From aeefb928607c577da4c729d75b3e5d49757cb417 Mon Sep 17 00:00:00 2001
From: timbl <123@qq.com>
Date: Wed, 1 Apr 2026 13:03:33 +0800
Subject: [PATCH 1/2] fix(webui): keep registration entry flow after restart

---
 webui.py | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/webui.py b/webui.py
index d50ab18e..d06cdc23 100644
--- a/webui.py
+++ b/webui.py
@@ -168,6 +168,18 @@ def main():
     parser.add_argument("--access-password", help="Web UI 访问密钥 (也可通过 WEBUI_ACCESS_PASSWORD 环境变量设置)")
     args = parser.parse_args()
 
+    # 先准备运行目录与数据库，再应用命令行/环境变量覆盖。
+    # 否则 update_settings 会在数据库未初始化时退回默认内存配置，
+    # 导致像 registration_entry_flow 这类已保存设置在重启后看起来“被重置”。
+    _load_dotenv()
+    data_dir = project_root / "data"
+    logs_dir = project_root / "logs"
+    data_dir.mkdir(exist_ok=True)
+    logs_dir.mkdir(exist_ok=True)
+    os.environ.setdefault("APP_DATA_DIR", str(data_dir))
+    os.environ.setdefault("APP_LOGS_DIR", str(logs_dir))
+    initialize_database()
+
     # 更新配置
     from src.config.settings import update_settings
 

From d49fff26f28ae58093e026bf3971d70dafc1abce Mon Sep 17 00:00:00 2001
From: timbl <123@qq.com>
Date: Wed, 1 Apr 2026 13:48:37 +0800
Subject: [PATCH 2/2] fix(upload): backfill chatgpt account fields for codex
 json

---
 src/core/register.py                |  61 ++++++++++++++-
 src/core/upload/cpa_upload.py       | 116 ++++++++++++++++++++++++++--
 src/database/crud.py                |   2 +
 tests/test_cpa_upload_token_json.py |  89 +++++++++++++++++++++
 4 files changed, 259 insertions(+), 9 deletions(-)
 create mode 100644 tests/test_cpa_upload_token_json.py

diff --git a/src/core/register.py b/src/core/register.py
index 08450164..29fd2562 100644
--- a/src/core/register.py
+++ b/src/core/register.py
@@ -12,7 +12,7 @@
 import uuid
 from typing import Optional, Dict, Any, Tuple, Callable, List
 from dataclasses import dataclass
-from datetime import datetime
+from datetime import datetime, timezone
 
 from curl_cffi import requests as cffi_requests
 
@@ -1858,6 +1858,52 @@ def _extract_account_id_from_access_token(self, access_token: str) -> str:
         except Exception:
             return ""
 
+    def _extract_token_timestamps(self, token: str) -> Tuple[Optional[datetime], Optional[datetime]]:
+        """
+        从 JWT（access_token / id_token）payload 提取 iat / exp，返回 naive UTC datetime。
+        """
+        try:
+            raw = str(token or "").strip()
+            if raw.count(".") < 2:
+                return None, None
+            payload = raw.split(".")[1]
+            import base64
+            pad = "=" * ((4 - (len(payload) % 4)) % 4)
+            decoded = base64.urlsafe_b64decode((payload + pad).encode("ascii"))
+            claims = json.loads(decoded.decode("utf-8"))
+            if not isinstance(claims, dict):
+                return None, None
+
+            iat = claims.get("iat")
+            exp = claims.get("exp")
+
+            issued_at = (
+                datetime.fromtimestamp(int(iat), tz=timezone.utc).replace(tzinfo=None)
+                if iat not in (None, "")
+                else None
+            )
+            expires_at = (
+                datetime.fromtimestamp(int(exp), tz=timezone.utc).replace(tzinfo=None)
+                if exp not in (None, "")
+                else None
+            )
+            return issued_at, expires_at
+        except Exception:
+            return None, None
+
+    @staticmethod
+    def _parse_iso_datetime_to_naive_utc(value: str) -> Optional[datetime]:
+        try:
+            text = str(value or "").strip()
+            if not text:
+                return None
+            parsed = datetime.fromisoformat(text.replace("Z", "+00:00"))
+            if parsed.tzinfo is None:
+                return parsed
+            return parsed.astimezone(timezone.utc).replace(tzinfo=None)
+        except Exception:
+            return None
+
     def _ensure_native_required_tokens(self, result: RegistrationResult) -> bool:
         """
         原生注册入口要求拿齐：
@@ -2995,6 +3041,17 @@ def save_to_database(
         try:
             # 获取默认 client_id
             settings = get_settings()
+            token_issued_at, token_expires_at = self._extract_token_timestamps(
+                result.access_token or result.id_token
+            )
+            metadata_expires_at = None
+            if isinstance(result.metadata, dict):
+                metadata_expires_at = self._parse_iso_datetime_to_naive_utc(
+                    str(result.metadata.get("expires") or "")
+                )
+            final_expires_at = token_expires_at or metadata_expires_at
+            now_utc_naive = datetime.now(timezone.utc).replace(tzinfo=None)
+            final_last_refresh = token_issued_at or (now_utc_naive if result.access_token else None)
 
             with get_db() as db:
                 # 保存账户信息
@@ -3013,6 +3070,8 @@ def save_to_database(
                     refresh_token=result.refresh_token,
                     id_token=result.id_token,
                     proxy_used=self.proxy_url,
+                    last_refresh=final_last_refresh,
+                    expires_at=final_expires_at,
                     extra_data=result.metadata,
                     source=result.source,
                     account_label=account_label,
diff --git a/src/core/upload/cpa_upload.py b/src/core/upload/cpa_upload.py
index a244eff4..fc941529 100644
--- a/src/core/upload/cpa_upload.py
+++ b/src/core/upload/cpa_upload.py
@@ -4,6 +4,7 @@
 
 import json
 import logging
+import base64
 from typing import List, Dict, Any, Tuple, Optional
 from datetime import datetime
 from urllib.parse import quote
@@ -14,11 +15,58 @@
 from ...database.session import get_db
 from ...database.models import Account
 from ...config.settings import get_settings
-from ..timezone_utils import utcnow_naive
+from ..timezone_utils import UTC, SHANGHAI_TZ, utcnow_naive
 
 logger = logging.getLogger(__name__)
 
 
+def _decode_jwt_payload_unverified(token: str) -> Dict[str, Any]:
+    text = str(token or "").strip()
+    if not text or text.count(".") < 2:
+        return {}
+    try:
+        payload_b64 = text.split(".")[1]
+        padding = "=" * (-len(payload_b64) % 4)
+        raw = base64.urlsafe_b64decode((payload_b64 + padding).encode("utf-8"))
+        data = json.loads(raw.decode("utf-8"))
+        return data if isinstance(data, dict) else {}
+    except Exception:
+        return {}
+
+
+def _parse_unix_ts_to_naive_utc(value: Any) -> Optional[datetime]:
+    try:
+        if value in (None, ""):
+            return None
+        ts = int(value)
+        return datetime.fromtimestamp(ts, tz=UTC).replace(tzinfo=None)
+    except Exception:
+        return None
+
+
+def _parse_iso_to_naive_utc(value: Any) -> Optional[datetime]:
+    try:
+        text = str(value or "").strip()
+        if not text:
+            return None
+        dt = datetime.fromisoformat(text.replace("Z", "+00:00"))
+        if dt.tzinfo is None:
+            return dt
+        return dt.astimezone(UTC).replace(tzinfo=None)
+    except Exception:
+        return None
+
+
+def _format_cpa_datetime(dt: Optional[datetime]) -> str:
+    if dt is None:
+        return ""
+    if dt.tzinfo is None:
+        dt = dt.replace(tzinfo=UTC)
+    else:
+        dt = dt.astimezone(UTC)
+    return dt.astimezone(SHANGHAI_TZ).strftime("%Y-%m-%dT%H:%M:%S+08:00")
+
+
 def _normalize_cpa_auth_files_url(api_url: str) -> str:
     """将用户填写的 CPA 地址规范化为 auth-files 接口地址。"""
     normalized = (api_url or "").strip().rstrip("/")
@@ -100,15 +148,67 @@ def generate_token_json(account: Account) -> dict:
     Returns:
         CPA 格式的 Token 字典
     """
+    access_token = str(account.access_token or "").strip()
+    id_token_raw = str(account.id_token or "").strip()
+    # 兼容历史下游：部分工具只从 id_token 提取 chatgpt_account_id。
+    # 当注册链路仅拿到 access_token 时，使用 access_token 作为兜底 id_token 供解析。
+    id_token = id_token_raw or access_token
+    refresh_token = str(account.refresh_token or "").strip()
+
+    claims = _decode_jwt_payload_unverified(access_token) or _decode_jwt_payload_unverified(id_token)
+    auth_claims = claims.get("https://api.openai.com/auth") if isinstance(claims, dict) else {}
+    auth_claims = auth_claims if isinstance(auth_claims, dict) else {}
+    profile_claims = claims.get("https://api.openai.com/profile") if isinstance(claims, dict) else {}
+    profile_claims = profile_claims if isinstance(profile_claims, dict) else {}
+
+    account_id = str(
+        account.account_id
+        or auth_claims.get("chatgpt_account_id")
+        or claims.get("chatgpt_account_id")
+        or ""
+    ).strip()
+    workspace_id = str(account.workspace_id or account_id).strip()
+    chatgpt_user_id = str(
+        auth_claims.get("chatgpt_user_id")
+        or auth_claims.get("user_id")
+        or claims.get("chatgpt_user_id")
+        or claims.get("user_id")
+        or ""
+    ).strip()
+
+    email = str(
+        account.email
+        or profile_claims.get("email")
+        or ""
+    ).strip()
+
+    token_issued_at = _parse_unix_ts_to_naive_utc(claims.get("iat"))
+    token_expires_at = _parse_unix_ts_to_naive_utc(claims.get("exp"))
+    extra_data = account.extra_data if isinstance(account.extra_data, dict) else {}
+    metadata_expires_at = _parse_iso_to_naive_utc(extra_data.get("expires"))
+
+    last_refresh = account.last_refresh or token_issued_at or account.registered_at
+    expires_at = account.expires_at or token_expires_at or metadata_expires_at
+
+    client_id = str(
+        claims.get("client_id")
+        or account.client_id
+        or ""
+    ).strip()
+
     return {
         "type": "codex",
-        "email": account.email,
-        "expired": account.expires_at.strftime("%Y-%m-%dT%H:%M:%S+08:00") if account.expires_at else "",
-        "id_token": account.id_token or "",
-        "account_id": account.account_id or "",
-        "access_token": account.access_token or "",
-        "last_refresh": account.last_refresh.strftime("%Y-%m-%dT%H:%M:%S+08:00") if account.last_refresh else "",
-        "refresh_token": account.refresh_token or "",
+        "email": email,
+        "expired": _format_cpa_datetime(expires_at),
+        "id_token": id_token,
+        "account_id": account_id,
+        "workspace_id": workspace_id,
+        "chatgpt_account_id": account_id,
+        "chatgpt_user_id": chatgpt_user_id,
+        "client_id": client_id,
+        "access_token": access_token,
+        "last_refresh": _format_cpa_datetime(last_refresh),
+        "refresh_token": refresh_token,
     }
 
 
diff --git a/src/database/crud.py b/src/database/crud.py
index 8b712b93..cd56be5f 100644
--- a/src/database/crud.py
+++ b/src/database/crud.py
@@ -52,6 +52,7 @@ def create_account(
     id_token: Optional[str] = None,
     cookies: Optional[str] = None,
     proxy_used: Optional[str] = None,
+    last_refresh: Optional['datetime'] = None,
     expires_at: Optional['datetime'] = None,
     extra_data: Optional[Dict[str, Any]] = None,
     status: Optional[str] = None,
@@ -88,6 +89,7 @@ def create_account(
         id_token=id_token,
         cookies=cookies,
         proxy_used=proxy_used,
+        last_refresh=last_refresh,
         expires_at=expires_at,
         extra_data=extra_data or {},
         status=status or 'active',
diff --git a/tests/test_cpa_upload_token_json.py b/tests/test_cpa_upload_token_json.py
new file mode 100644
index 00000000..dbc207cd
--- /dev/null
+++ b/tests/test_cpa_upload_token_json.py
@@ -0,0 +1,89 @@
+import base64
+import json
+from datetime import datetime
+
+from src.core.upload.cpa_upload import generate_token_json
+from src.database.models import Account
+
+
+def _jwt_with_payload(payload: dict) -> str:
+    header = {"alg": "none", "typ": "JWT"}
+
+    def _b64(obj: dict) -> str:
+        raw = json.dumps(obj, separators=(",", ":"), ensure_ascii=False).encode("utf-8")
+        return base64.urlsafe_b64encode(raw).decode("ascii").rstrip("=")
+
+    return f"{_b64(header)}.{_b64(payload)}."
+
+
+def test_generate_token_json_backfills_time_and_account_id_from_access_token():
+    payload = {
+        "iat": 1775019821,  # 2026-04-01T05:03:41Z
+        "exp": 1775883821,  # 2026-04-11T05:03:41Z
+        "https://api.openai.com/auth": {
+            "chatgpt_account_id": "acct_from_jwt",
+        },
+        "https://api.openai.com/profile": {
+            "email": "jwt@example.com",
+        },
+    }
+
+    account = Account(
+        email="ericvillegas9964@outlook.com",
+        email_service="luckmail",
+        access_token=_jwt_with_payload(payload),
+        account_id="",
+        last_refresh=None,
+        expires_at=None,
+    )
+
+    token_data = generate_token_json(account)
+
+    assert token_data["account_id"] == "acct_from_jwt"
+    assert token_data["email"] == "ericvillegas9964@outlook.com"
+    assert token_data["last_refresh"] == "2026-04-01T13:03:41+08:00"
+    assert token_data["expired"] == "2026-04-11T13:03:41+08:00"
+    assert token_data["id_token"] == account.access_token
+    assert token_data["workspace_id"] == "acct_from_jwt"
+    assert token_data["chatgpt_account_id"] == "acct_from_jwt"
+
+
+def test_generate_token_json_prefers_db_timestamps_over_jwt_claims():
+    payload = {
+        "iat": 1775019821,
+        "exp": 1775883821,
+        "https://api.openai.com/auth": {"chatgpt_account_id": "acct_from_jwt"},
+    }
+
+    account = Account(
+        email="db@example.com",
+        email_service="luckmail",
+        access_token=_jwt_with_payload(payload),
+        account_id="acct_from_db",
+        last_refresh=datetime(2026, 4, 2, 0, 0, 0),  # naive UTC
+        expires_at=datetime(2026, 4, 3, 0, 0, 0),  # naive UTC
+    )
+
+    token_data = generate_token_json(account)
+
+    assert token_data["account_id"] == "acct_from_db"
+    assert token_data["last_refresh"] == "2026-04-02T08:00:00+08:00"
+    assert token_data["expired"] == "2026-04-03T08:00:00+08:00"
+    assert token_data["chatgpt_account_id"] == "acct_from_db"
+
+
+def test_generate_token_json_uses_metadata_expires_when_db_and_jwt_missing():
+    account = Account(
+        email="meta@example.com",
+        email_service="luckmail",
+        access_token="",
+        account_id="acct_meta",
+        last_refresh=None,
+        expires_at=None,
+        extra_data={"expires": "2026-06-30T05:03:46.241Z"},
+    )
+
+    token_data = generate_token_json(account)
+
+    assert token_data["expired"] == "2026-06-30T13:03:46+08:00"
+    assert token_data["id_token"] == ""