beets fails to detect compilations when importing metadata from Deezer

[Czechball]

Problem

Beets is interpreting compilations as albums and is putting all artists into folder names, resulting in extremely long paths Standard import command:

$ beet im

Led to this problem:

Correcting tags from:
    A.Skillz - 2017 Fractal Forest Compilation
To:
    A.Skillz, Chali 2na, CMC & Silenta, Deekline, Defunk, DJ Brace, Fort Knox Five, K+Lab, Krafty Kuts, KWADI, Marten Hørger, Neighbour, OPIUO, Skratch Bastid, Slynk, Stanton Warriors, Stickybuds, The Funk Hunters, Dynamite MC, Father Funk, MC Rtkal, Mystro, Neon Steve, Shad, Malenda, Various Artists - 2017 Fractal Forest Compilation
URL:
    https://www.deezer.com/album/43959881
(Similarity: 81.4%) (tracks, artist, source) (Deezer, 2017, Westwood Recordings)
 * Real Big Deal (Original Mix)                            -> Real Big Deal (Original Mix) (source)
 * 7 Day Weekend (Original Mix)                            -> 7 Day Weekend (Original Mix) (source)
 * Limoncello feat. Shad (Original Mix)                    -> Limoncello feat. Shad (Original Mix) (source)
 * The Things That You Do feat. Father Funk (Original Mix) -> The Things That You Do feat. Father Funk (Original Mix) (source)
 * Right Now (Original Mix)                                -> Right Now (Original Mix) (source)
 * Flanjitsu (Original Mix)                                -> Flanjitsu (Original Mix) (source)
 * Flute Ting feat. MC Rtkal (Original Mix)                -> Flute Ting feat. MC Rtkal (Original Mix) (source)
 * Still Here (Stanton Warriors Shambhala Mix)             -> Still Here (Stanton Warriors Shambhala Mix) (source)
 * You Don't (Original Mix)                                -> You Don't (Original Mix) (source)
 * Shake My Head (Original Mix)                            -> Shake My Head (Original Mix) (source)
 * Don't Go (K+Lab Remix)                                  -> Don't Go (K+Lab Remix) (source)
 * Brazilian Hipster (Original Mix)                        -> Brazilian Hipster (Original Mix) (source)
 * Yeah Thats Right feat. Mystro & Malenda (Original Mix)  -> Yeah Thats Right feat. Mystro & Malenda (Original Mix) (source)
 * Right Right Up feat. CMC&Silenta and Verse Ital         -> Right Right Up feat. CMC&Silenta and Verse Ital (source)
 * Get Involved feat. Defunk                               -> Get Involved feat. Defunk (source)
 * Clap Ya Hands Now feat. KWADI (Original Mix)            -> Clap Ya Hands Now feat. KWADI (Original Mix) (source)

Full -vv output

Setup

• OS: Linux void 5.13.15_1
• Python version: Python 3.9.7
• beets version: beets version 1.5.0
• Turning off plugins made problem go away (yes/no): No

My configuration (output of beet config) is:

directory: /mnt/hdd/beets-export
library: /mnt/hdd/beets.db

plugins: spotify discogs deezer edit

paths:
    default: $albumartist - $album ($year) %if{$catalognum,{$catalognum$} }[Deezer WEB - $format]/$track. $artist - $title
    singleton: $artist - $album ($year) %if{$catalognum,{$catalognum$} [Deezer WEB - $format]/$track. $artist - $title
    comp: Various Artists - $album ($year) %if{$catalognum,{$catalognum$} [Deezer WEB - $format]/$track. $artist - $title
spotify:
    source_weight: 0.5
    mode: list
    tiebreak: popularity
    show_failures: no
    artist_field: albumartist
    album_field: album
    track_field: title
    region_filter:
    regex: []
    client_id: REDACTED
    client_secret: REDACTED
    tokenfile: spotify_token.json
discogs:
    apikey: REDACTED
    apisecret: REDACTED
    tokenfile: discogs_token.json
    source_weight: 0.5
    user_token: REDACTED
    separator: ', '
    index_tracks: no
edit:
    albumfields: album albumartist
    itemfields: track title artist album
    ignore_fields: id path
deezer:
    source_weight: 0.5

[wisp3rwind]

Here is the deezer api response if someone is going to look into this in detail: http://api.deezer.com/album/43959881. A first step would probably to research how deezers contributors and artist should be mapped to "all artists" and "album artist", and whether these are even used consistently in their database. Maybe git blame can tell why the contributors field was chosen for the artist in the first place?

[Czechball]

Right, so this is a Deezer related bug. I'll change the issue name

[wisp3rwind]

Right, so this is a Deezer related bug. I'll change the issue name

Yes, but I don't know whether it's simply beets not interpreting the deezer data very well, or whether the deezer response is hard to interpret unambiguously in the first place.

[ShreyBiswas]

Hi, I'd like to look into this. Just to make sure I understand the issue, is it specifically with compilation videos from the Deezer API? Where the 'Contributors' section holds multiple artists instead of, say, just a 'Various Artists' tag?

[wisp3rwind]

Hi, I'd like to look into this. Just to make sure I understand the issue, is it specifically with compilation videos from the Deezer API? Where the 'Contributors' section holds multiple artists instead of, say, just a 'Various Artists' tag?

Not really, I think. I don't know what the contributors section holds for different kinds of albums. The first problem here is figuring out the exact meaning of these API responses, that's not yet clear. See my comment above on interpreting the contributors and artist. Also, I'm not familiar with either the plugin or the deezer API. Maybe @rhlahuja can chime in here, he/she is the one who wrote the deezer plugin (see #3355).

A first step would really be to figure out whether there's an unambiguous way of telling that we're dealing with a compilation or whether we need a heuristic for that. There doesn't seem to be a compilation flag in the response (although there's a type: album field. Maybe, this shouldn't be tagged as compilation after all? What other values can type have?).Thus, probably you'd need to read the API documentation and/or compare responses for regular albums and compilations to figure out how to approach this.

My guess for this specific album is that it should be tagged with an albumartist of A. Skillz (the artist in the response rather than the first contributor as it is now) and marked as a compilation (although I'm not sure how to infer this from the response. The number of contributors being larger than 1 might not be enough evidence). Does that sound correct to other people here?

[9Mad-Max5]

In most cases, deezer api delivers a main tag in the contributors section. Like in this example: https://developers.deezer.com/api/explorer?url=album/302127 This makes it in the most use cases far more sophisticated over the spotify api. As this doesn't bother at all. I'm not actually sure how to handle it in this case. There have been a lot of contributors flagged as main. But as well, there is a contributor listed with ID 5080. I guess it is translated, but the ID should be sufficient to check if this is part of the contributors to flag the whole album as various artist.

I just stumbled across this issue as I was kind of disappointed as I noticed the deezer plugin tends to set all mentioned artist as albumartist..