#metabrainz

Heh. Looking at the raw output of the API showing all the tags really makes one wonder what people are thinking. Like seriously, one of the tags for the Various Artists artist... is an url to a Johnny Cash cover art... Like wtf? ^_^;

i hate tags

best part of them: nobody can remove a tag, except the one who entered it or something :p

Hmm. So https://wiki.musicbrainz.org/Cover_Art_Archive/API says that any request to fetch cover art must go through coverartarchive. I assume that does not cover artist cover art since that's handled under url-rels and coverartarchive doesn't have an artist context. Have I understood that right?

it's only valid for coverart of releases that has been uploaded to the CAA

Cheers :) Don't wanna get my code banned because I made some incorrect assumption :)

unless you do too many requests per seconds, it won't get you banned

many releases don't have coverart uploaded to the CAA yet

Don't think that'll be a problem. It uses a built in caching proxy with a 1sec minimum delay per request, so I guess it would require quite a number of users before hitting the rate limit :)

even my little script gets a 503 sometimes :D

but i'm possibly in the "anonymous" group

Well that might happen though I doubt popularity is high enough, but I will obviously have handlers for such situations. I just don't want to have the useragent outright banned

You could set up your own search server :-)

Well if the interest is high enough, I might. But this is mainly because I'm sick and tired of the last.fm agent in Plex...

Not sure I even want to release it to anyone else. Last time I did a couple of years ago, they broke the music API in plex and just left it broken for years >_<

How many users do you expect? :D

Well, I know there is some interest among the users on the Plex forum at least... How many, no idea.

:-)

bed time. good night :-)

https://musicbrainz.org/artist/f2b1690b-7b7a-43... <-- layout seems to break when wikipedia links are too long

d4rkie: Yes - though it may be because MB is not stripping off the # string. Easiest way to fix for this page is it edit the page and take out the # string.

hi! I'm confuse about the API. I want to get the full official discography for an artist, like what's on the artist Overview page on musicbrainz.org... but I can't figure out how to do it without sending multiple requests, one for each release-group

the docs seem to say that this: http://musicbrainz.org/ws/2/release-group?artis... should return both Album and EPs, but it does seem to only return albums?

specifically, the docs say the url above "all albums and EPs by Autechre"

*above should return

noci, can you give an example of something missing?

noci: I see three instances of 'type="EP"' in the XML I get from that query/lookup.

hm o__O

maybe I'm just blind

let me check again xD

you're right, hm, let me get back to my code and see, I must have mixed somehting up...

Should also note, that the reason there's only 3, is because of paging.

[musicbrainz-server] Sophist-UK opened pull request #505: MBS-9323: Fix long URLs distorting page layout. (master...patch-1) https://git.io/v9lkV

d4rkie: Fix submitted for web-page distortion when URL is too long.

noci, http://musicbrainz.org/ws/2/release-group?artis... gives you all 14 on a single page, though I don't suggest fetching 100 but it shows that it's not the query that is incorrect at least :)

yes this is highly embarrassing

I seem to have mixed my queries together

Dude. It beats my literally pulling my hair for three days for why a program wasn't reading a change in the config file... The program runs as a different user entirely... I could have changed the config till I was blue and it couldn't have cared less :)

hah

beats me*

hmm, is "release group" what I should be asking for by the way, and not "release"? (I want the official discography for an artist, like what's on the "Overview" page for an artist on musicbrainz.org)

I *think* I would want to query "release" with status=official instead

That depends. Do you differentiate between say the 12" Vinyl version of 'Anti EP' and the CD version? Or do you perhaps differentiate between the 2 different versions of Envane on 12" Vinyl?

ah no, I don't care about different versions really

Then release group is what you want.

okay

but there is no way to get only official releases on release-group is there?

Define official here?

has status=official instead of status=bootleg

Well you can... But if you make no difference between the individual versions, what does it even matter? That's a status on the releases, not the release-group isn't it?

yeah, if I append status=official on the release-group query I get "status is not a valid parameter unless release are requested"... but, like, if I do a query for release-group on a popular band like Iron Maiden, I get 100+ albums back - 90% bootlegs

No? http://musicbrainz.org/ws/2/release-group?artis... has only 2 results.

or no... That can't be the right

wrong artist ID I think

that's weird, I wonder what yours is

http://musicbrainz.org/ws/2/release-group?artis... that has to be the one. 122 hits... But I'm not seeing any bootlegs

hm

if you compare that to https://musicbrainz.org/artist/ca891d65-d9b0-42...

they have released a lot, but not really 122 albums and eps

I get it to about 50 or so

Right. But the 122 isn't querying for only albums and eps, it's quering for all types

or no, wait, album and ep only is 122 :)

uhm, but you have type=album|ep ?

It's 182 without that :)

ah :P

I think most Live albums are bootlegs in the XML

You can list all of that, if you browse https://musicbrainz.org/artist/ca891d65-d9b0-42... instead

ah yes

Naa. Some are, but far from all

so is there anyway I can do "all=0" ? in the XML response? ;)

I guess I can live with it...

but there will be a lot paging

Not afaik. You could go around it by getting all releases that isn't bootlegs, and include release group info and build an index of that.

yeah that could work, but I would really have to query the API a whole lot... I would like to be nice and query as little as possible

might be better if I just mirror the database

would be interesting to see what query is used to build the Overview page for each artist on musicbrainz.org tho

http://stackoverflow.com/questions/41983347/fil... might be of interest to you

ohum

at least I'm not alone in this endeavour then :D

that last url is almost exactly what I need I think, except I think I need first-release-date

dang

ruaok: ping

zong

and the ticket to fix it is 4 years old #feelsbadman

Sophist_UK: commented on it. break-all should work better, IMO

ruaok: we have a disk space issue on prince & boingo, related to search containers, but it looks something weird

space is mostly used by /var/lib/docker/aufs/

and it keeps growing, i have the feeling something isn't cleaned up properly

<== pre coffee

ruaok: np ;) i'm investigating but for now not sure about this issue and possible solutions (also i'm pre-second coffee)

without looking at it, I'm nearly certain (120% - 147%) that this is your problem. why? you touched the code last.

:-D

ahah, i think it is something more with docker, i have a feeling we have some kind of "leak"

i'll reboot prince to test something

d4rkie: Agreed - I have made the change.

zas: yeah, that was my feeling too. reboot.

Hmm... I must be missing something. https://musicbrainz.org/artist/183d6ef6-e161-47... has an image, but http://musicbrainz.org/ws/2/artist/183d6ef6-e16... shows no image relation? huh?

Ether_Man: The clue is the text below the picture on the web page. This is a picture that MBS automatically extracted from their Wikipedia page (in the same way that Releases can get pictures from Amazon.) I don't think that Artists can have pictures directly associated with them in the way that Releases can, so presumably this is the only way that MBS can get Artist images.

http://musicbrainz.org/ws/2/artist/f27ec8db-af0... Sure they can. Artist image url is rel type 221132e9-e30e-43f2-a741-15afc4c5fa7c

woho, I almost got it working now... playing with the API is fun, I'm surprised you're allowed to insert pure lucene syntax into your query

only thing missing is release date, but I guess I'll have to marry the result with a second query like someone suggested

ruaok: search server volume is unnamed (/home/search/indexes), is it intentional ?

ruaok: i reclaimed the wasted space (140Gb), but i had to stop docker, remove old /var/lib/docker, and start over, does the container sync the indexes on start ?

Ether_Man: The difference is that the Michael Jackson entry explicitly includes a picture URL whilst They Might Be Giants doesn't.

So I was wrong about not being able to add an explicit picture. But that is the explanation as to what is currently happening.

Yea. Just feels... Weird >_<

However, from a technical perspective, if MBS can find the picture itself, why shouldn't it be able to add it as an image entry in the JSON? I suggest you raise an Enhancement Ticket to see if this functionality can be added.

See also MBS-9223 for a similar but different issue about Amazon Cover Art.

MBS-9223: Enhance ws response for ASIN to include cover-art URL https://tickets.metabrainz.org/browse/MBS-9223

zas: the volume being unnamed is for no particular reason.

and yes, it picks-up and re-syncs on startup

Maybe. Want to get at least basic functionality in my Plex Agent first. Was mainly that like, the artist page had an image, was getting annoyed at why my agent was not fetching it :)