feat: support real browser auth import

backend/app/services/auth_capture_service.py

@@ -137,9 +137,12 @@ def _cookie_matches_hostname(cookie_domain: str, hostname: str) -> bool:
     """判断 cookie domain 是否适用于给定 hostname。
 
     支持带点前缀的 domain（如 `.saki.lat` 匹配 `api.saki.lat`）。
+    注意：hostname 为空时，调用方应跳过 cookie 收集而不是调用此函数。
     """
-    if not cookie_domain or not hostname:
-        return True  # 无 domain 限制时视为全域
+    if not cookie_domain:
+        return True  # 无 domain 限制的 cookie 对当前域有效
+    if not hostname:
+        return False  # 无法确定当前域，保守拒绝
     # 去掉前缀点
     domain = cookie_domain.lstrip(".")
     return hostname == domain or hostname.endswith("." + domain)
@@ -153,14 +156,22 @@ def _build_cookie_bundle(
 
     返回 (cookie_string, cookie_names_list)。
     cookie_string 格式：name1=value1; name2=value2; ...
-    过滤掉空值 cookie。
+    过滤掉空值 cookie。若 page_url 为空或无法解析 hostname，返回空结果
+    （不收集全域 cookie 以防误写入无关域凭证）。
     """
+    if not page_url:
+        logger.debug("_build_cookie_bundle: no page_url, skipping cookie collection")
+        return "", []
+
     hostname = ""
-    if page_url:
-        try:
-            hostname = urlparse(page_url).hostname or ""
-        except Exception:
-            pass
+    try:
+        hostname = urlparse(page_url).hostname or ""
+    except Exception:
+        pass
+
+    if not hostname:
+        logger.debug("_build_cookie_bundle: cannot parse hostname from %s, skipping", page_url[:80])
+        return "", []
 
     parts: list[str] = []
     names: list[str] = []
@@ -170,7 +181,7 @@ def _build_cookie_bundle(
         domain = c.get("domain", "")
         if not name or not value:
             continue
-        if hostname and not _cookie_matches_hostname(domain, hostname):
+        if not _cookie_matches_hostname(domain, hostname):
             continue
         parts.append(f"{name}={value}")
         names.append(name)

backend/app/services/browser_import_service.py

@@ -0,0 +1,159 @@
+"""One-time credential import sessions for real-browser auth capture."""
+from __future__ import annotations
+
+import hashlib
+import secrets
+import time
+from dataclasses import dataclass, field
+from typing import Any
+
+from app.services.auth_capture_service import _curate_candidates, _find_new_api_user
+
+
+IMPORT_SESSION_TTL_SECONDS = 600
+
+
+class ImportSessionError(RuntimeError):
+    pass
+
+
+@dataclass
+class BrowserImportSession:
+    id: str
+    secret_hash: str
+    target_url: str
+    created_by: str
+    expires_at: float
+    payload: dict[str, Any] | None = None
+    consumed: bool = False
+    created_at: float = field(default_factory=time.time)
+
+
+def _hash_secret(secret: str) -> str:
+    return hashlib.sha256(secret.encode("utf-8")).hexdigest()
+
+
+def _normalize_storage(value: Any) -> dict[str, str]:
+    if not isinstance(value, dict):
+        return {}
+    result: dict[str, str] = {}
+    for key, item in value.items():
+        if item is None:
+            continue
+        result[str(key)] = item if isinstance(item, str) else str(item)
+    return result
+
+
+def _normalize_cookies(value: Any) -> list[dict[str, Any]]:
+    if not isinstance(value, list):
+        return []
+    result: list[dict[str, Any]] = []
+    for item in value:
+        if not isinstance(item, dict):
+            continue
+        name = str(item.get("name") or "").strip()
+        cookie_value = str(item.get("value") or "")
+        if not name or not cookie_value:
+            continue
+        result.append({
+            "name": name,
+            "value": cookie_value,
+            "domain": str(item.get("domain") or ""),
+            "path": str(item.get("path") or "/"),
+            "httpOnly": bool(item.get("httpOnly", False)),
+            "secure": bool(item.get("secure", False)),
+        })
+    return result
+
+
+def _normalize_headers(value: Any) -> list[dict[str, str]]:
+    if not isinstance(value, list):
+        return []
+    result: list[dict[str, str]] = []
+    for item in value:
+        if not isinstance(item, dict):
+            continue
+        header_value = str(item.get("value") or "").strip()
+        if not header_value:
+            continue
+        result.append({
+            "type": str(item.get("type") or "authorization"),
+            "value": header_value,
+            "url": str(item.get("url") or ""),
+        })
+    return result
+
+
+def build_import_result(payload: dict[str, Any]) -> dict[str, Any]:
+    """Convert extension-submitted payload into auth-capture result shape."""
+    page_url = str(payload.get("page_url") or payload.get("url") or "")
+    cookies = _normalize_cookies(payload.get("cookies"))
+    storage = _normalize_storage(payload.get("local_storage") or payload.get("storage"))
+    session_storage = _normalize_storage(payload.get("session_storage"))
+    auth_headers = _normalize_headers(payload.get("auth_headers"))
+    new_api_user = _find_new_api_user(storage, session_storage)
+    candidates = _curate_candidates(
+        cookies=cookies,
+        local_storage=storage,
+        session_storage=session_storage,
+        auth_headers=auth_headers,
+        new_api_user=new_api_user,
+        page_url=page_url,
+    )
+    return {
+        "cookies": cookies,
+        "storage": storage,
+        "session_storage": session_storage,
+        "auth_headers": auth_headers,
+        "candidates": candidates,
+    }
+
+
+class BrowserImportService:
+    def __init__(self) -> None:
+        self._sessions: dict[str, BrowserImportSession] = {}
+
+    def create(self, target_url: str, created_by: str) -> tuple[BrowserImportSession, str]:
+        self.cleanup()
+        session_id = secrets.token_urlsafe(18)
+        secret = secrets.token_urlsafe(24)
+        session = BrowserImportSession(
+            id=session_id,
+            secret_hash=_hash_secret(secret),
+            target_url=target_url,
+            created_by=created_by,
+            expires_at=time.time() + IMPORT_SESSION_TTL_SECONDS,
+        )
+        self._sessions[session_id] = session
+        return session, secret
+
+    def get(self, session_id: str, created_by: str | None = None) -> BrowserImportSession:
+        self.cleanup()
+        session = self._sessions.get(session_id)
+        if not session:
+            raise ImportSessionError("import session not found")
+        if created_by is not None and session.created_by != created_by:
+            raise ImportSessionError("import session not found")
+        if session.expires_at <= time.time():
+            self._sessions.pop(session_id, None)
+            raise ImportSessionError("import session expired")
+        return session
+
+    def submit(self, session_id: str, secret: str, payload: dict[str, Any]) -> BrowserImportSession:
+        session = self.get(session_id)
+        if session.consumed:
+            raise ImportSessionError("import session already consumed")
+        if not secrets.compare_digest(session.secret_hash, _hash_secret(secret)):
+            raise ImportSessionError("invalid import secret")
+        session.payload = payload
+        session.consumed = True
+        return session
+
+    def cleanup(self) -> None:
+        now = time.time()
+        expired = [sid for sid, session in self._sessions.items() if session.expires_at <= now]
+        for sid in expired:
+            self._sessions.pop(sid, None)
+
+
+browser_imports = BrowserImportService()

backend/app/services/browser_session_service.py

@@ -68,6 +68,34 @@ class BrowserSessionService:
         self._last_event_at: dict[str, float] = {}
         self._evict_task: Optional[asyncio.Task[None]] = None
 
+    def _browser_launch_kwargs(self, width: int, height: int) -> dict[str, Any]:
+        return {
+            "headless": get_settings().browser_headless,
+            "viewport": {"width": width, "height": height},
+            "color_scheme": "dark",
+            "locale": "zh-CN",
+            "timezone_id": get_settings().tz,
+            "ignore_default_args": ["--enable-automation"],
+            "args": [
+                "--no-sandbox",
+                "--disable-dev-shm-usage",
+                "--disable-blink-features=AutomationControlled",
+                "--window-size=%d,%d" % (width, height),
+            ],
+        }
+
+    async def _install_browser_init_scripts(self, context: Any) -> None:
+        await context.add_init_script("""
+        (() => {
+          try {
+            Object.defineProperty(navigator, 'webdriver', { get: () => undefined });
+            Object.defineProperty(navigator, 'languages', { get: () => ['zh-CN', 'zh', 'en-US', 'en'] });
+            Object.defineProperty(navigator, 'plugins', { get: () => [1, 2, 3, 4, 5] });
+            window.chrome = window.chrome || { runtime: {} };
+          } catch (_) {}
+        })();
+        """)
+
     async def create(
         self,
         custom_page_id: int,
@@ -113,11 +141,9 @@ class BrowserSessionService:
 
             context = await self._playwright.chromium.launch_persistent_context(
                 str(self._profile_dir(profile_key)),
-                headless=get_settings().browser_headless,
-                viewport={"width": width, "height": height},
-                color_scheme="dark",
-                args=["--no-sandbox", "--disable-dev-shm-usage"],
+                **self._browser_launch_kwargs(width, height),
             )
+            await self._install_browser_init_scripts(context)
             await self._restore_session_state(context, profile_key)
             # Grant clipboard access for the page origin
             try:
@@ -773,11 +799,9 @@ class BrowserSessionService:
             profile_key = f"auth-capture-{session_id[:12]}"
             context = await self._playwright.chromium.launch_persistent_context(
                 str(self._profile_dir(profile_key)),
-                headless=get_settings().browser_headless,
-                viewport={"width": width, "height": height},
-                color_scheme="dark",
-                args=["--no-sandbox", "--disable-dev-shm-usage"],
+                **self._browser_launch_kwargs(width, height),
             )
+            await self._install_browser_init_scripts(context)
             # Grant clipboard access for the page origin
             try:
                 parsed = urlparse(url)

backend/app/services/upstream_client.py

@@ -2,6 +2,7 @@
 from __future__ import annotations
 
 import json
+import time
 from typing import Any, Optional
 from urllib.parse import urljoin
 import httpx
@@ -13,6 +14,9 @@ class UpstreamError(RuntimeError):
     pass
 
 
+NEW_API_DEFAULT_QUOTA_PER_UNIT = 500000
+
+
 def _find_token(value: Any) -> str:
     if isinstance(value, str) and value.count(".") >= 2:
         return value
@@ -74,6 +78,8 @@ def mask_secret(value: Any) -> str:
 def _unwrap_data(value: Any) -> Any:
     if isinstance(value, dict) and "data" in value and ("code" in value or "message" in value):
         return value.get("data")
+    if isinstance(value, dict) and "data" in value and "success" in value:
+        return value.get("data")
     return value
 
 
@@ -105,6 +111,20 @@ def _extract_key_value(value: Any) -> str:
     return ""
 
 
+def _is_success_response(value: Any) -> bool:
+    if not isinstance(value, dict) or "success" not in value:
+        return True
+    return value.get("success") is True
+
+
+def _response_message(value: Any, fallback: str = "") -> str:
+    if isinstance(value, dict):
+        msg = value.get("message") or value.get("detail")
+        if msg:
+            return str(msg)
+    return fallback
+
+
 def _unwrap_list(value: Any) -> Optional[list[dict[str, Any]]]:
     def _normalize(lst: list) -> list[dict[str, Any]]:
         out = []
@@ -288,6 +308,17 @@ class UpstreamClient:
         prefix = f"/{self.api_prefix}" if self.api_prefix else ""
         return f"{self.base_url}{prefix}/{path.lstrip('/')}"
 
+    def _is_new_api_user_mode(self) -> bool:
+        login_path = str(self.auth_config.get("login_path") or "")
+        return (
+            self.api_prefix == ""
+            and (
+                bool(self.auth_config.get("new_api_user"))
+                or login_path == "/api/user/login"
+                or self.auth_type == "cookie"
+            )
+        )
+
     def _headers(self, auth: bool = True) -> dict[str, str]:
         headers: dict[str, str] = {
             "Accept": "application/json",
@@ -348,6 +379,119 @@ class UpstreamClient:
             raise UpstreamError(f"{method} {path} returned HTML, not JSON")
         return resp.json()
 
+    def _ensure_api_success(self, payload: Any, action: str) -> None:
+        if not _is_success_response(payload):
+            raise UpstreamError(_response_message(payload, f"{action} failed"))
+
+    def _new_api_quota_per_unit(self) -> int:
+        try:
+            payload = self._request("GET", "/api/status", auth=False)
+            data = _unwrap_data(payload)
+            if isinstance(data, dict):
+                value = data.get("quota_per_unit")
+                quota_per_unit = int(float(value))
+                if quota_per_unit > 0:
+                    return quota_per_unit
+        except Exception:
+            pass
+        return NEW_API_DEFAULT_QUOTA_PER_UNIT
+
+    @staticmethod
+    def _normalize_key_record(record: dict[str, Any]) -> dict[str, Any]:
+        out = dict(record)
+        if not out.get("name") and out.get("key_name"):
+            out["name"] = out.get("key_name")
+        if not out.get("group_id") and out.get("group"):
+            out["group_id"] = str(out.get("group"))
+        if not out.get("group_name") and out.get("group"):
+            out["group_name"] = str(out.get("group"))
+        return out
+
+    def _list_new_api_tokens(
+        self,
+        search: str = "",
+        group_id: str | int | None = None,
+    ) -> list[dict[str, Any]]:
+        if search:
+            path = "/api/token/search"
+            params = {"keyword": search, "token": "", "p": 1, "size": 100}
+        else:
+            path = "/api/token/"
+            params = {"p": 1, "size": 100}
+        resp = self._client.request(
+            "GET",
+            self._url(path),
+            params=params,
+            headers=self._headers(),
+            cookies=self._cookies,
+        )
+        self._cookies.update(dict(resp.cookies))
+        resp.raise_for_status()
+        data = resp.json()
+        self._ensure_api_success(data, "list New-API tokens")
+        nested = _unwrap_data(data)
+        items: list[dict[str, Any]] | None = None
+        if isinstance(nested, list):
+            items = [i for i in nested if isinstance(i, dict)]
+        elif isinstance(nested, dict):
+            for key in ("items", "tokens", "list", "records"):
+                value = nested.get(key)
+                if isinstance(value, list):
+                    items = [i for i in value if isinstance(i, dict)]
+                    break
+        if items is None:
+            raise UpstreamError("unexpected New-API token list response")
+        normalized = [self._normalize_key_record(i) for i in items]
+        if group_id is not None:
+            gid = str(group_id)
+            normalized = [i for i in normalized if str(i.get("group_id") or i.get("group") or "") == gid]
+        return normalized
+
+    def _get_new_api_token_key(self, token_id: str | int) -> str:
+        payload = self._request("POST", f"/api/token/{token_id}/key")
+        self._ensure_api_success(payload, "get New-API token key")
+        key_value = _extract_key_value(_unwrap_data(payload))
+        if not key_value:
+            raise UpstreamError("New-API token key response did not include key")
+        return key_value
+
+    def _create_new_api_token(
+        self,
+        name: str,
+        group_id: str | int,
+        quota: float = 0,
+        expires_in_days: int | None = None,
+    ) -> dict[str, Any]:
+        unlimited = quota <= 0
+        body: dict[str, Any] = {
+            "name": name,
+            "remain_quota": 0 if unlimited else int(round(quota * self._new_api_quota_per_unit())),
+            "unlimited_quota": unlimited,
+            "expired_time": int(time.time()) + expires_in_days * 86400 if expires_in_days else -1,
+            "model_limits_enabled": False,
+            "model_limits": "",
+            "allow_ips": "",
+            "group": str(group_id),
+            "cross_group_retry": False,
+        }
+        payload = self._request("POST", "/api/token/", body)
+        self._ensure_api_success(payload, "create New-API token")
+
+        matches = self._list_new_api_tokens(search=name, group_id=group_id)
+        token = next((i for i in matches if str(i.get("name") or "").strip() == name.strip()), None)
+        if not token:
+            raise UpstreamError("New-API token was created but could not be found by name")
+        token_id = token.get("id")
+        if token_id is None:
+            raise UpstreamError("New-API token list response did not include id")
+        key_value = self._get_new_api_token_key(token_id)
+        return {
+            "id": str(token_id),
+            "key": key_value,
+            "masked_key": mask_secret(key_value),
+            "raw": self._normalize_key_record(token),
+        }
+
     def login(self) -> None:
         if self.auth_type != "login_password":
             return
@@ -412,6 +556,9 @@ class UpstreamClient:
         endpoint: str = "/keys",
     ) -> list[dict[str, Any]]:
         """查询远端上游 Key 列表，支持按名称搜索、分组筛选、状态筛选。"""
+        if endpoint in {"/api/token", "/api/token/"} or (endpoint == "/keys" and self._is_new_api_user_mode()):
+            return self._list_new_api_tokens(search=search, group_id=group_id)
+
         params: dict[str, Any] = {}
         if search:
             params["search"] = search
@@ -446,7 +593,7 @@ class UpstreamClient:
             for key in ("items", "keys", "list", "records"):
                 val = data.get(key)
                 if isinstance(val, list):
-                    return val
+                    return [self._normalize_key_record(i) for i in val if isinstance(i, dict)]
         raise UpstreamError(f"unexpected keys response type: {type(data).__name__}")
 
     def delete_api_key(self, key_id: str, endpoint: str = "/keys") -> None:
@@ -486,6 +633,14 @@ class UpstreamClient:
         rate_limit_7d: float = 0,
         endpoint: str = "/keys",
     ) -> dict[str, Any]:
+        if endpoint in {"/api/token", "/api/token/"} or (endpoint == "/keys" and self._is_new_api_user_mode()):
+            return self._create_new_api_token(
+                name,
+                group_id,
+                quota=quota,
+                expires_in_days=expires_in_days,
+            )
+
         body: dict[str, Any] = {
             "name": name,
             "group_id": int(group_id) if str(group_id).isdigit() else group_id,