fix(lastfm): clean up Last.fm content by removing "Read more" links from descriptions and bios
Signed-off-by: Deluan <deluan@navidrome.org>
This commit is contained in:
@@ -31,6 +31,12 @@ var ignoredContent = []string{
|
||||
`<a href="https://www.last.fm/music/`,
|
||||
}
|
||||
|
||||
var lastFMReadMoreRegex = regexp.MustCompile(`\s*<a href="https://www\.last\.fm/music/[^"]*">Read more on Last\.fm</a>\.?`)
|
||||
|
||||
func cleanContent(content string) string {
|
||||
return strings.TrimSpace(lastFMReadMoreRegex.ReplaceAllString(content, ""))
|
||||
}
|
||||
|
||||
type lastfmAgent struct {
|
||||
ds model.DataStore
|
||||
sessionKeys *agents.SessionKeys
|
||||
@@ -95,7 +101,7 @@ func (l *lastfmAgent) GetAlbumInfo(ctx context.Context, name, artist, mbid strin
|
||||
resp.MBID = a.MBID
|
||||
resp.URL = a.URL
|
||||
if isValidContent(a.Description.Summary) {
|
||||
resp.Description = strings.TrimSpace(a.Description.Summary)
|
||||
resp.Description = cleanContent(a.Description.Summary)
|
||||
return &resp, nil
|
||||
}
|
||||
log.Debug(ctx, "LastFM/album.getInfo returned empty/ignored description, trying next language", "album", name, "artist", artist, "lang", lang)
|
||||
@@ -171,7 +177,7 @@ func (l *lastfmAgent) GetArtistBiography(ctx context.Context, id, name, mbid str
|
||||
return "", err
|
||||
}
|
||||
if isValidContent(a.Bio.Summary) {
|
||||
return strings.TrimSpace(a.Bio.Summary), nil
|
||||
return cleanContent(a.Bio.Summary), nil
|
||||
}
|
||||
log.Debug(ctx, "LastFM/artist.getInfo returned empty/ignored biography, trying next language", "artist", name, "lang", lang)
|
||||
}
|
||||
|
||||
@@ -80,7 +80,7 @@ var _ = Describe("lastfmAgent", func() {
|
||||
It("returns the biography", func() {
|
||||
f, _ := os.Open("tests/fixtures/lastfm.artist.getinfo.json")
|
||||
httpClient.Res = http.Response{Body: f, StatusCode: 200}
|
||||
Expect(agent.GetArtistBiography(ctx, "123", "U2", "")).To(Equal("U2 é uma das mais importantes bandas de rock de todos os tempos. Formada em 1976 em Dublin, composta por Bono (vocalista e guitarrista), The Edge (guitarrista, pianista e backing vocal), Adam Clayton (baixista), Larry Mullen, Jr. (baterista e percussionista).\n\nDesde a década de 80, U2 é uma das bandas mais populares no mundo. Seus shows são únicos e um verdadeiro festival de efeitos especiais, além de serem um dos que mais arrecadam anualmente. <a href=\"https://www.last.fm/music/U2\">Read more on Last.fm</a>"))
|
||||
Expect(agent.GetArtistBiography(ctx, "123", "U2", "")).To(Equal("U2 é uma das mais importantes bandas de rock de todos os tempos. Formada em 1976 em Dublin, composta por Bono (vocalista e guitarrista), The Edge (guitarrista, pianista e backing vocal), Adam Clayton (baixista), Larry Mullen, Jr. (baterista e percussionista).\n\nDesde a década de 80, U2 é uma das bandas mais populares no mundo. Seus shows são únicos e um verdadeiro festival de efeitos especiais, além de serem um dos que mais arrecadam anualmente."))
|
||||
Expect(httpClient.RequestCount).To(Equal(1))
|
||||
Expect(httpClient.SavedRequest.URL.Query().Get("artist")).To(Equal("U2"))
|
||||
})
|
||||
@@ -535,7 +535,7 @@ var _ = Describe("lastfmAgent", func() {
|
||||
Expect(agent.GetAlbumInfo(ctx, "Believe", "Cher", "03c91c40-49a6-44a7-90e7-a700edf97a62")).To(Equal(&agents.AlbumInfo{
|
||||
Name: "Believe",
|
||||
MBID: "03c91c40-49a6-44a7-90e7-a700edf97a62",
|
||||
Description: "Believe is the twenty-third studio album by American singer-actress Cher, released on November 10, 1998 by Warner Bros. Records. The RIAA certified it Quadruple Platinum on December 23, 1999, recognizing four million shipments in the United States; Worldwide, the album has sold more than 20 million copies, making it the biggest-selling album of her career. In 1999 the album received three Grammy Awards nominations including \"Record of the Year\", \"Best Pop Album\" and winning \"Best Dance Recording\" for the single \"Believe\". It was released by Warner Bros. Records at the end of 1998. The album was executive produced by Rob <a href=\"https://www.last.fm/music/Cher/Believe\">Read more on Last.fm</a>.",
|
||||
Description: "Believe is the twenty-third studio album by American singer-actress Cher, released on November 10, 1998 by Warner Bros. Records. The RIAA certified it Quadruple Platinum on December 23, 1999, recognizing four million shipments in the United States; Worldwide, the album has sold more than 20 million copies, making it the biggest-selling album of her career. In 1999 the album received three Grammy Awards nominations including \"Record of the Year\", \"Best Pop Album\" and winning \"Best Dance Recording\" for the single \"Believe\". It was released by Warner Bros. Records at the end of 1998. The album was executive produced by Rob",
|
||||
URL: "https://www.last.fm/music/Cher/Believe",
|
||||
}))
|
||||
Expect(httpClient.RequestCount).To(Equal(1))
|
||||
|
||||
Reference in New Issue
Block a user