A Wrapped / Replay like for teal.fm and rocksky.app (currently on hiatus)
first attempt at musicbrainz lookup
+124
-1
+11
-1
src/mbz/init_fts.sql
+11
-1
src/mbz/init_fts.sql
···
1
1
PRAGMA create_fts_index('mbz.artist_credit', 'id', 'name');
2
2
PRAGMA create_fts_index('mbz.recording', 'id', 'name');
3
-
PRAGMA create_fts_index('mbz.release', 'id', 'name', 'comment');
3
+
PRAGMA create_fts_index('mbz.release', 'id', 'name', 'comment');
4
+
5
+
create macro if not exists mbz.fts_release (term, gid) AS CASE
6
+
WHEN gid IS NOT NULL THEN 10
7
+
ELSE fts_mbz_release.match_bm25 (release.id, term) * if (lower(release.name) = lower(term), 2.5, 1)
8
+
END;
9
+
10
+
create macro if not exists mbz.fts_recording (term, gid) AS CASE
11
+
WHEN gid IS NOT NULL THEN 10
12
+
ELSE fts_mbz_recording.match_bm25 (recording.id, term) * if (lower(recording.name) = lower(term), 2.5, 1)
13
+
END;
+90
src/mbz/mod.rs
+90
src/mbz/mod.rs
···
1
+
use duckdb::{Connection, params};
2
+
1
3
mod replica;
2
4
3
5
pub use replica::start_replication;
6
+
7
+
#[derive(Debug, Default)]
8
+
pub struct FindMbzData<'a> {
9
+
pub release_name: Option<&'a str>,
10
+
pub release_mbid: Option<&'a str>,
11
+
pub release_discrim: Option<&'a str>,
12
+
pub recording_mbid: Option<&'a str>,
13
+
pub artist_name: Option<&'a str>,
14
+
pub track_discrim: Option<&'a str>,
15
+
}
16
+
17
+
#[derive(Debug)]
18
+
pub struct MbzQueryRes {
19
+
pub track: String,
20
+
pub track_gid: String,
21
+
pub release_gid: Option<String>,
22
+
pub release: Option<String>,
23
+
pub recording_gid: Option<String>,
24
+
pub release_group_gid: Option<String>,
25
+
pub release_group: Option<String>,
26
+
pub artists: Option<String>,
27
+
pub debug: String,
28
+
}
29
+
30
+
#[derive(Debug)]
31
+
struct MbzDebug {
32
+
release: f64,
33
+
recording: f64,
34
+
artist: f64,
35
+
}
36
+
37
+
impl std::fmt::Display for MbzDebug {
38
+
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
39
+
write!(
40
+
f,
41
+
"release={:.3};recording={:.3};artist={:.3}",
42
+
self.release, self.recording, self.artist
43
+
)
44
+
}
45
+
}
46
+
47
+
pub fn try_find_mbz_data(
48
+
conn: &Connection,
49
+
track_name: &str,
50
+
opts: &FindMbzData,
51
+
) -> duckdb::Result<Option<MbzQueryRes>> {
52
+
let mut prep = conn.prepare_cached(include_str!("query.sql"))?;
53
+
54
+
let mut rows = prep
55
+
.query_map(
56
+
params![
57
+
opts.release_name,
58
+
opts.release_mbid,
59
+
track_name,
60
+
opts.recording_mbid,
61
+
opts.artist_name,
62
+
],
63
+
|row| {
64
+
let debug = MbzDebug {
65
+
release: row.get(8)?,
66
+
recording: row.get(9)?,
67
+
artist: row.get(10)?,
68
+
};
69
+
70
+
Ok(MbzQueryRes {
71
+
track: row.get(0)?,
72
+
track_gid: row.get(1)?,
73
+
release_gid: row.get(2)?,
74
+
release: row.get(3)?,
75
+
recording_gid: row.get(4)?,
76
+
release_group_gid: row.get(5)?,
77
+
release_group: row.get(6)?,
78
+
artists: row.get(7)?,
79
+
debug: debug.to_string(),
80
+
})
81
+
},
82
+
)?
83
+
.collect::<Result<Vec<_>, _>>()?;
84
+
85
+
if !rows.is_empty() {
86
+
let row = rows.pop();
87
+
return Ok(row);
88
+
}
89
+
90
+
tracing::debug!("couldn't find '{track_name}' // {:?}", opts);
91
+
92
+
Ok(None)
93
+
}
+23
src/mbz/query.sql
+23
src/mbz/query.sql
···
1
+
with releases as (select *, rel_score: mbz.fts_release($1, $2)
2
+
from mbz.release
3
+
where (rel_score is not null and rel_score > 5 and $2 IS NULL)
4
+
or ($2 IS NOT NULL and release.gid = $2)),
5
+
recordings as (select *, rec_score: mbz.fts_recording($3, $4)
6
+
from mbz.recording
7
+
where (rec_score is not null and rec_score > 5 and $4 IS NULL)
8
+
or ($4 IS NOT NULL and recording.gid = $4)),
9
+
artists as (select *, artist_score: flashback.fts_mbz_artist_credit.match_bm25(artist_credit.id, $5),
10
+
from mbz.artist_credit
11
+
where artist_score is not null
12
+
and artist_score > 2.5)
13
+
select track_name: track.name, track_gid: track.gid, release_gid: releases.gid, release_name: releases.name, recording_gid: recordings.gid, release_grp_gid: release_group.gid, release_grp_name: release_group.name, artists: artists.name, rel_score,
14
+
rec_score,
15
+
artist_score
16
+
from mbz.track
17
+
inner join recordings on recordings.id = track.recording
18
+
inner join mbz.medium on medium.id = track.medium
19
+
inner join releases on releases.id = medium.release
20
+
inner join mbz.release_group on release_group.id = releases.release_group
21
+
inner join artists on artists.id = recordings.artist_credit
22
+
where is_data_track = false
23
+
order by rel_score desc, rec_score desc, artist_score desc limit 10;