Discogs plugin going south

[JOJ0]

I am wondering what is going wrong here. Musicbrainz, and I think with the help of chroma plugin as well, is finding out perfectly what this incomplete album is (Sides A+B of this Vinyl rip is missing, I just never did it ;-).

Discogs is my 3rd resort of finding matches, it seems it's finding completely unrelated albums even though I am sure there is several perfectly fine versions of this album on discogs. Weird ha? How would I proceed to debug this, if it's something in the Discogs plugin I might as well look into it myself because I'm interested in the usage of python3-discogs-client here anyway :-)

But first of all I need a debugging starting point:

    ~  beet -v import --set=genre_orig_dir=DowntempoTriphopDubAcidjazz dsmusic/DowntempoTriphopDubAcidjazz/0-ALBUMS/Thievery\ Corporation\ -\ The\ Mirror\ Conspiracy\ C+D                                                          ✔ 
user configuration: /home/jojo/.config/beets/config.yaml
data directory: /home/jojo/.config/beets
plugin paths: 
artresizer: method is (2, (7, 1, 0), False)
thumbnails: using IM to write metadata
thumbnails: using GIO to compute URIs
Sending event: pluginload
library database: /home/jojo/jtbeets.db
library directory: /home/jojo/Music
Sending event: library_opened
Sending event: import_begin
Sending event: import_task_created
Sending event: import_task_start
Looking up: /home/jojo/dsmusic/DowntempoTriphopDubAcidjazz/0-ALBUMS/Thievery Corporation - The Mirror Conspiracy C+D
Tagging  - 
No album ID found.
Search terms:  - 
Album might be VA: True
Requesting MusicBrainz release 01fdaeb5-f4fc-46d6-a43c-4b2f61e95564
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_album_extract
Sending event: albuminfo_received
Requesting MusicBrainz release 1de73c08-5755-3646-ad2c-a6657aa00fd4
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_album_extract
Sending event: albuminfo_received
Requesting MusicBrainz release d8e7f1f0-0d17-47a7-8fcf-0e297adbae97
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_album_extract
Sending event: albuminfo_received
Requesting MusicBrainz release 6f062e31-e362-4b9b-9730-b039f144097e
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_album_extract
Sending event: albuminfo_received
Requesting MusicBrainz release 84aa3182-8a27-4ed8-8773-2a4c626a6340
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_track_extract
Sending event: mb_album_extract
Sending event: albuminfo_received
chroma: acoustid album candidates: 5
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (01fdaeb5-f4fc-46d6-a43c-4b2f61e95564)
Computing track assignment...
...done.
Success. Distance: 0.55
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (1de73c08-5755-3646-ad2c-a6657aa00fd4)
Computing track assignment...
...done.
Success. Distance: 0.51
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (d8e7f1f0-0d17-47a7-8fcf-0e297adbae97)
Computing track assignment...
...done.
Success. Distance: 0.51
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (6f062e31-e362-4b9b-9730-b039f144097e)
Computing track assignment...
...done.
Success. Distance: 0.54
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (84aa3182-8a27-4ed8-8773-2a4c626a6340)
Computing track assignment...
...done.
Success. Distance: 0.57
discogs: Searching for master release 66073
discogs: Searching for master release 1065916
discogs: Searching for master release 17140
discogs: Searching for master release 81002
discogs: Searching for master release 122677
Sending event: albuminfo_received
Candidate: Leftfield - Leftism (3301)
Computing track assignment...
...done.
Success. Distance: 0.87
Sending event: albuminfo_received
Candidate: Yello - Toy (9121346)
Computing track assignment...
...done.
Success. Distance: 0.84
Sending event: albuminfo_received
Candidate: NOFX - 22 Songs That Weren't Good Enough To Go On Our Other Records (826145)
Computing track assignment...
...done.
Success. Distance: 0.88
Sending event: albuminfo_received
Candidate: The J. Geils Band - Bloodshot (412870)
Computing track assignment...
...done.
Success. Distance: 0.82
Sending event: albuminfo_received
Candidate: High Contrast - Basement Track (163369)
Computing track assignment...
...done.
Success. Distance: 0.87
Evaluating 10 candidates.

/home/jojo/dsmusic/DowntempoTriphopDubAcidjazz/0-ALBUMS/Thievery Corporation - The Mirror Conspiracy C+D (7 items)
Sending event: import_task_before_choice
Sending event: before_choose_candidate
Correcting tags from:
    (unknown album)
To:
    Thievery Corporation - The Mirror Conspiracy
URL:
    https://musicbrainz.org/release/1de73c08-5755-3646-ad2c-a6657aa00fd4
(Similarity: 49.0%) (missing tracks, album, artist, tracks) (CD, 2000, GB, 4AD, CAD2K06CD)
 * Thievery Corporation - C1 - So Com Voce (#0)           -> Só com você (#7) (title)
 * Thievery Corporation - C2 - Samba Tranquille (#0)      -> Samba Tranquille (#8) (title)
 * Thievery Corporation - C3 - Shadows Of Ourselves (#0)  -> Shadows of Ourselves (#9) (title)
 * Thievery Corporation - C4 - The Hong Kong Triad (#0)   -> The Hong Kong Triad (#10) (title)
 * Thievery Corporation - D1 - Illumination (#0)          -> Illumination (#11) (title)
 * Thievery Corporation - D2 - The Mirror Conspiracy (#0) -> The Mirror Conspiracy (#12) (title)
 * Thievery Corporation - D3 - Tomorrow (#0)              -> Tomorrow (#13) (title)
Missing tracks (6/13 - 46.2%):
 ! Treasures       (# 1) (2:25)
 ! Le Monde        (# 2) (3:11)
 ! Indra           (# 3) (5:22)
 ! Lebanese Blonde (# 4) (4:48)
 ! Focus on Sight  (# 5) (3:47)
 ! Air Batucada    (# 6) (4:46)
[A]pply, More candidates, Skip, Use as-is, as Tracks, Group albums,
Enter search, enter Id, aBort, eDit, edit Candidates? m
Finding tags for album " - ".
Candidates:
1. Thievery Corporation - The Mirror Conspiracy (49.0%) (missing tracks, album, artist, ...) (CD, 2000, GB, 4AD, CAD2K06CD)
2. Thievery Corporation - The Mirror Conspiracy (49.0%) (missing tracks, album, artist, ...) (CD, 2000, US, Eighteenth Street Lounge Music, ESL 033)
3. Thievery Corporation - The Mirror Conspiracy (45.5%) (missing tracks, album, artist, ...) (2000, GB)
4. Thievery Corporation - The Mirror Conspiracy (45.4%) (missing tracks, album, artist, ...) (2x12" Vinyl, 2000, GB, 4AD, CAD 2K06)
5. Thievery Corporation - The Mirror Conspiracy (42.6%) (missing tracks, album, artist, ...) (Enhanced CD, 2000, US, Eighteenth Street Lounge Music, ESL 033)
6. The J. Geils Band - Bloodshot (18.2%) (tracks, album, artist, ...) (Discogs, Vinyl, 1973, US, Atlantic, SD 7260)
7. Yello - Toy (16.3%) (tracks, missing tracks, album, ...) (Discogs, 2xVinyl, 2016, Europe, Polydor, 00602547602619)
8. High Contrast - Basement Track (13.3%) (tracks, album, artist, ...) (Discogs, Vinyl, 2003, UK, Hospital Records, NHS60)
9. Leftfield - Leftism (12.7%) (tracks, missing tracks, album, ...) (Discogs, CD, UK & Europe, Hard Hands, HANDCD2)
10. NOFX - 22 Songs That Weren't Good Enough To Go On Our Other Records (12.4%) (missing tracks, tracks, album, ...) (Discogs, Vinyl, 2002, US, Fat Wreck Chords, FAT641-1)
# selection (default 1), Skip, Use as-is, as Tracks, Group albums,
Enter search, enter Id, aBort, eDit, edit Candidates? 

[JOJ0]

This is the interesting part. What part of the process is passing these discogs master releases ids to the discogs plugin? Where are they from? that's the source of evil:

Sending event: mb_album_extract
Sending event: albuminfo_received
chroma: acoustid album candidates: 5
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (01fdaeb5-f4fc-46d6-a43c-4b2f61e95564)
Computing track assignment...
...done.
Success. Distance: 0.55
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (1de73c08-5755-3646-ad2c-a6657aa00fd4)
Computing track assignment...
...done.
Success. Distance: 0.51
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (d8e7f1f0-0d17-47a7-8fcf-0e297adbae97)
Computing track assignment...
...done.
Success. Distance: 0.51
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (6f062e31-e362-4b9b-9730-b039f144097e)
Computing track assignment...
...done.
Success. Distance: 0.54
Sending event: albuminfo_received
Candidate: Thievery Corporation - The Mirror Conspiracy (84aa3182-8a27-4ed8-8773-2a4c626a6340)
Computing track assignment...
...done.
Success. Distance: 0.57
discogs: Searching for master release 66073
discogs: Searching for master release 1065916
discogs: Searching for master release 17140
discogs: Searching for master release 81002
discogs: Searching for master release 122677
Sending event: albuminfo_received
Candidate: Leftfield - Leftism (3301)
Computing track assignment...
...done.
Success. Distance: 0.87
Sending event: albuminfo_received
Candidate: Yello - Toy (9121346)
Computing track assignment...
...done.
Success. Distance: 0.84
Sending event: albuminfo_received
Candidate: NOFX - 22 Songs That Weren't Good Enough To Go On Our Other Records (826145)
Computing track assignment...
...done.
Success. Distance: 0.88
Sending event: albuminfo_received
Candidate: The J. Geils Band - Bloodshot (412870)
Computing track assignment...
...done.
Success. Distance: 0.82
Sending event: albuminfo_received
Candidate: High Contrast - Basement Track (163369)
Computing track assignment...
...done.
Success. Distance: 0.87
Evaluating 10 candidates

[JOJ0]

This might be the answer: The original mp3 files don't have any tags at all (a very very lazy vinyl rip I did back then ;-)))
Also this output proofs that there is not much to search for:

Looking up: /home/jojo/dsmusic/DowntempoTriphopDubAcidjazz/0-ALBUMS/Thievery Corporation - The Mirror Conspiracy C+D
Tagging  - 
No album ID found.
Search terms:  - 
Album might be VA: True

Anyway, it seems musicbrainz (or is it chroma??) can handle this, discogs plugin can't. Does Musicbrainz built-in plugin somehow use the information provided via folder and filenames too? And Discgos obviously does not. This could then be improved in the Discogs plugin I guess.

Or am I reading the output wrong and the only reason why a match was found on musicbrainz is because chroma plugin figured it out via acoustid fingerprint.

Please help me learn to read the log! :-))

[JOJ0]

We don't currently have an option like that; sorry.

I would love to make this configurable. IMHO it does not make sense for a plugin to search with nothing. It could be made configurable whether the plugin should even try to search if all or certain tags are empty and state this clearly in the log. What do you think of such an addition to the Discogs plugin? If you point me to a starting point, that would be helpful.

[sampsyo]

It's not entirely clear to me exactly what the discogs plugin does when metadata is missing, but digging around in discogs.py would be the place to start to find out.

[sampsyo]

What is the best practice order for using the fromfilename plugin? Should I put it as the very first plugin to use? Since it only reacts on empty tags, that should work right?

FWIW, there is no priority ordering on plugins. Changing the order on your plugins: line will have no effect. The fromfilename plugin triggers when music is completely devoid of metadata and should affect every metadata source.

[JOJ0]

FWIW, there is no priority ordering on plugins. Changing the order on your plugins: line will have no effect. The fromfilename plugin triggers when music is completely devoid of metadata and should affect every metadata sour

I was aware / wouldn't have guessed that the order of the plugins: values matters haha :-)

But what I am realissing now is that I did have a wrong assumption: I assumed plugins run in the order that source_weight is specifiying - highest number = runs first. That was an entirely wrong assumption, that correct?

So then it's rather like: Each plugin does it's own thing in a specific order but is not dependent on each others outputs/results? So implementing things like: Wait for plugin a and b for their results, and then take that in plugin c and do something with it is not easily possible right?

[sampsyo]

Ah, great! Yep, that's correct—all the sources get queried in parallel, and then the weights apply just to prioritize the resulting matches (i.e., compute their "distance").

[JOJ0]

Fixed in PR #4227