fix lastfm extraction regression and honor no-db

This commit is contained in:
2026-04-21 20:40:28 +02:00
parent 4c7e6f5792
commit dfa8095e1d
3 changed files with 147 additions and 38 deletions

View File

@@ -552,6 +552,7 @@ func main() {
os.Exit(1)
}
defer func() { _ = mainApp.Close() }()
mainApp.IgnoreDB = gopts.noDB
title, tracks, err := fetchLastFMPlaylist(ctx, cfg.Session.Downloads.VerifySSL, opts.PlaylistURL)
if err != nil {
@@ -1422,12 +1423,13 @@ type resolvedLastFMTrack struct {
}
var (
lastFMTitleTagsRe = regexp.MustCompile(`<a\b[^>]*\btitle=(?:"([^"]+)"|'([^']+)')`)
lastFMTotalTracksRe = regexp.MustCompile(`data-playlisting-entry-count="(\d+)"`)
lastFMPlaylistTitleRe = regexp.MustCompile(`<h1[^>]*class="[^"]*playlisting-playlist-header-title[^"]*"[^>]*>([^<]+)</h1>`)
lastFMMirrorTitleRe = regexp.MustCompile(`^Title:\s*(.+?)\s+\|`)
lastFMMirrorLinkTextRe = regexp.MustCompile(`\[([^\]]+)\]\(`)
errLastFMInvalidSource = "unsupported source"
lastFMTitleTagsRe = regexp.MustCompile(`<a\b[^>]*\btitle=(?:"([^"]+)"|'([^']+)')`)
lastFMDataTrackArtistRe = regexp.MustCompile(`data-track-name=(?:"([^"]+)"|'([^']+)')[^>]*data-artist-name=(?:"([^"]+)"|'([^']+)')`)
lastFMTotalTracksRe = regexp.MustCompile(`data-playlisting-entry-count="(\d+)"`)
lastFMPlaylistTitleRe = regexp.MustCompile(`<h1[^>]*class="[^"]*playlisting-playlist-header-title[^"]*"[^>]*>([^<]+)</h1>`)
lastFMMirrorTitleRe = regexp.MustCompile(`^Title:\s*(.+?)\s+\|`)
lastFMMirrorLinkTextRe = regexp.MustCompile(`\[([^\]]+)\]\(`)
errLastFMInvalidSource = "unsupported source"
)
func addURLToQueue(ctx context.Context, mainApp *app.Main, raw string) bool {
@@ -1828,11 +1830,30 @@ func extractLastFMPlaylistInfo(page string) (string, int, error) {
}
func extractLastFMTitleArtistPairs(page string) []lastFMTrack {
dataPairs := lastFMDataTrackArtistRe.FindAllStringSubmatch(page, -1)
if len(dataPairs) > 0 {
out := make([]lastFMTrack, 0, len(dataPairs))
for _, m := range dataPairs {
title := html.UnescapeString(strings.TrimSpace(firstNonEmpty(m[1], m[2])))
artist := html.UnescapeString(strings.TrimSpace(firstNonEmpty(m[3], m[4])))
if title == "" || artist == "" {
continue
}
out = append(out, lastFMTrack{Title: title, Artist: artist})
}
if len(out) > 0 {
return out
}
}
titles := lastFMTitleTagsRe.FindAllStringSubmatch(page, -1)
out := make([]lastFMTrack, 0, len(titles)/2)
for i := 0; i+1 < len(titles); i += 2 {
titleRaw := strings.TrimSpace(firstNonEmpty(titles[i][1], titles[i][2]))
artistRaw := strings.TrimSpace(firstNonEmpty(titles[i+1][1], titles[i+1][2]))
if strings.EqualFold(titleRaw, "Play on YouTube") || strings.EqualFold(artistRaw, "Play on YouTube") {
continue
}
title := html.UnescapeString(titleRaw)
artist := html.UnescapeString(artistRaw)
if title == "" || artist == "" {