From bfda2ef79704fcd0c1d949f370f02abb18c6f81d Mon Sep 17 00:00:00 2001
From: jxxghp <jxxghp@gmail.com>
Date: Wed, 9 Sep 2026 07:23:26 +0800
Subject: [PATCH] fix(indexer): limit userdata schema fallbacks

---
 app/modules/indexer/__init__.py         | 16 +++++-
 tests/test_indexer_userdata_fallback.py | 67 +++++++++++++++++++++++++
 2 files changed, 81 insertions(+), 2 deletions(-)
 create mode 100644 tests/test_indexer_userdata_fallback.py

diff --git a/app/modules/indexer/__init__.py b/app/modules/indexer/__init__.py
index f64d1953b..3c789a661 100644
--- a/app/modules/indexer/__init__.py
+++ b/app/modules/indexer/__init__.py
@@ -12,7 +12,7 @@ from app.domain.context import Context, SubtitleInfo, TorrentInfo
 from app.foundation import text as text_tools
 from app.foundation.reflection import ModuleHelper
 from app.modules import _ModuleBase
-from app.modules.indexer.parser import SiteParserBase
+from app.modules.indexer.parser import SiteParserBase, SiteSchema
 from app.modules.indexer.spider import SiteSpider
 from app.modules.indexer.spider.haidan import HaiDanSpider
 from app.modules.indexer.spider.hddolby import HddolbySpider
@@ -51,6 +51,14 @@ _SPECIALIZED_SEARCH_ARGUMENTS = {
 
 _UNKNOWN_SEARCH_FAILURE_MESSAGE = "站点请求或页面解析失败"
 
+# NexusPHP、Gazelle 和 Unit3d 覆盖大多数站点，其余模型多为专站适配，
+# 不参与自动探测，避免对同一站点重复请求专用接口。
+_COMMON_SITE_USERDATA_FALLBACK_SCHEMAS = frozenset({
+    SiteSchema.Gazelle.value,
+    SiteSchema.NexusPhp.value,
+    SiteSchema.Unit3d.value,
+})
+
 
 @dataclass(frozen=True)
 class _IndexerSearchRequest:
@@ -758,7 +766,11 @@ class IndexerModule(_ModuleBase):
             if not site_obj.userid and not site.get("public"):
                 tried = {site.get("schema")}
                 for site_schema in self._site_schemas:
-                    if not site_schema.schema or site_schema.schema.value in tried:
+                    if (
+                        not site_schema.schema
+                        or site_schema.schema.value in tried
+                        or site_schema.schema.value not in _COMMON_SITE_USERDATA_FALLBACK_SCHEMAS
+                    ):
                         continue
                     tried.add(site_schema.schema.value)
                     logger.info(f"站点 {site.get('name')} schema {site.get('schema')} 解析失败, "
diff --git a/tests/test_indexer_userdata_fallback.py b/tests/test_indexer_userdata_fallback.py
new file mode 100644
index 000000000..45136b7e3
--- /dev/null
+++ b/tests/test_indexer_userdata_fallback.py
@@ -0,0 +1,67 @@
+# -*- coding: utf-8 -*-
+from types import SimpleNamespace
+from unittest.mock import Mock
+
+from app.modules.indexer import IndexerModule
+from app.modules.indexer.parser import SiteSchema
+
+
+def _build_parser_constructor(schema_value: str, calls: list[str]) -> Mock:
+    """构造记录解析调用的站点解析器桩。"""
+    constructor = Mock(name=f"{schema_value}Parser")
+    constructor.schema = SimpleNamespace(value=schema_value)
+
+    def build_parser(**_kwargs):
+        """构造一个解析失败且不发起真实请求的解析器实例。"""
+        parser = Mock()
+        parser.userid = None
+        parser.username = None
+        parser.user_level = None
+        parser.join_at = None
+        parser.upload = 0
+        parser.download = 0
+        parser.ratio = 0
+        parser.bonus = 0
+        parser.seeding = 0
+        parser.seeding_size = 0
+        parser.seeding_info = []
+        parser.leeching = 0
+        parser.leeching_size = 0
+        parser.message_unread = 0
+        parser.message_unread_contents = []
+        parser.err_msg = None
+        parser.parse.side_effect = lambda: calls.append(schema_value)
+        return parser
+
+    constructor.side_effect = build_parser
+    return constructor
+
+
+def test_refresh_userdata_fallback_only_uses_common_schemas(monkeypatch):
+    """声明模型失败时只尝试常用模型，不触发专用模型兜底请求。"""
+    calls = []
+    parser_constructors = [
+        _build_parser_constructor(SiteSchema.NexusPhp.value, calls),
+        _build_parser_constructor(SiteSchema.Gazelle.value, calls),
+        _build_parser_constructor(SiteSchema.Unit3d.value, calls),
+        _build_parser_constructor(SiteSchema.TNode.value, calls),
+        _build_parser_constructor(SiteSchema.HDDolby.value, calls),
+    ]
+    monkeypatch.setattr(IndexerModule, "_site_schemas", parser_constructors)
+
+    module = object.__new__(IndexerModule)
+    result = module.refresh_userdata({
+        "name": "音乐乌托邦",
+        "url": "https://www.musopia.vip/",
+        "schema": SiteSchema.NexusPhp.value,
+        "public": False,
+    })
+
+    assert result.userid is None
+    assert calls == [
+        SiteSchema.NexusPhp.value,
+        SiteSchema.Gazelle.value,
+        SiteSchema.Unit3d.value,
+    ]
+    assert not parser_constructors[3].called
+    assert not parser_constructors[4].called