            return []
+        return [item for item in members if isinstance(item, dict)]
+
+    async def _get_message_text(self, message_id: str) -> str | None:
+        if not self._token:
+            return None
+        status, payload = await self._request(
+            "GET",
+            f"/messages/{message_id}",
+            token=self._token,
+        )
+        if status != 200 or not isinstance(payload, dict):
+            return None
+
+        parts = [
+            payload.get("subject"),
+            payload.get("intro"),
+            payload.get("text"),
+            payload.get("html"),
+        ]
+        text = "\n".join(str(part) for part in parts if part)
+        return text or None
+
+    async def get_latest_message(self) -> str | None:
+        if not self._token:
+            raise RuntimeError("mail.tm provider is not initialized with mailbox token")
+
+        for _ in range(45):
+            messages = await self._list_messages()
+            if messages:
+                latest = messages[0]
+                message_id = latest.get("id")
+                if isinstance(message_id, str) and message_id:
+                    full_message = await self._get_message_text(message_id)
+                    if full_message:
+                        logger.info("[mail.tm] Latest message received")
+                        return full_message
+
+            await asyncio.sleep(2)
+
+        logger.warning("[mail.tm] No messages received within timeout")
+        return None

src/email_providers/temp_mail_org.py

@@ -2,9 +2,10 @@ import asyncio
 import logging
 import re
 
-from playwright.async_api import BrowserContext, Page
+from playwright.async_api import BrowserContext, Error as PlaywrightError, Page
 
 from .base import BaseProvider
+from .utils import ensure_page
 
 logger = logging.getLogger(__name__)
 
@@ -15,8 +16,7 @@ class TempMailOrgProvider(BaseProvider):
         self.page: Page | None = None
 
     async def _ensure_page(self) -> Page:
-        if self.page is None or self.page.is_closed():
-            self.page = await self.browser_session.new_page()
+        self.page = await ensure_page(self.browser_session, self.page)
         return self.page
 
     async def get_new_email(self) -> str:
@@ -44,7 +44,7 @@ class TempMailOrgProvider(BaseProvider):
                                 value,
                             )
                             return value
-                except Exception:
+                except PlaywrightError:
                     continue
 
             try:
@@ -53,16 +53,16 @@ class TempMailOrgProvider(BaseProvider):
                 if found:
                     logger.info("[temp-mail.org] email found by body scan: %s", found)
                     return found
-            except Exception:
-                pass
+            except PlaywrightError:
+                logger.debug("Failed to scan body text for email")
 
             await asyncio.sleep(1)
 
         raise RuntimeError("Could not get temp email from temp-mail.org")
 
-    async def get_latest_message(self, email: str) -> str | None:
+    async def get_latest_message(self) -> str | None:
         page = await self._ensure_page()
-        logger.info("[temp-mail.org] Waiting for latest message for %s", email)
+        logger.info("[temp-mail.org] Waiting for latest message")
 
         if page.is_closed():
             raise RuntimeError("temp-mail.org tab was closed unexpectedly")
@@ -76,7 +76,7 @@ class TempMailOrgProvider(BaseProvider):
             try:
                 count = await items.count()
                 logger.info("[temp-mail.org] inbox items: %s", count)
-            except Exception:
+            except PlaywrightError:
                 count = 0
 
             if count > 0:
@@ -87,30 +87,30 @@ class TempMailOrgProvider(BaseProvider):
                             continue
                         text = (await item.inner_text()).strip().replace("\n", " ")
                         logger.info("[temp-mail.org] item[%s]: %s", idx, text[:160])
-                    except Exception:
+                    except PlaywrightError:
                         continue
                     if text:
                         try:
                             await item.click()
                             logger.info("[temp-mail.org] opened item[%s]", idx)
-                        except Exception:
-                            pass
+                        except PlaywrightError:
+                            logger.debug("Failed to open inbox item[%s]", idx)
 
                         message_text = text
                         try:
                             content = await page.content()
                             if content and "Your ChatGPT code is" in content:
                                 message_text = content
-                        except Exception:
-                            pass
+                        except PlaywrightError:
+                            logger.debug("Failed to read opened message content")
 
                         try:
                             await page.go_back(
                                 wait_until="domcontentloaded", timeout=5000
                             )
                             logger.info("[temp-mail.org] returned back to inbox")
-                        except Exception:
-                            pass
+                        except PlaywrightError:
+                            logger.debug("Failed to return back to inbox")
 
                         return message_text
 

src/email_providers/ten_minute_mail.py

@@ -4,6 +4,7 @@ import logging
 from playwright.async_api import BrowserContext, Page
 
 from .base import BaseProvider
+from .utils import ensure_page
 
 logger = logging.getLogger(__name__)
 
@@ -14,8 +15,7 @@ class TenMinuteMailProvider(BaseProvider):
         self.page: Page | None = None
 
     async def _ensure_page(self) -> Page:
-        if self.page is None or self.page.is_closed():
-            self.page = await self.browser_session.new_page()
+        self.page = await ensure_page(self.browser_session, self.page)
         return self.page
 
     async def get_new_email(self) -> str:
@@ -34,9 +34,9 @@ class TenMinuteMailProvider(BaseProvider):
         logger.info("[10min] New email acquired: %s", email)
         return email
 
-    async def get_latest_message(self, email: str) -> str | None:
+    async def get_latest_message(self) -> str | None:
         page = await self._ensure_page()
-        logger.info("[10min] Waiting for latest message for %s", email)
+        logger.info("[10min] Waiting for latest message")
 
         seen_count = 0
         for attempt in range(60):

src/email_providers/utils.py

@@ -0,0 +1,10 @@
+from playwright.async_api import BrowserContext, Page
+
+
+async def ensure_page(
+    browser_session: BrowserContext,
+    page: Page | None,
+) -> Page:
+    if page is None or page.is_closed():
+        return await browser_session.new_page()
+    return page

src/providers/base.py

@@ -52,3 +52,36 @@ class Provider(ABC):
     def save_tokens(self, tokens: ProviderTokens) -> None:
         """Save tokens to storage"""
         pass
+
+    async def force_recreate_token(self) -> str | None:
+        """Force-create a new active token when normal acquisition fails."""
+        return None
+
+    async def maybe_rotate_account(self, usage_percent: int) -> bool:
+        """Rotate active account/token if provider policy requires it."""
+        return False
+
+    @property
+    def prepare_threshold(self) -> int:
+        """Usage percent when provider should prepare standby account/token."""
+        return 100
+
+    @property
+    def switch_threshold(self) -> int | None:
+        """Usage percent when provider may switch active account/token."""
+        return None
+
+    def should_prepare_standby(self) -> bool:
+        """Whether standby preparation should be triggered for current usage."""
+        return False
+
+    async def ensure_standby_account(
+        self,
+        usage_percent: int,
+    ) -> None:
+        """Prepare standby account/token asynchronously when needed."""
+        return None
+
+    async def startup_prepare(self) -> None:
+        """Optional provider-specific startup preparation."""
+        return None

src/providers/chatgpt/provider.py

@@ -1,14 +1,14 @@
 import asyncio
 import logging
-import os
 from typing import Any
 from typing import Callable
 
 from playwright.async_api import BrowserContext
 
 from email_providers import BaseProvider
-from email_providers import TempMailOrgProvider
+from email_providers import MailTmProvider
 from providers.base import Provider, ProviderTokens
+from utils.env import parse_int_env
 from .tokens import (
     clear_next_tokens,
     load_next_tokens,
@@ -16,7 +16,7 @@ from .tokens import (
     load_tokens,
     promote_next_tokens,
     refresh_tokens,
-    save_next_tokens,
+    save_state,
     save_tokens,
 )
 from .usage import get_usage_data
@@ -24,7 +24,13 @@ from .registration import register_chatgpt_account
 
 logger = logging.getLogger(__name__)
 CHATGPT_REGISTRATION_MAX_ATTEMPTS = 4
-CHATGPT_SWITCH_THRESHOLD = int(os.environ.get("CHATGPT_SWITCH_THRESHOLD", "95"))
+CHATGPT_PREPARE_THRESHOLD = parse_int_env("CHATGPT_PREPARE_THRESHOLD", 85, 0, 100)
+CHATGPT_SWITCH_THRESHOLD = parse_int_env(
+    "CHATGPT_SWITCH_THRESHOLD",
+    95,
+    0,
+    100,
+)
 
 
 class ChatGPTProvider(Provider):
@@ -34,9 +40,17 @@ class ChatGPTProvider(Provider):
         self,
         email_provider_factory: Callable[[BrowserContext], BaseProvider] | None = None,
     ):
-        self.email_provider_factory = email_provider_factory or TempMailOrgProvider
+        self.email_provider_factory = email_provider_factory or MailTmProvider
         self._token_write_lock = asyncio.Lock()
 
+    @property
+    def prepare_threshold(self) -> int:
+        return CHATGPT_PREPARE_THRESHOLD
+
+    @property
+    def switch_threshold(self) -> int | None:
+        return CHATGPT_SWITCH_THRESHOLD
+
     async def _register_with_retries(self) -> bool:
         for attempt in range(1, CHATGPT_REGISTRATION_MAX_ATTEMPTS + 1):
             logger.info(
@@ -44,10 +58,14 @@ class ChatGPTProvider(Provider):
                 attempt,
                 CHATGPT_REGISTRATION_MAX_ATTEMPTS,
             )
-            success = await self.register_new_account()
-            if success:
+            generated_tokens = await register_chatgpt_account(
+                email_provider_factory=self.email_provider_factory,
+            )
+            if generated_tokens:
+                save_tokens(generated_tokens)
                 return True
             logger.warning("Registration attempt %s failed", attempt)
+            await asyncio.sleep(1.5 * attempt)
         return False
 
     async def _create_next_account_under_lock(self) -> bool:
@@ -56,23 +74,28 @@ class ChatGPTProvider(Provider):
             return True
 
         logger.info("Creating next account")
-        success = await self._register_with_retries()
-        if not success:
-            return False
+        for attempt in range(1, CHATGPT_REGISTRATION_MAX_ATTEMPTS + 1):
+            logger.info(
+                "Next-account registration attempt %s/%s",
+                attempt,
+                CHATGPT_REGISTRATION_MAX_ATTEMPTS,
+            )
+            generated_tokens = await register_chatgpt_account(
+                email_provider_factory=self.email_provider_factory,
+            )
+            if generated_tokens:
+                if active_before:
+                    save_state(active_before, generated_tokens)
+                else:
+                    save_state(generated_tokens, None)
+                logger.info("Next account is ready")
+                return True
+            logger.warning("Next-account registration attempt %s failed", attempt)
+            await asyncio.sleep(1.5 * attempt)
 
-        generated_active = load_tokens()
-        if not generated_active:
-            return False
-
-        # Registration writes new tokens as active; restore old active and keep
-        # generated account as next.
-        if active_before:
-            save_tokens(active_before)
-        else:
-            clear_next_tokens()
-        save_next_tokens(generated_active)
-        logger.info("Next account is ready")
-        return True
+        if active_before or next_before:
+            save_state(active_before, next_before)
+        return False
 
     async def force_recreate_token(self) -> str | None:
         async with self._token_write_lock:
@@ -85,17 +108,30 @@ class ChatGPTProvider(Provider):
                 return None
             return tokens.access_token
 
+    async def startup_prepare(self) -> None:
+        await self.ensure_next_account()
+
     async def ensure_next_account(self) -> bool:
         next_tokens = load_next_tokens()
-        if next_tokens and not next_tokens.is_expired:
+        if next_tokens:
             return True
 
         async with self._token_write_lock:
             next_tokens = load_next_tokens()
-            if next_tokens and not next_tokens.is_expired:
+            if next_tokens:
                 return True
             return await self._create_next_account_under_lock()
 
+    def should_prepare_standby(self) -> bool:
+        return bool(load_next_tokens())
+
+    async def ensure_standby_account(
+        self,
+        usage_percent: int,
+    ) -> None:
+        if usage_percent >= self.prepare_threshold:
+            await self.ensure_next_account()
+
     async def maybe_switch_active_account(self, usage_percent: int) -> bool:
         if usage_percent < CHATGPT_SWITCH_THRESHOLD:
             return False
@@ -119,6 +155,9 @@ class ChatGPTProvider(Provider):
                 )
             return switched
 
+    async def maybe_rotate_account(self, usage_percent: int) -> bool:
+        return await self.maybe_switch_active_account(usage_percent)
+
     @property
     def name(self) -> str:
         return "chatgpt"
@@ -154,13 +193,17 @@ class ChatGPTProvider(Provider):
 
     async def register_new_account(self) -> bool:
         """Register a new ChatGPT account"""
-        return await register_chatgpt_account(
+        generated_tokens = await register_chatgpt_account(
             email_provider_factory=self.email_provider_factory,
         )
+        if not generated_tokens:
+            return False
+        save_tokens(generated_tokens)
+        return True
 
     async def get_usage_info(self, access_token: str) -> dict[str, Any]:
         """Get usage information for the current token"""
-        usage_data = get_usage_data(access_token)
+        usage_data = await get_usage_data(access_token)
         if not usage_data:
             return {"error": "Failed to get usage"}
 

src/providers/chatgpt/registration.py

@@ -5,7 +5,6 @@ import logging
 import random
 import re
 import secrets
-import string
 import time
 from datetime import datetime
 from pathlib import Path
@@ -14,12 +13,18 @@ from typing import Callable
 from urllib.parse import parse_qs, urlencode, urlparse
 
 import aiohttp
-from playwright.async_api import async_playwright, Page, BrowserContext
+from playwright.async_api import (
+    async_playwright,
+    Error as PlaywrightError,
+    Page,
+    BrowserContext,
+)
 
 from browser import launch as launch_browser
 from email_providers import BaseProvider
 from providers.base import ProviderTokens
-from .tokens import CLIENT_ID, save_tokens
+from utils.randoms import generate_password
+from .tokens import CLIENT_ID
 
 logger = logging.getLogger(__name__)
 
@@ -46,14 +51,9 @@ async def save_error_screenshot(page: Page | None, step: str):
         filename = screenshots_dir / f"error_{step}_{timestamp}.png"
         try:
             await page.screenshot(path=str(filename))
-            logger.error(f"Screenshot saved: {filename}")
-        except:
-            pass
-
-
-def generate_password(length: int = 20) -> str:
-    alphabet = string.ascii_letters + string.digits
-    return "".join(random.choice(alphabet) for _ in range(length))
+            logger.error("Screenshot saved: %s", filename)
+        except PlaywrightError as e:
+            logger.warning("Failed to save screenshot at step %s: %s", step, e)
 
 
 def generate_name() -> str:
@@ -204,8 +204,7 @@ def generate_state() -> str:
     return secrets.token_urlsafe(32)
 
 
-def build_authorize_url(verifier: str, challenge: str, state: str) -> str:
-    del verifier
+def build_authorize_url(challenge: str, state: str) -> str:
     params = {
         "response_type": "code",
         "client_id": CLIENT_ID,
@@ -222,30 +221,37 @@ def build_authorize_url(verifier: str, challenge: str, state: str) -> str:
 
 
 async def exchange_code_for_tokens(code: str, verifier: str) -> ProviderTokens:
-    async with aiohttp.ClientSession() as session:
-        payload = {
-            "grant_type": "authorization_code",
-            "client_id": CLIENT_ID,
-            "code": code,
-            "code_verifier": verifier,
-            "redirect_uri": REDIRECT_URI,
-        }
-        async with session.post(TOKEN_URL, data=payload) as resp:
-            if not resp.ok:
-                text = await resp.text()
-                raise RuntimeError(f"Token exchange failed: {resp.status} {text}")
-            body = await resp.json()
+    payload = {
+        "grant_type": "authorization_code",
+        "client_id": CLIENT_ID,
+        "code": code,
+        "code_verifier": verifier,
+        "redirect_uri": REDIRECT_URI,
+    }
+    timeout = aiohttp.ClientTimeout(total=20)
+    try:
+        async with aiohttp.ClientSession(timeout=timeout) as session:
+            async with session.post(TOKEN_URL, data=payload) as resp:
+                if not resp.ok:
+                    text = await resp.text()
+                    raise RuntimeError(f"Token exchange failed: {resp.status} {text}")
+                body = await resp.json()
+    except (aiohttp.ClientError, TimeoutError) as e:
+        raise RuntimeError(f"Token exchange request error: {e}") from e
 
-    expires_in = int(body["expires_in"])
-    return ProviderTokens(
-        access_token=body["access_token"],
-        refresh_token=body["refresh_token"],
-        expires_at=time.time() + expires_in,
-    )
+    try:
+        expires_in = int(body["expires_in"])
+        return ProviderTokens(
+            access_token=body["access_token"],
+            refresh_token=body["refresh_token"],
+            expires_at=time.time() + expires_in,
+        )
+    except (KeyError, TypeError, ValueError) as e:
+        raise RuntimeError(f"Token exchange response parse error: {e}") from e
 
 
-async def get_latest_code(email_provider: BaseProvider, email: str) -> str | None:
-    message = await email_provider.get_latest_message(email)
+async def get_latest_code(email_provider: BaseProvider) -> str | None:
+    message = await email_provider.get_latest_message()
     if not message:
         return None
     return extract_verification_code(message)
@@ -270,6 +276,51 @@ async def click_continue(page: Page, timeout_ms: int = 10000):
     await btn.click()
 
 
+async def oauth_needs_email_check(page: Page) -> bool:
+    marker = page.get_by_text("Check your inbox", exact=False)
+    return await marker.count() > 0
+
+
+async def fill_oauth_code_if_present(page: Page, code: str) -> bool:
+    candidates = [
+        page.get_by_placeholder("Code"),
+        page.get_by_label("Code"),
+        page.locator(
+            'input[name*="code" i], input[id*="code" i], '
+            'input[autocomplete="one-time-code"], input[inputmode="numeric"]'
+        ),
+    ]
+
+    for locator in candidates:
+        if await locator.count() == 0:
+            continue
+        try:
+            await locator.first.wait_for(state="visible", timeout=1500)
+            await locator.first.fill(code)
+            return True
+        except PlaywrightError:
+            continue
+    return False
+
+
+async def click_first_visible_button(
+    page: Page,
+    labels: list[str],
+    timeout_ms: int = 2000,
+) -> bool:
+    for label in labels:
+        button = page.get_by_role("button", name=label)
+        if await button.count() == 0:
+            continue
+        try:
+            await button.first.wait_for(state="visible", timeout=timeout_ms)
+            await button.first.click(timeout=timeout_ms)
+            return True
+        except PlaywrightError:
+            continue
+    return False
+
+
 async def wait_for_signup_stabilization(
     page: Page,
     source_url: str,
@@ -288,12 +339,12 @@ async def wait_for_signup_stabilization(
 
 async def register_chatgpt_account(
     email_provider_factory: Callable[[BrowserContext], BaseProvider] | None = None,
-) -> bool:
+) -> ProviderTokens | None:
     logger.info("=== Starting ChatGPT account registration ===")
 
     if email_provider_factory is None:
         logger.error("No email provider factory configured")
-        return False
+        return None
 
     birth_month, birth_day, birth_year = generate_birthdate_90s()
 
@@ -321,7 +372,7 @@ async def register_chatgpt_account(
             full_name = generate_name()
             verifier, challenge = generate_pkce_pair()
             oauth_state = generate_state()
-            authorize_url = build_authorize_url(verifier, challenge, oauth_state)
+            authorize_url = build_authorize_url(challenge, oauth_state)
 
             logger.info("[2/5] Registering ChatGPT for %s", email)
             chatgpt_page = await context.new_page()
@@ -347,24 +398,23 @@ async def register_chatgpt_account(
             )
 
             logger.info("[3/5] Getting verification message from email provider...")
-            code = await get_latest_code(email_provider, email)
+            code = await get_latest_code(email_provider)
             if not code:
                 raise AutomationError(
                     "email_provider", "Email provider returned no verification message"
                 )
-            logger.info("[3/5] Verification code extracted: %s", code)
+            logger.info("[3/5] Verification code extracted")
 
             await chatgpt_page.bring_to_front()
             code_input = chatgpt_page.get_by_placeholder("Code")
-            if await code_input.count() > 0:
-                await code_input.fill(code)
+            await code_input.first.wait_for(state="visible", timeout=10000)
+            await code_input.first.fill(code)
             await click_continue(chatgpt_page)
 
             logger.info("[4/5] Setting profile...")
             name_input = chatgpt_page.get_by_placeholder("Full name")
             await name_input.first.wait_for(state="visible", timeout=20000)
-            if await name_input.count() > 0:
-                await name_input.fill(full_name)
+            await name_input.first.fill(full_name)
 
             await fill_date_field(chatgpt_page, birth_month, birth_day, birth_year)
             profile_url = chatgpt_page.url
@@ -409,25 +459,45 @@ async def register_chatgpt_account(
                 if await continue_button.count() > 0:
                     await continue_button.first.click()
 
-            for label in ["Continue", "Allow", "Authorize"]:
-                button = oauth_page.get_by_role("button", name=label)
-                if await button.count() > 0:
-                    try:
-                        await button.first.click(timeout=5000)
-                        await oauth_page.wait_for_timeout(500)
-                    except Exception:
-                        pass
+            last_oauth_email_code = code
+            oauth_deadline = asyncio.get_running_loop().time() + 60
+            while asyncio.get_running_loop().time() < oauth_deadline:
+                if redirect_url_captured:
+                    break
+
+                if await oauth_needs_email_check(oauth_page):
+                    logger.info("OAuth requested email confirmation code")
+                    new_code = await get_latest_code(email_provider)
+                    if new_code and new_code != last_oauth_email_code:
+                        filled = await fill_oauth_code_if_present(oauth_page, new_code)
+                        if filled:
+                            last_oauth_email_code = new_code
+                            logger.info("Filled OAuth email confirmation code")
+                        else:
+                            logger.warning(
+                                "OAuth inbox challenge detected but code field not found"
+                            )
 
-            if not redirect_url_captured:
                 try:
-                    await oauth_page.wait_for_timeout(4000)
                     current_url = oauth_page.url
                     if "localhost:1455" in current_url and "code=" in current_url:
                         redirect_url_captured = current_url
                         logger.info("Captured OAuth redirect from page URL")
+                        break
                 except Exception:
                     pass
 
+                clicked = await click_first_visible_button(
+                    oauth_page,
+                    ["Continue", "Allow", "Authorize", "Verify"],
+                    timeout_ms=2000,
+                )
+
+                if clicked:
+                    await oauth_page.wait_for_timeout(500)
+                else:
+                    await oauth_page.wait_for_timeout(1000)
+
             if not redirect_url_captured:
                 raise AutomationError(
                     "oauth", "OAuth redirect with code was not captured", oauth_page
@@ -446,20 +516,18 @@ async def register_chatgpt_account(
                 raise AutomationError("oauth", "OAuth state mismatch", oauth_page)
 
             tokens = await exchange_code_for_tokens(auth_code, verifier)
-            save_tokens(tokens)
-            logger.info("OAuth tokens saved successfully")
+            logger.info("OAuth tokens fetched successfully")
 
-            return True
+            return tokens
 
     except AutomationError as e:
         logger.error(f"Error at step [{e.step}]: {e.message}")
         await save_error_screenshot(e.page, e.step)
-        return False
+        return None
     except Exception as e:
         logger.error(f"Unexpected error: {e}")
         await save_error_screenshot(current_page, "unexpected")
-        return False
+        return None
     finally:
         if managed:
-            await asyncio.sleep(2)
             await managed.close()

src/providers/chatgpt/tokens.py

@@ -1,6 +1,7 @@
 import json
 import logging
 import os
+import tempfile
 import time
 from pathlib import Path
 from typing import Any
@@ -35,7 +36,7 @@ def _dict_to_tokens(data: dict[str, Any] | None) -> ProviderTokens | None:
             refresh_token=data["refresh_token"],
             expires_at=data["expires_at"],
         )
-    except KeyError, TypeError:
+    except (KeyError, TypeError):
         return None
 
 
@@ -54,8 +55,20 @@ def _load_raw() -> dict[str, Any] | None:
 
 def _save_raw(data: dict[str, Any]) -> None:
     TOKENS_FILE.parent.mkdir(parents=True, exist_ok=True)
-    with open(TOKENS_FILE, "w") as f:
-        json.dump(data, f, indent=2)
+    fd, tmp_path = tempfile.mkstemp(
+        prefix=f"{TOKENS_FILE.name}.",
+        suffix=".tmp",
+        dir=str(TOKENS_FILE.parent),
+    )
+    try:
+        with os.fdopen(fd, "w") as f:
+            json.dump(data, f, indent=2)
+            f.flush()
+            os.fsync(f.fileno())
+        os.replace(tmp_path, TOKENS_FILE)
+    finally:
+        if os.path.exists(tmp_path):
+            os.unlink(tmp_path)
 
 
 def _normalize_state(data: dict[str, Any] | None) -> dict[str, Any]:
@@ -68,7 +81,6 @@ def _normalize_state(data: dict[str, Any] | None) -> dict[str, Any]:
             "next_account": data.get("next_account"),
         }
 
-    # Backward compatibility with old flat schema
     return {"active": data, "next_account": None}
 
 
@@ -79,9 +91,7 @@ def load_state() -> tuple[ProviderTokens | None, ProviderTokens | None]:
     return active, next_account
 
 
-def save_state(
-    active: ProviderTokens | None, next_account: ProviderTokens | None
-) -> None:
+def save_state(active: ProviderTokens | None, next_account: ProviderTokens | None) -> None:
     payload = {
         "active": _tokens_to_dict(active) if active else None,
         "next_account": _tokens_to_dict(next_account) if next_account else None,
@@ -104,13 +114,8 @@ def save_tokens(tokens: ProviderTokens):
     save_state(tokens, next_account)
 
 
-def save_next_tokens(tokens: ProviderTokens):
-    active, _ = load_state()
-    save_state(active, tokens)
-
-
 def promote_next_tokens() -> bool:
-    active, next_account = load_state()
+    _, next_account = load_state()
     if not next_account:
         return False
     save_state(next_account, None)
@@ -123,42 +128,34 @@ def clear_next_tokens():
 
 
 async def refresh_tokens(refresh_token: str) -> ProviderTokens | None:
-    async with aiohttp.ClientSession() as session:
-        data = {
-            "grant_type": "refresh_token",
-            "refresh_token": refresh_token,
-            "client_id": CLIENT_ID,
-        }
-        async with session.post(TOKEN_URL, data=data) as resp:
-            if not resp.ok:
-                text = await resp.text()
-                logger.warning("Token refresh failed: %s %s", resp.status, text)
-                return None
-            json_resp = await resp.json()
-            expires_in = json_resp["expires_in"]
-            return ProviderTokens(
-                access_token=json_resp["access_token"],
-                refresh_token=json_resp["refresh_token"],
-                expires_at=time.time() + expires_in,
-            )
-
-
-async def get_valid_tokens() -> ProviderTokens | None:
-    tokens = load_tokens()
-    if not tokens:
-        logger.info("No tokens found")
+    data = {
+        "grant_type": "refresh_token",
+        "refresh_token": refresh_token,
+        "client_id": CLIENT_ID,
+    }
+    timeout = aiohttp.ClientTimeout(total=15)
+    try:
+        async with aiohttp.ClientSession(timeout=timeout) as session:
+            async with session.post(TOKEN_URL, data=data) as resp:
+                if not resp.ok:
+                    text = await resp.text()
+                    logger.warning("Token refresh failed: %s %s", resp.status, text)
+                    return None
+                json_resp = await resp.json()
+    except (aiohttp.ClientError, TimeoutError) as e:
+        logger.warning("Token refresh request error: %s", e)
+        return None
+    except Exception as e:
+        logger.warning("Token refresh unexpected error: %s", e)
         return None
 
-    if tokens.is_expired:
-        logger.info("Token expired, refreshing...")
-        if not tokens.refresh_token:
-            logger.info("No refresh token available")
-            return None
-        new_tokens = await refresh_tokens(tokens.refresh_token)
-        if not new_tokens:
-            logger.warning("Failed to refresh token")
-            return None
-        save_tokens(new_tokens)
-        return new_tokens
-
-    return tokens
+    try:
+        expires_in = int(json_resp["expires_in"])
+        return ProviderTokens(
+            access_token=json_resp["access_token"],
+            refresh_token=json_resp["refresh_token"],
+            expires_at=time.time() + expires_in,
+        )
+    except (KeyError, TypeError, ValueError) as e:
+        logger.warning("Token refresh response parse error: %s", e)
+        return None

src/providers/chatgpt/usage.py

@@ -1,14 +1,15 @@
-import json
-import socket
-import urllib.error
-import urllib.request
+import logging
 from typing import Any
 
+import aiohttp
+
+logger = logging.getLogger(__name__)
+
 
 def clamp_percent(value: Any) -> int:
     try:
         num = float(value)
-    except TypeError, ValueError:
+    except (TypeError, ValueError):
         return 0
     if num < 0:
         return 0
@@ -28,30 +29,36 @@ def _parse_window(window: dict[str, Any] | None) -> dict[str, int] | None:
     }
 
 
-def get_usage_data(access_token: str, timeout_ms: int = 10000) -> dict[str, Any] | None:
+async def get_usage_data(
+    access_token: str,
+    timeout_ms: int = 10000,
+) -> dict[str, Any] | None:
     headers = {
         "Authorization": f"Bearer {access_token}",
         "User-Agent": "CodexProxy",
         "Accept": "application/json",
     }
 
-    req = urllib.request.Request(
-        "https://chatgpt.com/backend-api/wham/usage",
-        headers=headers,
-        method="GET",
-    )
+    timeout = aiohttp.ClientTimeout(total=timeout_ms / 1000)
+    url = "https://chatgpt.com/backend-api/wham/usage"
 
     try:
-        with urllib.request.urlopen(req, timeout=timeout_ms / 1000) as res:
-            body = res.read().decode("utf-8", errors="replace")
-    except urllib.error.HTTPError:
+        async with aiohttp.ClientSession(timeout=timeout) as session:
+            async with session.get(url, headers=headers) as res:
+                if not res.ok:
+                    body = await res.text()
+                    logger.warning(
+                        "Usage fetch failed: status=%s body=%s",
+                        res.status,
+                        body[:300],
+                    )
+                    return None
+                data = await res.json()
+    except (aiohttp.ClientError, TimeoutError) as e:
+        logger.warning("Usage fetch request error: %s", e)
         return None
-    except urllib.error.URLError, socket.timeout:
-        return None
-
-    try:
-        data = json.loads(body)
-    except json.JSONDecodeError:
+    except Exception as e:
+        logger.warning("Usage fetch unexpected error: %s", e)
         return None
 
     rate_limit = data.get("rate_limit") or {}
@@ -76,10 +83,3 @@ def get_usage_data(access_token: str, timeout_ms: int = 10000) -> dict[str, Any]
         "limit_reached": bool(rate_limit.get("limit_reached")),
         "allowed": bool(rate_limit.get("allowed", True)),
     }
-
-
-def get_usage_percent(access_token: str, timeout_ms: int = 10000) -> int:
-    data = get_usage_data(access_token, timeout_ms=timeout_ms)
-    if not data:
-        return -1
-    return int(data["used_percent"])

src/server.py

@@ -1,46 +1,57 @@
 import asyncio
 import logging
-import os
 
-from aiohttp import web
+from aiohttp import web, web_log
 
 from providers.chatgpt import ChatGPTProvider
-
-PORT = int(os.environ.get("PORT", "8080"))
-CHATGPT_PREPARE_THRESHOLD = int(os.environ.get("CHATGPT_PREPARE_THRESHOLD", "85"))
-LIMIT_EXHAUSTED_PERCENT = 100
+from providers.base import Provider
+from utils.env import parse_int_env
 
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
+PORT = parse_int_env("PORT", 8080, 1, 65535)
+LIMIT_EXHAUSTED_PERCENT = 100
 
-# Registry of available providers
-PROVIDERS = {
+PROVIDERS: dict[str, Provider] = {
     "chatgpt": ChatGPTProvider(),
 }
 
-refresh_locks = {name: asyncio.Lock() for name in PROVIDERS.keys()}
-background_refresh_tasks: dict[str, asyncio.Task | None] = {
-    name: None for name in PROVIDERS.keys()
-}
+background_tasks: dict[str, asyncio.Task | None] = {name: None for name in PROVIDERS}
 
 
-@web.middleware
-async def request_log_middleware(request: web.Request, handler):
-    response = await handler(request)
-    logger.info("%s %s -> %s", request.method, request.path_qs, response.status)
-    return response
+class AccessLogger(web_log.AccessLogger):
+    def log(
+        self, request: web.BaseRequest, response: web.StreamResponse, time: float
+    ) -> None:
+        if request.path == "/health":
+            return
+        super().log(request, response, time)
 
 
-def build_limit(usage_percent: int) -> dict[str, int | bool]:
+def build_limit(usage_percent: int, prepare_threshold: int) -> dict[str, int | bool]:
     remaining = max(0, 100 - usage_percent)
     return {
         "used_percent": usage_percent,
         "remaining_percent": remaining,
         "exhausted": usage_percent >= LIMIT_EXHAUSTED_PERCENT,
-        "needs_refresh": usage_percent >= CHATGPT_PREPARE_THRESHOLD,
+        "needs_prepare": usage_percent >= prepare_threshold,
     }
 
 
+def get_prepare_threshold(provider_name: str) -> int:
+    provider = PROVIDERS.get(provider_name)
+    if not provider:
+        return 100
+    return provider.prepare_threshold
+
+
+def should_trigger_standby_prepare(provider_name: str, usage_percent: int) -> bool:
+    provider = PROVIDERS.get(provider_name)
+    if not provider:
+        return False
+    return provider.should_prepare_standby()
+
+
 async def ensure_provider_token_ready(provider_name: str):
     provider = PROVIDERS.get(provider_name)
     if not provider:
@@ -52,135 +63,95 @@ async def ensure_provider_token_ready(provider_name: str):
         logger.warning(
             "[%s] Startup token check failed, forcing recreation", provider_name
         )
-        if isinstance(provider, ChatGPTProvider):
-            token = await provider.force_recreate_token()
+        token = await provider.force_recreate_token()
 
     if not token:
         logger.error("[%s] Could not prepare token at startup", provider_name)
         return
 
-    if isinstance(provider, ChatGPTProvider):
-        await provider.ensure_next_account()
-
     usage_info = await provider.get_usage_info(token)
-    if "error" not in usage_info:
-        logger.info("[%s] Startup token is ready", provider_name)
-        return
-
-    logger.warning(
-        "[%s] Startup token invalid for usage, forcing recreation", provider_name
-    )
-    if isinstance(provider, ChatGPTProvider):
+    if "error" in usage_info:
+        logger.warning(
+            "[%s] Startup token invalid for usage, forcing recreation", provider_name
+        )
         token = await provider.force_recreate_token()
-        if token:
-            logger.info("[%s] Startup token recreated successfully", provider_name)
+        if not token:
+            logger.error("[%s] Startup token recreation failed", provider_name)
             return
 
-    logger.error("[%s] Startup token recreation failed", provider_name)
+    await provider.startup_prepare()
+    logger.info("[%s] Startup token is ready", provider_name)
 
 
-async def on_startup(app: web.Application):
-    del app
-    for provider_name in PROVIDERS.keys():
-        await ensure_provider_token_ready(provider_name)
-
-
-async def issue_new_token(provider_name: str) -> str | None:
+async def ensure_standby_task(provider_name: str, usage_percent: int, reason: str):
     provider = PROVIDERS.get(provider_name)
     if not provider:
-        return None
+        return
 
-    async with refresh_locks[provider_name]:
-        logger.info(f"[{provider_name}] Generating new token")
-        success = await provider.register_new_account()
-        if not success:
-            logger.error(f"[{provider_name}] Token generation failed")
-            return None
+    if not provider.should_prepare_standby():
+        return
 
-    token = await provider.get_token()
-    if not token:
-        logger.error(f"[{provider_name}] Token was generated but not available")
-        return None
-
-    return token
-
-
-async def background_refresh_worker(provider_name: str, reason: str):
     try:
-        logger.info(f"[{provider_name}] Starting background token refresh ({reason})")
-        new_token = await issue_new_token(provider_name)
-        if new_token:
-            logger.info(f"[{provider_name}] Background token refresh completed")
-        else:
-            logger.error(f"[{provider_name}] Background token refresh failed")
+        logger.info("[%s] Preparing standby in background (%s)", provider_name, reason)
+        await provider.ensure_standby_account(usage_percent)
     except Exception:
-        logger.exception(
-            f"[{provider_name}] Unhandled error in background token refresh"
-        )
+        logger.exception("[%s] Unhandled standby preparation error", provider_name)
 
 
-def trigger_background_refresh(provider_name: str, reason: str):
-    task = background_refresh_tasks.get(provider_name)
+def trigger_standby_prepare(provider_name: str, usage_percent: int, reason: str):
+    task = background_tasks.get(provider_name)
     if task and not task.done():
         logger.info(
-            f"[{provider_name}] Background refresh already running, skip ({reason})"
+            "[%s] Standby prep already running, skip (%s)", provider_name, reason
         )
         return
-    background_refresh_tasks[provider_name] = asyncio.create_task(
-        background_refresh_worker(provider_name, reason)
+    background_tasks[provider_name] = asyncio.create_task(
+        ensure_standby_task(provider_name, usage_percent, reason)
     )
 
 
 async def token_handler(request: web.Request) -> web.Response:
     provider_name = request.match_info.get("provider", "chatgpt")
-
     provider = PROVIDERS.get(provider_name)
     if not provider:
         return web.json_response(
-            {"error": f"Unknown provider: {provider_name}"},
-            status=404,
+            {"error": f"Unknown provider: {provider_name}"}, status=404
         )
 
-    # Get or create token
     token = await provider.get_token()
     if not token:
-        return web.json_response(
-            {"error": "Failed to get active token"},
-            status=503,
-        )
+        return web.json_response({"error": "Failed to get active token"}, status=503)
 
-    # Get usage info
     usage_info = await provider.get_usage_info(token)
     if "error" in usage_info:
-        return web.json_response(
-            {"error": usage_info["error"]},
-            status=503,
+        return web.json_response({"error": usage_info["error"]}, status=503)
+
+    usage_percent = int(usage_info.get("used_percent", 0))
+    switched = await provider.maybe_rotate_account(usage_percent)
+    if switched:
+        token = await provider.get_token()
+        if not token:
+            return web.json_response(
+                {"error": "Failed to get active token after account switch"},
+                status=503,
+            )
+        usage_info = await provider.get_usage_info(token)
+        if "error" in usage_info:
+            return web.json_response({"error": usage_info["error"]}, status=503)
+        usage_percent = int(usage_info.get("used_percent", 0))
+        logger.info("[%s] Active account switched before response", provider_name)
+
+    prepare_threshold = get_prepare_threshold(provider_name)
+    if should_trigger_standby_prepare(provider_name, usage_percent):
+        trigger_standby_prepare(
+            provider_name,
+            usage_percent,
+            f"usage {usage_percent}% reached standby policy",
         )
 
-    usage_percent = usage_info.get("used_percent", 0)
-    remaining_percent = usage_info.get("remaining_percent", max(0, 100 - usage_percent))
-
-    if isinstance(provider, ChatGPTProvider):
-        switched = await provider.maybe_switch_active_account(usage_percent)
-        if switched:
-            token = await provider.get_token()
-            if not token:
-                return web.json_response(
-                    {"error": "Failed to get active token after account switch"},
-                    status=503,
-                )
-            usage_info = await provider.get_usage_info(token)
-            if "error" in usage_info:
-                return web.json_response(
-                    {"error": usage_info["error"]},
-                    status=503,
-                )
-            usage_percent = usage_info.get("used_percent", 0)
-            remaining_percent = usage_info.get(
-                "remaining_percent", max(0, 100 - usage_percent)
-            )
-            logger.info("[%s] Active account switched before response", provider_name)
-
+    remaining_percent = int(
+        usage_info.get("remaining_percent", max(0, 100 - usage_percent))
+    )
     logger.info(
         "[%s] token issued, used=%s%% remaining=%s%%",
         provider_name,
@@ -207,20 +178,10 @@ async def token_handler(request: web.Request) -> web.Response:
             secondary_window.get("reset_after_seconds", 0),
         )
 
-    # Trigger background refresh if needed
-    if usage_percent >= CHATGPT_PREPARE_THRESHOLD:
-        if isinstance(provider, ChatGPTProvider):
-            await provider.ensure_next_account()
-        else:
-            trigger_background_refresh(
-                provider_name,
-                f"usage {usage_percent}% >= threshold {CHATGPT_PREPARE_THRESHOLD}%",
-            )
-
     return web.json_response(
         {
             "token": token,
-            "limit": build_limit(usage_percent),
+            "limit": build_limit(usage_percent, prepare_threshold),
             "usage": {
                 "primary_window": primary_window,
                 "secondary_window": secondary_window,
@@ -229,22 +190,52 @@ async def token_handler(request: web.Request) -> web.Response:
     )
 
 
+async def health_handler(request: web.Request) -> web.Response:
+    del request
+    return web.Response(text="ok")
+
+
+async def on_startup(app: web.Application):
+    del app
+    for provider_name in PROVIDERS:
+        await ensure_provider_token_ready(provider_name)
+
+
+async def on_cleanup(app: web.Application):
+    del app
+    for task in background_tasks.values():
+        if task and not task.done():
+            task.cancel()
+    pending = [t for t in background_tasks.values() if t is not None]
+    if pending:
+        await asyncio.gather(*pending, return_exceptions=True)
+
+
 def create_app() -> web.Application:
-    app = web.Application(middlewares=[request_log_middleware])
     app.on_startup.append(on_startup)
-    # New route: /{provider}/token
+    app.on_cleanup.append(on_cleanup)
+    app.router.add_get("/health", health_handler)
     app.router.add_get("/{provider}/token", token_handler)
-    # Legacy route for backward compatibility
     app.router.add_get("/token", token_handler)
     return app
 
 
 if __name__ == "__main__":
     logger.info("Starting token service on port %s", PORT)
-    logger.info(
-        "ChatGPT prepare-next threshold: %s%%",
-        CHATGPT_PREPARE_THRESHOLD,
-    )
+    chatgpt_provider = PROVIDERS.get("chatgpt")
+    if chatgpt_provider:
+        logger.info(
+            "ChatGPT prepare-next threshold: %s%%", chatgpt_provider.prepare_threshold
+        )
+        if chatgpt_provider.switch_threshold is not None:
+            logger.info(
+                "ChatGPT switch threshold: %s%%", chatgpt_provider.switch_threshold
+            )
     logger.info("Available providers: %s", ", ".join(PROVIDERS.keys()))
     app = create_app()
-    web.run_app(app, host="0.0.0.0", port=PORT)
+    web.run_app(
+        app,
+        host="0.0.0.0",
+        port=PORT,
+        access_log_class=AccessLogger,
+    )

src/utils/__init__.py

@@ -0,0 +1,4 @@
+from .env import parse_int_env
+from .randoms import generate_password
+
+__all__ = ["parse_int_env", "generate_password"]

src/utils/env.py

@@ -0,0 +1,22 @@
+import os
+
+
+def parse_int_env(
+    name: str,
+    default: int,
+    minimum: int,
+    maximum: int,
+) -> int:
+    raw = os.environ.get(name)
+    if raw is None:
+        return default
+
+    try:
+        value = int(raw)
+    except ValueError:
+        return default
+
+    if value < minimum or value > maximum:
+        return default
+
+    return value

src/utils/randoms.py

@@ -0,0 +1,13 @@
+import random
+import secrets
+import string
+
+
+def generate_password(
+    length: int = 20,
+    *,
+    secure: bool = True,
+) -> str:
+    alphabet = string.ascii_letters + string.digits
+    chooser = secrets.choice if secure else random.choice
+    return "".join(chooser(alphabet) for _ in range(length))

tests/conftest.py

@@ -0,0 +1,12 @@
+import sys
+from pathlib import Path
+
+
+def _add_src_to_path() -> None:
+    root = Path(__file__).resolve().parents[1]
+    src = root / "src"
+    if str(src) not in sys.path:
+        sys.path.insert(0, str(src))
+
+
+_add_src_to_path()

tests/test_registration_unit.py

@@ -0,0 +1,37 @@
+from providers.chatgpt.registration import (
+    build_authorize_url,
+    extract_verification_code,
+    generate_birthdate_90s,
+    generate_name,
+)
+
+
+def test_generate_name_shape():
+    name = generate_name()
+    parts = name.split(" ")
+    assert len(parts) == 2
+    assert all(p.isalpha() for p in parts)
+
+
+def test_generate_birthdate_90s_range():
+    month, day, year = generate_birthdate_90s()
+    assert 1 <= int(month) <= 12
+    assert 1 <= int(day) <= 28
+    assert 1990 <= int(year) <= 1999
+
+
+def test_extract_verification_code_prefers_chatgpt_phrase():
+    text = "foo 123456 bar Your ChatGPT code is 654321"
+    assert extract_verification_code(text) == "654321"
+
+
+def test_extract_verification_code_fallback_last_code():
+    text = "codes 111111 and 222222"
+    assert extract_verification_code(text) == "222222"
+
+
+def test_build_authorize_url_contains_required_params():
+    url = build_authorize_url("challenge", "state123")
+    assert "response_type=code" in url
+    assert "code_challenge=challenge" in url
+    assert "state=state123" in url

tests/test_server_unit.py

@@ -0,0 +1,165 @@
+import asyncio
+import json
+
+import server
+from providers.base import Provider, ProviderTokens
+from utils.env import parse_int_env
+
+
+class FakeRequest:
+    def __init__(self, provider: str):
+        self.match_info = {"provider": provider}
+
+
+class FakeProvider(Provider):
+    def __init__(
+        self,
+        token: str | None = "tok",
+        usage: dict | None = None,
+        rotate: bool = False,
+    ):
+        self._token = token
+        self._usage = usage or {
+            "used_percent": 10,
+            "remaining_percent": 90,
+            "primary_window": None,
+            "secondary_window": None,
+        }
+        self._rotate = rotate
+        self._prepare_threshold = 80
+        self.get_token_calls = 0
+        self.standby_calls = 0
+
+    @property
+    def prepare_threshold(self) -> int:
+        return self._prepare_threshold
+
+    def should_prepare_standby(self) -> bool:
+        return False
+
+    @property
+    def name(self) -> str:
+        return "fake"
+
+    async def get_token(self) -> str | None:
+        self.get_token_calls += 1
+        return self._token
+
+    async def register_new_account(self) -> bool:
+        return True
+
+    async def get_usage_info(self, access_token: str) -> dict:
+        _ = access_token
+        return dict(self._usage)
+
+    def load_tokens(self) -> ProviderTokens | None:
+        return None
+
+    def save_tokens(self, tokens: ProviderTokens) -> None:
+        _ = tokens
+
+    async def maybe_rotate_account(self, usage_percent: int) -> bool:
+        _ = usage_percent
+        return self._rotate
+
+    async def ensure_standby_account(
+        self,
+        usage_percent: int,
+    ) -> None:
+        _ = usage_percent
+        self.standby_calls += 1
+
+
+def _response_json(resp) -> dict:
+    return json.loads(resp.body.decode("utf-8"))
+
+
+def test_parse_int_env_defaults(monkeypatch):
+    monkeypatch.delenv("X_TEST", raising=False)
+    assert parse_int_env("X_TEST", 10, 1, 20) == 10
+
+
+def test_parse_int_env_invalid(monkeypatch):
+    monkeypatch.setenv("X_TEST", "abc")
+    assert parse_int_env("X_TEST", 10, 1, 20) == 10
+
+
+def test_parse_int_env_out_of_range(monkeypatch):
+    monkeypatch.setenv("X_TEST", "999")
+    assert parse_int_env("X_TEST", 10, 1, 20) == 10
+
+
+def test_build_limit_fields():
+    limit = server.build_limit(90, 85)
+    assert limit == {
+        "used_percent": 90,
+        "remaining_percent": 10,
+        "exhausted": False,
+        "needs_prepare": True,
+    }
+
+
+def test_get_prepare_threshold():
+    assert (
+        server.get_prepare_threshold("chatgpt")
+        == server.PROVIDERS["chatgpt"].prepare_threshold
+    )
+    assert server.get_prepare_threshold("unknown") == 100
+
+
+def test_token_handler_unknown_provider(monkeypatch):
+    monkeypatch.setattr(server, "PROVIDERS", {})
+    resp = asyncio.run(server.token_handler(FakeRequest("missing")))
+    assert resp.status == 404
+
+
+def test_health_handler_ok():
+    resp = asyncio.run(server.health_handler(object()))
+    assert resp.status == 200
+    assert resp.text == "ok"
+
+
+def test_token_handler_success(monkeypatch):
+    provider = FakeProvider()
+    monkeypatch.setattr(server, "PROVIDERS", {"fake": provider})
+    monkeypatch.setattr(server, "background_tasks", {"fake": None})
+    resp = asyncio.run(server.token_handler(FakeRequest("fake")))
+    data = _response_json(resp)
+
+    assert resp.status == 200
+    assert data["token"] == "tok"
+    assert data["limit"]["needs_prepare"] is False
+
+
+def test_token_handler_triggers_standby(monkeypatch):
+    provider = FakeProvider(usage={"used_percent": 90, "remaining_percent": 10})
+    monkeypatch.setattr(server, "PROVIDERS", {"fake": provider})
+    monkeypatch.setattr(server, "background_tasks", {"fake": None})
+
+    called = {"value": False}
+
+    def fake_trigger(name, usage_percent, reason):
+        assert name == "fake"
+        assert usage_percent == 90
+        assert "standby policy" in reason
+        called["value"] = True
+
+    monkeypatch.setattr(server, "trigger_standby_prepare", fake_trigger)
+
+    resp = asyncio.run(server.token_handler(FakeRequest("fake")))
+    assert resp.status == 200
+    assert called["value"] is True
+
+
+def test_token_handler_rotation_path(monkeypatch):
+    provider = FakeProvider(
+        usage={"used_percent": 96, "remaining_percent": 4},
+        rotate=True,
+    )
+    monkeypatch.setattr(server, "PROVIDERS", {"fake": provider})
+    monkeypatch.setattr(server, "background_tasks", {"fake": None})
+    monkeypatch.setattr(server, "trigger_standby_prepare", lambda *_: None)
+
+    resp = asyncio.run(server.token_handler(FakeRequest("fake")))
+    assert resp.status == 200
+    assert provider.get_token_calls >= 2

tests/test_tokens_unit.py

@@ -0,0 +1,60 @@
+import json
+from pathlib import Path
+
+from providers.base import ProviderTokens
+from providers.chatgpt import tokens as t
+
+
+def test_normalize_state_backward_compatible():
+    raw = {"access_token": "a", "refresh_token": "r", "expires_at": 1}
+    normalized = t._normalize_state(raw)
+    assert normalized["active"]["access_token"] == "a"
+    assert normalized["next_account"] is None
+
+
+def test_promote_next_tokens(tmp_path, monkeypatch):
+    file_path = tmp_path / "chatgpt_tokens.json"
+    monkeypatch.setattr(t, "TOKENS_FILE", file_path)
+
+    active = ProviderTokens("a1", "r1", 100)
+    nxt = ProviderTokens("a2", "r2", 200)
+    t.save_state(active, nxt)
+
+    assert t.promote_next_tokens() is True
+    cur, next_cur = t.load_state()
+    assert cur is not None
+    assert cur.access_token == "a2"
+    assert next_cur is None
+
+
+def test_save_tokens_preserves_next(tmp_path, monkeypatch):
+    file_path = tmp_path / "chatgpt_tokens.json"
+    monkeypatch.setattr(t, "TOKENS_FILE", file_path)
+
+    active = ProviderTokens("a1", "r1", 100)
+    nxt = ProviderTokens("a2", "r2", 200)
+    t.save_state(active, nxt)
+
+    t.save_tokens(ProviderTokens("a3", "r3", 300))
+    cur, next_cur = t.load_state()
+    assert cur is not None and cur.access_token == "a3"
+    assert next_cur is not None and next_cur.access_token == "a2"
+
+
+def test_atomic_write_produces_valid_json(tmp_path, monkeypatch):
+    file_path = tmp_path / "chatgpt_tokens.json"
+    monkeypatch.setattr(t, "TOKENS_FILE", file_path)
+
+    t.save_state(ProviderTokens("x", "y", 123), None)
+    with open(file_path) as f:
+        data = json.load(f)
+    assert "active" in data
+    assert data["active"]["access_token"] == "x"
+
+
+def test_load_state_from_missing_file(tmp_path, monkeypatch):
+    file_path = tmp_path / "missing.json"
+    monkeypatch.setattr(t, "TOKENS_FILE", file_path)
+    active, nxt = t.load_state()
+    assert active is None
+    assert nxt is None

tests/test_usage_unit.py

@@ -0,0 +1,32 @@
+from providers.chatgpt.usage import _parse_window, clamp_percent
+
+
+def test_clamp_percent_bounds():
+    assert clamp_percent(-1) == 0
+    assert clamp_percent(150) == 100
+    assert clamp_percent(49.6) == 50
+
+
+def test_clamp_percent_invalid():
+    assert clamp_percent(None) == 0
+    assert clamp_percent("bad") == 0
+
+
+def test_parse_window_valid():
+    window = {
+        "used_percent": 34.4,
+        "limit_window_seconds": 3600,
+        "reset_after_seconds": 120,
+        "reset_at": 999,
+    }
+    parsed = _parse_window(window)
+    assert parsed == {
+        "used_percent": 34,
+        "limit_window_seconds": 3600,
+        "reset_after_seconds": 120,
+        "reset_at": 999,
+    }
+
+
+def test_parse_window_none():
+    assert _parse_window(None) is None

uv.lock

@@ -83,6 +83,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/3a/2a/7cc015f5b9f5db42b7d48157e23356022889fc354a2813c15934b7cb5c0e/attrs-25.4.0-py3-none-any.whl", hash = "sha256:adcf7e2a1fb3b36ac48d97835bb6d8ade15b8dcce26aba8bf1d14847b57a3373", size = 67615, upload-time = "2025-10-06T13:54:43.17Z" },
 ]
 
+[[package]]
+name = "colorama"
+version = "0.4.6"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/d8/53/6f443c9a4a8358a93a6792e2acffb9d9d5cb0a5cfd8802644b7b1c9a02e4/colorama-0.4.6.tar.gz", hash = "sha256:08695f5cb7ed6e0531a20572697297273c47b8cae5a63ffc6d6ed5c201be6e44", size = 27697, upload-time = "2022-10-25T02:36:22.414Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/d1/d6/3965ed04c63042e047cb6a3e6ed1a63a35087b6a609aa3a15ed8ac56c221/colorama-0.4.6-py2.py3-none-any.whl", hash = "sha256:4f1d9991f5acc0ca119f9d443620b77f9d6b33703e51011c16baf57afb285fc6", size = 25335, upload-time = "2022-10-25T02:36:20.889Z" },
+]
+
 [[package]]
 name = "frozenlist"
 version = "1.8.0"
@@ -158,6 +167,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/0e/61/66938bbb5fc52dbdf84594873d5b51fb1f7c7794e9c0f5bd885f30bc507b/idna-3.11-py3-none-any.whl", hash = "sha256:771a87f49d9defaf64091e6e6fe9c18d4833f140bd19464795bc32d966ca37ea", size = 71008, upload-time = "2025-10-12T14:55:18.883Z" },
 ]
 
+[[package]]
+name = "iniconfig"
+version = "2.3.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/72/34/14ca021ce8e5dfedc35312d08ba8bf51fdd999c576889fc2c24cb97f4f10/iniconfig-2.3.0.tar.gz", hash = "sha256:c76315c77db068650d49c5b56314774a7804df16fee4402c1f19d6d15d8c4730", size = 20503, upload-time = "2025-10-18T21:55:43.219Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/cb/b1/3846dd7f199d53cb17f49cba7e651e9ce294d8497c8c150530ed11865bb8/iniconfig-2.3.0-py3-none-any.whl", hash = "sha256:f631c04d2c48c52b84d0d0549c99ff3859c98df65b3101406327ecc7d53fbf12", size = 7484, upload-time = "2025-10-18T21:55:41.639Z" },
+]
+
 [[package]]
 name = "megapt"
 version = "0.1.0"
@@ -168,12 +186,19 @@ dependencies = [
     { name = "playwright" },
 ]
 
+[package.optional-dependencies]
+dev = [
+    { name = "pytest" },
+]
+
 [package.metadata]
 requires-dist = [
     { name = "aiohttp", specifier = "==3.13.3" },
     { name = "pkce", specifier = "==1.0.3" },
     { name = "playwright", specifier = "==1.58.0" },
+    { name = "pytest", marker = "extra == 'dev'", specifier = ">=8.0.0" },
 ]
+provides-extras = ["dev"]
 
 [[package]]
 name = "multidict"
@@ -220,6 +245,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/81/08/7036c080d7117f28a4af526d794aab6a84463126db031b007717c1a6676e/multidict-6.7.1-py3-none-any.whl", hash = "sha256:55d97cc6dae627efa6a6e548885712d4864b81110ac76fa4e534c03819fa4a56", size = 12319, upload-time = "2026-01-26T02:46:44.004Z" },
 ]
 
+[[package]]
+name = "packaging"
+version = "26.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/65/ee/299d360cdc32edc7d2cf530f3accf79c4fca01e96ffc950d8a52213bd8e4/packaging-26.0.tar.gz", hash = "sha256:00243ae351a257117b6a241061796684b084ed1c516a08c48a3f7e147a9d80b4", size = 143416, upload-time = "2026-01-21T20:50:39.064Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/b7/b9/c538f279a4e237a006a2c98387d081e9eb060d203d8ed34467cc0f0b9b53/packaging-26.0-py3-none-any.whl", hash = "sha256:b36f1fef9334a5588b4166f8bcd26a14e521f2b55e6b9de3aaa80d3ff7a37529", size = 74366, upload-time = "2026-01-21T20:50:37.788Z" },
+]
+
 [[package]]
 name = "pkce"
 version = "1.0.3"
@@ -248,6 +282,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/c8/c4/cc0229fea55c87d6c9c67fe44a21e2cd28d1d558a5478ed4d617e9fb0c93/playwright-1.58.0-py3-none-win_arm64.whl", hash = "sha256:32ffe5c303901a13a0ecab91d1c3f74baf73b84f4bedbb6b935f5bc11cc98e1b", size = 33085919, upload-time = "2026-01-30T15:09:45.71Z" },
 ]
 
+[[package]]
+name = "pluggy"
+version = "1.6.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/f9/e2/3e91f31a7d2b083fe6ef3fa267035b518369d9511ffab804f839851d2779/pluggy-1.6.0.tar.gz", hash = "sha256:7dcc130b76258d33b90f61b658791dede3486c3e6bfb003ee5c9bfb396dd22f3", size = 69412, upload-time = "2025-05-15T12:30:07.975Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/54/20/4d324d65cc6d9205fabedc306948156824eb9f0ee1633355a8f7ec5c66bf/pluggy-1.6.0-py3-none-any.whl", hash = "sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746", size = 20538, upload-time = "2025-05-15T12:30:06.134Z" },
+]
+
 [[package]]
 name = "propcache"
 version = "0.4.1"
@@ -299,6 +342,31 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/a0/c4/b4d4827c93ef43c01f599ef31453ccc1c132b353284fc6c87d535c233129/pyee-13.0.1-py3-none-any.whl", hash = "sha256:af2f8fede4171ef667dfded53f96e2ed0d6e6bd7ee3bb46437f77e3b57689228", size = 15659, upload-time = "2026-02-14T21:12:26.263Z" },
 ]
 
+[[package]]
+name = "pygments"
+version = "2.19.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/b0/77/a5b8c569bf593b0140bde72ea885a803b82086995367bf2037de0159d924/pygments-2.19.2.tar.gz", hash = "sha256:636cb2477cec7f8952536970bc533bc43743542f70392ae026374600add5b887", size = 4968631, upload-time = "2025-06-21T13:39:12.283Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/c7/21/705964c7812476f378728bdf590ca4b771ec72385c533964653c68e86bdc/pygments-2.19.2-py3-none-any.whl", hash = "sha256:86540386c03d588bb81d44bc3928634ff26449851e99741617ecb9037ee5ec0b", size = 1225217, upload-time = "2025-06-21T13:39:07.939Z" },
+]
+
+[[package]]
+name = "pytest"
+version = "9.0.2"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "iniconfig" },
+    { name = "packaging" },
+    { name = "pluggy" },
+    { name = "pygments" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/d1/db/7ef3487e0fb0049ddb5ce41d3a49c235bf9ad299b6a25d5780a89f19230f/pytest-9.0.2.tar.gz", hash = "sha256:75186651a92bd89611d1d9fc20f0b4345fd827c41ccd5c299a868a05d70edf11", size = 1568901, upload-time = "2025-12-06T21:30:51.014Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/3b/ab/b3226f0bd7cdcf710fbede2b3548584366da3b19b5021e74f5bde2a8fa3f/pytest-9.0.2-py3-none-any.whl", hash = "sha256:711ffd45bf766d5264d487b917733b453d917afd2b0ad65223959f59089f875b", size = 374801, upload-time = "2025-12-06T21:30:49.154Z" },
+]
+
 [[package]]
 name = "typing-extensions"
 version = "4.15.0"