fix(site): recognize Gazelle login state (#6464)

2026-09-08 00:46:57 +08:00 · 2026-08-26 21:09:19 +08:00
parent 88703d645a
commit 0a5b6a637d
2 changed files with 45 additions and 0 deletions
@@ -12,8 +12,29 @@ from app.modules.indexer.parser import SiteParserBase, SiteSchema


 class GazelleSiteUserInfo(SiteParserBase):
+    """Gazelle 站点用户数据解析器。"""
+
    schema = SiteSchema.Gazelle

+    def _parse_logged_in(self, html_text: str) -> bool:
+        """
+        识别 Gazelle 登录态，兼容不展示退出入口的站点首页。
+
+        :param html_text: 站点首页 HTML
+        :return: 已登录返回 True，否则返回 False
+        """
+        html = etree.HTML(html_text)
+        try:
+            if DomUtils.has_child_elements(html) and html.xpath(
+                '//a[contains(@href, "user.php?id=")]'
+            ):
+                return True
+        finally:
+            if html is not None:
+                del html
+
+        return super()._parse_logged_in(html_text)
+
    def _parse_user_base_info(self, html_text: str):
        html_text = self._prepare_html_text(html_text)
        html = etree.HTML(html_text)
@@ -179,3 +179,27 @@ def test_gazelle_seeding_keeps_legacy_column_fallback():
    assert parser.seeding == 1
    assert parser.seeding_size == 2147483648
    assert parser.seeding_info == [[45, 2147483648]]
+
+
+def test_gazelle_login_accepts_current_user_link_without_logout_marker():
+    """Gazelle 首页仅暴露当前用户链接时也应识别为已登录。"""
+    parser = _build_parser()
+    logged_in_html = (
+        '<html><body><a href="user.php?id=1234">tester</a>'
+        '<a href="bonus.php">积分 (42.0)</a></body></html>'
+    )
+
+    assert parser._parse_logged_in(logged_in_html) is True
+    assert parser.err_msg is None
+
+
+def test_gazelle_login_rejects_password_form_without_current_user_link():
+    """Gazelle 登录页不得因普通用户相关文本被误判为已登录。"""
+    parser = _build_parser()
+    login_html = (
+        '<html><body><form action="login.php">'
+        '<input type="password" name="password"></form></body></html>'
+    )
+
+    assert parser._parse_logged_in(login_html) is False
+    assert parser.err_msg == "未检测到已登陆，请检查cookies是否过期"