mirror of
https://github.com/navidrome/navidrome.git
synced 2026-10-15 05:47:13 +02:00
da0899de2b
UTF8Reader decoded non-BOM input strictly as UTF-8, so a byte that was not valid UTF-8 became the replacement character U+FFFD. Playlists and lyrics saved by Windows editors are commonly Windows-1252/Latin-1, where accented characters are single high bytes (for example é as 0xE9). Those bytes were mangled into U+FFFD, so an M3U entry like 'PokéMon' no longer matched the scanned library path and the track was silently skipped, while pure-ASCII entries in the same playlist imported fine. UTF8Reader now keeps valid UTF-8 untouched and decodes any invalid byte as Windows-1252 (a superset of Latin-1), recovering the accented characters instead of losing them. BOM handling and UTF-16 transcoding are unchanged. Closes #6037 Signed-off-by: Steve <stevejason639@gmail.com> |
||
|---|---|---|
| .. | ||
| import.go | ||
| import_test.go | ||
| parse_m3u.go | ||
| parse_m3u_test.go | ||
| parse_nsp.go | ||
| parse_nsp_test.go | ||
| playlists.go | ||
| playlists_suite_test.go | ||
| playlists_test.go | ||
| rest_adapter.go | ||
| rest_adapter_test.go | ||