Lyrics: LyricWiki and capitalization

[ghost]

[15:07:51] About the lyrics plugin, there is something quite annoying with french songs. On lyricswiki people sometimes use capital letters for all the words of the song title, while on musicbrainz only the first word of the title has a capital letter. The result is that beets doesn't find the song "Abc def", because on lyricswiki it is "Abc Def". [15:27:36] Actually it is not "sometimes", it seems to be standart: http://lyrics.wikia.com/LyricWiki_talk:Page_Names#Capitalization [16:39:08] Skeptix: Interesting... so we should capitalize all strings before querying LW? [16:47:47] sampsyo: I am not sure, it seems quite messy. Some artists follow LW's guidelines, some don't... [16:48:19] Ugh; that's a drag. [16:49:27] sampsyo: I guess an option that would capitalize would be interesting, so that with two pass (one with and the other without) you would manage to get pretty much everything [16:51:00] Right; I suppose that might help, at the cost of more time spent when the song is not present.

[Kraymer]

@obcd pas directement lié, mais as tu essayé le google backend, particulièrement efficace en ajoutant des sources francophones telles que lacoccinelle.net ou paroles.net ?

[ghost]

@KraYmer J'ai essayé sans rien ajouter et ça m'en a trouvé pas mal. C'est excellent, merci ! Sinon pour ajouter des sources francophones, il faut les ajouter en moteurs de recheche sur le site de Google, c'est ça ?

[Kraymer]

@obcd yes, you can create your own custom search engine and add (more) french sources, then put your _google_engineID in config.yaml

FYI, the plugin use a default search engine that is configured to search only following sites:

• http://www.chartlyrics.com
• http://www.songmeanings.net/
• http://paroles.net
• http://www.lyricsontop.com/
• http://www.lacoccinelle.net
• http://songlyrics.com/
• http://smartlyrics.com/
• http://www.lyrics007.com/
• http://lyrics.wikia.com/
• http://www.elyricsworld.com/

So the current list is biased towards EN+FR results as these are the langages i'm interested in. From my current limited experience, the more the better. The only --but important-- prerequisite before adding permanently a site is to check that its content is scrapeable.

So, my conclusions :

• maybe we could extend the whitelist to contain sites covering more langages
• unit tests would be useful to check that listed sites are, indeed, scrapeable
• beets users could create their own custom search engine and make requests to integrate site in default plugin engine whitelist after they have checked the site is scrapeable?

[sampsyo]

It seems that the Google and Lyrics.com backends may suffice here. (The latter, fortunately, is case-insensitive!)