[Eden] CU Lyrics unusable?

Hi,

I am currently playing with the lyrics plugin, and found that it works, but only if the song title matches exactly(!) the entry on the CU lyrics site. If there is any additional String in the song name, for example (Radio Edit), or something else, or a minor misspelling, it finds simply nothing.

Is it possible to manually alter the searchstring? Or make thw whole search a bit more unsharp, like google does by default when it suggests hits, even if you misstyped them?

This may seem like a minor issue, but if you take a look at an average music collection, you will find many song titles not exactly tagged like required by the database. is it REM or R.E.M? Kesha or Ke$ha?

Any way to make this more usable?

You'd have to do some Python coding.
I made some minor mods to the file
Line 152 generates the search string. I've posted mine here - it's a minor tweak but it helped quite a bit. Still wouldn't snag Ke$sha though lol. Modify that line to what you want.

DDDamian Wrote:You'd have to do some Python coding.
I made some minor mods to the file

Code:

C:\Users\Damian\AppData\Roaming\XBMC\addons\script.cu.lyrics\resources\lib\scrap​ers\lyricstime\lyricsScraper.py

Line 152 generates the search string. I've posted mine here - it's a minor tweak but it helped quite a bit. Still wouldn't snag Ke$sha though lol. Modify that line to what you want.

Code:

[url=http://pastebin.com/6YpKkyDx]http://pastebin.com/6YpKkyDx[/url]

Does this mod work in Dharma v10.1 ?

When I use CU Lyrics, all I get is:
"Fetching lyrics from LyricWiki.org, API Failed

Now, I'm wondering if that just means it can't find a match?

Continuing on my quest to find a working Lyrics (USA English) solution for my new XBMC install.

Use LyricsTime instead of LyricsWiki. Go to Settings>Addons>Enabled Addons>Lyrics and right-click on CU Lyrics to configure.

Modding that line I reference will work on Dharma, pre-Eden or Eden beta. You can basically make the parser behave how you want, but fuzzy logic is a bit beyond my Python skills

And yes, almost as a rule, if you get the "API failed" message it simply means that it didn't find a match.

DDDamian Wrote:Use LyricsTime instead of LyricsWiki. Go to Settings>Addons>Enabled Addons>Lyrics and right-click on CU Lyrics to configure.

Modding that line I reference will work on Dharma, pre-Eden or Eden beta. You can basically make the parser behave how you want, but fuzzy logic is a bit beyond my Python skills

And yes, almost as a rule, if you get the "API failed" message it simply means that it didn't find a match.

Firstly, thanks for recommending LyricsTime scraper. That improved results alot.

I also tried your code ... seems to work fine. It's not that easy to switch back-and-forth between it and the original code, so I'm not sure how much better it works ... but I'll take your word for it.

Can you think of why this filename doesn't match:

07 Rush - 2112; Overture-The Temples of Syrinx

But this one does "The Temples of Syrinx" (at LyricsTime.com search box anyway)

The rest of the Chronicles mostly match ... seems like this one should parse out correctly also.

Glad I could help! You can basically just use my code change - it won't hurt and definitely helps with certain issues, notably anything with "&" or " ' " in it(e.g. George Thorogood & The Destroyers, Don't Fear the Reaper, etc).

For the Rush tune (great pick!) and others, everything depends on how LyricsTime does it's "fuzzy" search, and from what I've seen it ain't exactly google! So no Python scraper in the world is gonna help much there.

Perhaps one of the skilled programmers out there can help this script out by allowing the user to enter a search string containing some of the lyrics, and submit that to LyricsTime!

Sorry I couldn't be of more help, but at least that should improve things for ya a lot!

Cheers, Damian

DDDamian Wrote:Sorry I couldn't be of more help, but at least that should improve things for ya a lot!

Just to verify what I did ...
- I copied your code from that link (from the bottom without the line numbers)
- pasted it into a plain text file.
- backed up the original file
- renamed the new file to lyricsScraper.py and copied it over the old one

To be clear ... you have helped me a lot. I finally have working Lyrics on my new XBMC install. This is great.

And while it's just my preference ... I'd rather the Lyrics to match or not ... but I don't like when it asks me questions (a prompt, pick from a list, etc.)

Thanks again.

It's less than perfect for sure, but it does seem to work ~80-90% at least with my collection. And as long as you placed the code in the LyricsTime directory then yep, you did that right.

Glad I could help - enjoy!

is that something worth including in a CU Lyrics? does it have any downsides?

Hi Amet - glad you saw this outside of the normal CULyrics thread. I have not seen any downside - it's simply an expansion of the parser line.

I found I was having issues with certain artist or song names as listed above, and they formed a reasonable percentage of the failed scrapes.

I have not had any issues as a result of the change, and a considerably higher hit rate, so I would recommend it.

Be nice to be able to give back a bit! Thanks, by the way, for all your incredible work!