fix: 放宽音乐资源标题匹配

app/chain/search.py

@@ -8,6 +8,7 @@ from concurrent.futures import FIRST_COMPLETED, ThreadPoolExecutor, as_completed
 from datetime import datetime
 from typing import AsyncIterator, Any, Dict, Iterable, Tuple
 from typing import List, Optional
+from unicodedata import normalize
 
 from fastapi.concurrency import run_in_threadpool
 
@@ -40,6 +41,7 @@ from app.utils.media import (
     resolve_media_identity,
 )
 from app.utils.string import StringUtils
+from app.utils.zhconv import convert as zhconv_convert
 
 
 class SearchChain(ChainBase):
@@ -60,17 +62,17 @@ class SearchChain(ChainBase):
 
     @classmethod
     def music_site_keywords(cls, music: MetaMusic | MusicInfo) -> list[str]:
-        """按单曲或专辑实体生成站点关键词，避免单曲优先命中所属整专。"""
+        """按实体生成站点关键词，先扩大名称召回，再由结果匹配校验艺术家。"""
         artists = music.artists or []
         artist = artists[0] if artists else music.album_artist
         values: list[Optional[str]] = []
         if getattr(music, "music_type", None) == MUSIC_ENTITY_ALBUM:
             album = music.album or music.title
-            values.extend([f"{artist} {album}" if artist and album else None, album])
+            values.extend([album, f"{artist} {album}" if artist and album else None])
         else:
             values.extend([
-                f"{artist} {music.title}" if artist and music.title else None,
                 music.title,
+                f"{artist} {music.title}" if artist and music.title else None,
             ])
         return cls._unique_music_texts(values)
 
@@ -81,8 +83,8 @@ class SearchChain(ChainBase):
             resource_title: str,
             resource_description: Optional[str] = None,
     ) -> bool:
-        """校验站点资源标题同时包含目标音乐名称和已知艺术家。"""
+        """校验标题与副标题同时命中目标专辑/曲名和艺术家。"""
-        normalized_resource = MetaMusic.compact_text(
+        normalized_resource = cls._normalize_music_match_text(
             f"{resource_title or ''} {resource_description or ''}"
         )
         if not normalized_resource:
@@ -94,23 +96,37 @@ class SearchChain(ChainBase):
             ])
         else:
             candidates = cls._unique_music_texts([music.title])
-        if not any(
+        normalized_candidates = [
-                MetaMusic.compact_text(candidate) in normalized_resource
+            cls._normalize_music_match_text(candidate)
-                for candidate in candidates
+            for candidate in candidates
-                if MetaMusic.compact_text(candidate)
+        ]
-        ):
+        if not any(candidate in normalized_resource for candidate in normalized_candidates if candidate):
             return False
         artists = cls._unique_music_texts([
             music.artist,
             music.album_artist,
             *(music.artists or []),
         ])
-        return not artists or any(
+        normalized_artists = [
-            MetaMusic.compact_text(artist) in normalized_resource
+            cls._normalize_music_match_text(artist)
             for artist in artists
-            if MetaMusic.compact_text(artist)
+        ]
+        return bool(normalized_artists) and any(
+            artist in normalized_resource
+            for artist in normalized_artists
+            if artist
         )
 
+    @staticmethod
+    def _normalize_music_match_text(value: Optional[str]) -> str:
+        """去除音乐名称干扰字符并转换为简体小写文本。"""
+        compact_text = "".join(
+            char
+            for char in normalize("NFKC", str(value or "")).casefold()
+            if char.isalnum()
+        )
+        return zhconv_convert(compact_text, "zh-hans")
+
     @staticmethod
     def _unique_music_texts(values: Iterable[Optional[str]]) -> list[str]:
         """按清理后的文本去重，并保留站点搜索词原始顺序。"""
@@ -1469,7 +1485,7 @@ class SearchChain(ChainBase):
             torrents: Optional[List[TorrentInfo]],
             mediainfo: MusicInfo,
     ) -> List[TorrentInfo]:
-        """筛出音乐分类且标题包含目标单曲或专辑名称的站点资源。"""
+        """筛出音乐分类且标题、副标题匹配目标名称与艺术家的站点资源。"""
         return [
             torrent
             for torrent in torrents or []

tests/test_music_workflows.py

@@ -70,8 +70,8 @@ def test_parse_query_keeps_non_artist_suffix():
     assert meta.title == "思念 - 现场版"
 
 
-def test_build_site_keywords_prefers_artist_album():
+def test_build_site_keywords_searches_album_before_combined_artist():
-    """专辑订阅只按艺术家与专辑名搜索，不混入其中某首单曲。"""
+    """专辑先按专辑名扩大召回，再用艺术家与专辑名组合词回退。"""
     info = MusicInfo(
         music_type="album",
         title="Get Lucky",
@@ -80,13 +80,13 @@ def test_build_site_keywords_prefers_artist_album():
     )
 
     assert SearchChain.music_site_keywords(info) == [
-        "Daft Punk Random Access Memories",
         "Random Access Memories",
+        "Daft Punk Random Access Memories",
     ]
 
 
-def test_build_site_keywords_keeps_recording_out_of_album_search():
+def test_build_site_keywords_searches_track_before_combined_artist():
-    """单曲订阅只按艺术家与曲名搜索，不能优先命中所属整张专辑。"""
+    """单曲先按曲名扩大召回，且关键词不能混入所属专辑名。"""
     info = MusicInfo(
         music_type="recording",
         title="Get Lucky",
@@ -95,17 +95,18 @@ def test_build_site_keywords_keeps_recording_out_of_album_search():
     )
 
     assert SearchChain.music_site_keywords(info) == [
-        "Daft Punk Get Lucky",
         "Get Lucky",
+        "Daft Punk Get Lucky",
     ]
 
 
 def test_album_resource_match_requires_selected_album_title():
-    """专辑订阅只接受包含目标专辑名的站点资源，忽略大小写、空格和标点差异。"""
+    """专辑订阅只接受同时包含目标专辑名和艺术家的站点资源。"""
     album = MusicInfo(
         music_type="album",
         title="Random Access Memories",
         album="Random Access Memories",
+        artists=["Daft Punk"],
         names=["Random-Access Memories"],
     )
 
@@ -122,6 +123,7 @@ def test_recording_resource_match_does_not_treat_album_name_as_track_alias():
         music_type="recording",
         title="Get Lucky",
         album="Random Access Memories",
+        artists=["Daft Punk"],
         names=["Get Lucky", "Random Access Memories"],
     )
 
@@ -145,6 +147,44 @@ def test_resource_match_requires_artist_when_target_artist_is_known():
     assert SearchChain.matches_music_resource(recording, "晴天 FLAC") is False
 
 
+def test_album_resource_match_combines_title_description_and_converts_traditional_chinese():
+    """专辑名与艺术家可分处标题和副标题，繁简及干扰符号不影响匹配。"""
+    album = MusicInfo(
+        music_type="album",
+        title="永远是朋友",
+        album="永远是朋友",
+        artists=["周华健"],
+    )
+
+    assert SearchChain.matches_music_resource(
+        album,
+        "【永遠・是朋友】24bit／96kHz",
+        "專輯藝人：周華健；無損音樂",
+    ) is True
+
+
+def test_recording_resource_match_combines_title_description_and_converts_traditional_chinese():
+    """单曲按曲名与艺术家匹配，并统一繁简及全角符号。"""
+    recording = MusicInfo(
+        music_type="recording",
+        title="晴天",
+        artists=["周杰伦"],
+    )
+
+    assert SearchChain.matches_music_resource(
+        recording,
+        "０１．晴 天［FLAC］",
+        "演唱：周杰倫",
+    ) is True
+
+
+def test_music_resource_match_rejects_target_without_artist():
+    """目标缺少艺术家时不能仅凭同名专辑或单曲放行。"""
+    recording = MusicInfo(music_type="recording", title="晴天")
+
+    assert SearchChain.matches_music_resource(recording, "周杰伦 - 晴天 FLAC") is False
+
+
 def test_normalize_candidates_deduplicates_source_identity():
     """同一来源和媒体 ID 的音乐候选应只保留一次。"""
     results = MediaChain.normalize_music_candidates(