danvics/soulsync
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

Add raw non-ASCII query for CJK track search on Soulseek

Browse source 
When artist or title contains non-ASCII characters (Japanese, Chinese,
Korean, etc.), prepend the original un-romanized text as the first
search query. unidecode converts Japanese kanji to Chinese pinyin
(e.g. "藤澤慶昌" → "wu zhi zhuan sheng") which never matches on
Soulseek. The original characters match filenames directly.

Romanized fallback queries are still generated after for coverage.
Zero impact on ASCII-only tracks (isascii check skips them).
This commit is contained in:
Broque Thomas 2026-03-29 18:34:39 -07:00
parent 3c47281ce5
commit 7a2bc49458
1 changed files with 13 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
core/matching_engine.py
View file

@ -420,15 +420,25 @@ class MusicMatchingEngine:
Returns queries in order of preference (cleaned titles first, then original). Returns queries in order of preference (cleaned titles first, then original).
""" """
queries = [] queries = []
if not spotify_track.artists: if not spotify_track.artists:
# No artist info - just use track name variations # No artist info - just use track name variations
queries.append(self.clean_title(spotify_track.name)) queries.append(self.clean_title(spotify_track.name))
return queries return queries
# If artist or title contains non-ASCII (e.g. Japanese, Chinese, Korean),
# add a raw query first — Soulseek filenames often use original characters,
# and unidecode mangles CJK text into wrong romanizations (Chinese pinyin for Japanese kanji).
raw_artist = spotify_track.artists[0].strip()
raw_title = spotify_track.name.strip()
if raw_artist and raw_title and not (raw_artist + raw_title).isascii():
raw_query = f"{raw_artist} {raw_title}".strip()
queries.append(raw_query)
logger.debug(f"NON-ASCII: Raw original query: '{raw_query}'")
artist = self.clean_artist(spotify_track.artists[0]) artist = self.clean_artist(spotify_track.artists[0])
original_title = spotify_track.name original_title = spotify_track.name
# Get album name if available - try multiple attribute names # Get album name if available - try multiple attribute names
album_name = None album_name = None
for attr in ['album', 'album_name', 'album_title']: for attr in ['album', 'album_name', 'album_title']: