2,054,148 Pages

LF instrumentals

So last week, after noticing UserBot was back creating LF pages, I thought to myself, if its contributions are visible through API but not frontend, then maybe I can block it through API even though frontend claims no such user exists. Tried it, it worked, and page creations stopped. Then I started deleting all lyricless LF pages (as I'm sure you noticed).

You can imagine I wasn't very happy when I noticed today that UB was back creating pages… even though it's still blocked (and block won't expire until tomorrow). So either bots aren't affected by blocks (but at least ÜberBot is), or someone "fixed" it to run anyway (despite Sean's claim that no one knows anything about this script)…

So, since deleting lyricless LF pages is pointless, and redirecting them to the correct ns0 page (which in many cases doesn't even exist) is more trouble than it's worth, how about the following hack (wouldn't want to call it a solution): create a dummy LF page, e.g. LyricFind:Instrumental, and redirect all lyricless LF pages to that? The idea makes me cringe, but if LF namespace will eventually be deleted anyway (note I'm writing "if", not "when"), it's only a temporary pain. What do you think? — 6×9 (Talk) 16:57, June 24, 2014 (UTC)

  • Yes, I did notice your new attack on ns 222 :). Now, after your comment, I finally understand the reason, why have you decided that (n+1)th attemtp may be more succesfull than n previous... :).
  • Yes, nothing's changing in wikia stuff - either nobody knows anything or just don't wanna share their knowledge
  • Yes, "solution" hardly may be qualified as impeccable from systematic point of view, but that's exactly why I like it very much ;) Just let me know if you need some Lwt assistance in your job.
Well Done!, --Senvaikis (talk) 20:00, June 24, 2014 (UTC)
We'll see how far I get :-) For now, I'm limiting it to pages with size <= 200 and deleting (now redirecting) those where text between tags is either whitespace-only (vast majority) or contains string "instrumental". Most of what was left after that was redirects on which ÜB had slapped a merge template... — 6×9 (Talk) 20:45, June 24, 2014 (UTC)
Increased maxsize to 500; vast majority still lyricless. Maybe we should keep this page around, just to show to people who want to turn LW into a discography site… — 6×9 (Talk) 14:50, July 1, 2014 (UTC)
Why then don't you want to select LFs by a given max-length/pattern of content between <lyricfind> tags, - with 100% accuracy?. Once again, let me know if Lwt may help you here. Nice sample, btw - we should definitely keep it ;) --Senvaikis (talk) 17:18, July 1, 2014 (UTC)
I set maxsize originally because I wasn't aware of how huge the extra-lyrical content could get and wanted to avoid loading hundreds of pages just to redirect one. But by now I've extended the script a bit, so I might as well run it across the whole set.
Here's what the script currently does:
  • redirect to LF:I if length(lyrics) < 10 OR lyrics == song title OR (contains "instrumental" or "no lyrics") and length < 50 length limit because these words might appear in regular lyrics OR lyric has 3 or less unique chars quite a number of pages had "asdasdasd", "sdfsdfsdf" or "qweqweqwe"… not sure if LF is trolling us, or someone is trolling them…
  • else try to find (main) page, if found:
    • if lyrics are identical, redirect LF to (main)
      • if LF has songwriter="…" and (main) has no CreditBox, add CB to (main)
    • else if LF doesn't have tl:merge, add it to LF and (main) page
Any more checks I could add? Maybe I should look for GN pages of the same name while I'm at it, and delete them if lyrics are identical.
Oh, and if Lwt is bored :-), it could start to tackle Category:Requests For Moves – if artist & title match an external source, move the page & add templates. — 6×9 (Talk) 08:14, July 2, 2014 (UTC)
Yep, that's a Job - respect(condolence)!
As always, the most interesting for me are such "plain & simple" parts as
  • "find the matching main page" (matching with what precision? what is your algorithm for such matching?)
  • "check if lyrics are identical" (the same Q)
  • "add credits from LF" (should bot trust all LF credits or make some recheck? Should it add links to author LW page, if available? (There are lot of LF-authors, having lw-pages, but under different name)... etc)
I've reviewed some your recent contribs trying to find an answers to above questions, and seems to me you still may tighten your "matching conditions". So, stripping spaces and lines, matching some patterns before checking your condition for moving to LF:I would lead to increasing the number of moved pages (see LyricFind:Elton John:Highlander). The same should be said about the checking of lyrics identity. Seems you treat this identity literally. But do we really need this tiresome balast of merging just for some negligible differences in lyrics punctuation? (see Electric Light Orchestra:Second Time Around).
--Senvaikis (talk) 15:18, July 2, 2014 (UTC)
Main priority (at first) was just the lyricless pages, all the other stuff was added and tweaked while I ran it, so the code is looks horrible right now. Could be worse though – max. indent level is 7.
  • Finding main page: follow any redirects; first a direct match, then inicap each word, if song title has "(" strip that (from right to left), if artist is redirect get actual artist and try all of the above again with that one.
  • Identical lyrics: I've tweaked the comparison to only compare alphanumericals, apostrophes (replacing ’ ´ ‘ `) and whitespace. I guess I could go one further and treat those with Levenshtein distance / length < 0.01 as identical :-) ELO was unfortunate example, btw, because a couple words are different as well. Haven't tested if it's below 0.01 though :-)
  • Credits: Maybe it's best to just ignore them entirely; LF is way too unreliable. As soon as there are two songs by different artists with same title, they're almost sure to mix up either lyrics, credits or both. — 6×9 (Talk) 16:39, July 2, 2014 (UTC)
  • #2: Yes, example was unhappy, even after my attempts to avoid mistake, checking texts with minimal possible leva,b value ;). Well, let's take another sample then: if you strip all punktuation from the lyrics, result will be identical for both lw & lf versions: "weightlessmybabysasleepsomethingsjustdisappearweightlessmybabysasleepfloatingdreamingnodangerjustsweet"
  • #3: I'd like such (or similar) decision. For example, we could define lf credits merging requirement as optional in such cases, when lf author doesn't have lw page. Above Enigma sample may be taken as an example of such situation: almost all possible aliases of Michael Cretu are used by lf (Cretu, Michael Cretu, Curly M.C...), making correct (lw-linked) merge of such credits de facto impossible (at least for bot).
But then our instruction for Hercules should at least mention a posibility of such mitigation... :)

--Senvaikis (talk) 13:28, July 3, 2014 (UTC)

#2: I do that sort of "lax" comparison when checking if lyric == song title; I deliberately didn't use it for comparing lyrics, because I don't want to redirect well-formed lyrics to badly-formed (no line breaks, all lowercase etc.) ones. The sanitizing function could use further tweaking though (like ignoring lines with words like "chorus" or "verse" and little to nothing else). Running both checks should be a good way of finding pairs with near-identical but differently formatted lyrics though.
#3: Plus, LF credits are often (but not always) all-uppercase, and mostly (but not always) in the format "Lastname, Firstname Middlename1 Middlename2 etc." I suspect they sometimes make middle names up just to make it look more impressive; anyway we don't have an artist page for John Davies Cale, and usually separated by " / " but sometimes only "/" (though that's the smallest problem)…
Btw, I finally got myself a bot account – pity 101010 was already taken… — 6×9 (Talk) 16:32, July 3, 2014 (UTC)
Maybe 0.01 is too strict… LyricFind:*NSYNC:Crazy for You vs. *NSYNC:Crazy For You has 0.019 and differences are negligible. (Yeah, sample of 1, very scientific…) I'll increase it to 0.02 and see how it goes…
LF linebreaks aren't always the bee's knees either where does that saying even come from??, if lyrics don't already have paragraphs they seem to insert a blank for each 4 lines – see LF vs. main. — 6×9 (Talk) 18:23, July 3, 2014 (UTC)
Your last observation about lf-lyrics "format" is one more argument in favor of simplified (stripped) comparison imo. I'd better agree to add/remove a few punctuation marks in existing lyrics instead of tedious merging, especially keeping in mind that lf's formatting not necessarily should be correct and therefore just can't be accepted without rechecking. What is the point then to leave such merge request?. Well, in the end - it's your affair, I just said my opinion.
That was nice idea to ask me about the genesis of cliche "bee's knees" - I've had to learn its meaning for beginning :)
Congrats with running a bot finally! Hope that should remarkably increase understanding between your double and my dual identities :)
I know you don't like to disclose any personal info, but sometimes I'm also curious, just contrary to Rx, I'd like to know your profession, not age. A gallery of odd musical instruments in your profile shouldn't mislead nobody - your math knowledge is too good for musician. But your inclination to philosophy and specific sense of humor is hardly imaginable for mathematician. If you were a programmer, you'd run 2A earlier... Remains only one choice - you are natur-philosoph, or just physicist. Am I right?
nite, --Senvaikis (talk) 21:56, July 3, 2014 (UTC)
Turns out the sanitizing function already strips blank lines… side effect of ignoring lines with only 2 chars or less. So at least my subconscious was listening to you :-) Still don't want to ignore line breaks completely, since some lyrics are just blobs of text without any; now I ignore them unless LF has at least twice as many (again, ignoring blank lines).
After setting the bot flag, semanticdrifter warned me to "make sure that it does not achieve sentience"; I replied that our expierence with semi-sentient bots so far was positive :-) Dunno if they got the hidden joke; after all, there is no cyborg flag yet…
Yep, I'm definitely not a musician. (Maybe odd that, the way I love music, I don't play any instrument…) That link is mainly there because I'd found the site shortly before I created my userpage, and I've been to lazy to update it since. All other suggestions are partially right, or not too far off the mark – I love mathematics, but not nearly enough to become a mathematician; I once studied chemistry, but that didn't work out for various reasons; and now I'm a sysadmin, but the whole IT dept. having only 3 people (including me) means I have to do a little bit of everything (which I actually like that way).
Hmm, "2A" could be short for "to answer", as in "… the question"; wonder if DA was aware of that… — 6×9 (Talk) 08:03, July 4, 2014 (UTC)

Imcomplete lyrics

Hi. I checked this page Kix:She Dropped Me The Bomb and the lyrics are imcomplete. I think i saw in another lyric an warning like: "this lyrics is incomplete. you can help complete". Is it possible to add a warning? If yes, how can i do? Thanks. Songsbr (talk) 04:10, July 4, 2014 (UTC)

You may use template {{Partial}} in such cases, --Senvaikis (talk) 04:48, July 4, 2014 (UTC)
o/ thanks xD Songsbr (talk) 06:05, July 4, 2014 (UTC)

API caching

Is there a way to bypass/purge the API cache? So that when I update a page and then fetch it again a short while later, I don't get the pre-update version? "action=purge" doesn't help… I'd like to avoid cases like this one when processing two LF pages for the same song. Any ideas? — 6×9 (Talk) 17:50, July 12, 2014 (UTC)

I'm not sure if that's universal "solution", but in most cases adding some unique fake attribute to a query should help.
I use "&qq=" & Now.Ticks.ToString for that :) --Senvaikis (talk) 18:22, July 12, 2014 (UTC)
Wow, thanks! That trick did the trick! Since I'm using the pywikibot framework I had to change one of its methods; gotta remember that when I update it! — 6×9 (Talk) 18:51, July 12, 2014 (UTC)

Album Covers

Thank you for fixing the graphics for the 2 'We're About 9' albums I added. I was trying to get the formatting correct but every time I tried, something else went wrong.

Jctobin (talk) 18:22, July 13, 2014 (UTC)jctobin

You are welcome! Feel free to ask any questions or help if needed.
Happy editing, --Senvaikis (talk) 19:19, July 13, 2014 (UTC)

Bruno Lauzi:Io Canterò Politico

Hi! The link to Amazon not works, can you fix it, please? --Dante81 (talk) 23:06, July 15, 2014 (UTC)

No, I can fix only links to existing external sources - don't know where this asin was taken from, but it's definitely invalid, thus should have been just removed.
Hope you don't mind some albs, added to BL page as a small compensation for deleted asin :) --Senvaikis (talk) 08:41, July 16, 2014 (UTC)
P.S. Sorry, - haven't noticed your link to the source asin was taken from :). That makes things clearer - the link leads to Italian Amazon site ( Unfortunatelly, contrary to {{iTunes}} , external links to national Amazon sites aren't (yet?) supported in {{asin}}, thus you should limit your amazon-links to only.
Ok, I see. Thank you for reply and thank you very much for the albums :) --Dante81 (talk) 13:37, July 16, 2014 (UTC)

Loretta Goggi

But the title of the song is "Vieni via con me (Taratapunzi-e...)", see the original single. "Vieni via con me" is the album title instead, and the short title of the song. Is good if the song page have the right title. (Note: the 2008 album title is "I grandi successi originali") --Dante81 (talk) 23:50, July 17, 2014 (UTC)

  • Yes, the title of the single song was "Vieni via con me (Taratapunzi-e...)", and I do know that :). But later this song has been included into multiple releases, - sometime under the same, but more and more often with the time passing by - under a shortened title (see wp, mb, amg, iT, offsite etc). You definitely know one of the main LW pagenaming policy rules: "One song - one page". So we should decide which title should be used for such song page. You may disagree, but song searchability should be considered as one of the most deciding factor in such choice imo. The higher probability of searching under particular title - the higher searchability of page, using this title. That's why I decided to redirect VVCM(T) to VVCM, not vice versa. I'm not a big expert in Italian pop music, thus may be wrong. If you have some solid arguments that redirect should be reverted - feel free to do that by yourself or just let me know :).
  • The similar situation is with the title of mentioned compilation. If you have noticed, album cover title contains two different-colored parts: "I grandi successi" & "originali". Usually such technique is used to separate album title from some additional notation ("Originals", "Remastered" etc), which should not always be included into album page title - that's why it was ommited. But considering existence of another same-titled 2008 compilation, seems this decision wasn't the best :). More over - now I'm not even sure if "...originali" was released in 2008 - some sources (amz, simpatyrecords etc) point to 2002... So, I agree that this album should be renamed. Would you mind to do that?
Thanks for collaboration and these notes, and happy editing, --Senvaikis (talk) 07:58, July 18, 2014 (UTC)
1) I don't know which is the right solution but this is the Wikipedia page for the single. I think the songs should have the right complete title and other titles redirects there. 2) "I grandi successi originali" is a large collection of compilations, I think they should have the same (right) title. Ok, now I'll move it. Thank you and sorry for my English. --Dante81 (talk) 15:54, July 18, 2014 (UTC)

Tom Hooker

Hi. This singer called Tom Hooker (!?!) released in 1985 a single called "Swiss Boy" but he used the name Lou Sern. I have to create a new page to Lou Sern or I add in the page Tom Hooker? Can u please tell how it's the correct way because its my first here with an aliase I don't know if is Tom Hooker:Swiss Boy or Lou Sern:Swiss Boy then the page Lou sern will be redirect... i can't find a help page about this.. thank you. Songsbr (talk) 22:40, July 20, 2014 (UTC)

Both we aren't very good at English, thus I just made an example for you, - hope it will be more useful than my long explanations :).
Here's a short list of things you'd notice:
  • creating album/songs under alias name
  • using "Alias" in {{Album}}
  • using "alias" and "albumartist" in {{Song}}
  • using "albumartist" in {{Album Art}}
  • creating redirects "alias" -> "artist"
hth, --Senvaikis (talk) 05:57, July 21, 2014 (UTC)
Again, thank you so much for your help =D Songsbr (talk) 16:01, July 21, 2014 (UTC)

Difficulty contacting...

Hi there V - having some difficulty contacting you re: adding to my pages. Can you get in touch? Cheers - 10:38, July 30, 2014 (UTC)

Community content is available under Copyright unless otherwise noted.