feat(lookup): Replace DistroKid placeholder labels with [no label]

Author: kellnerd

harmonizer/release_label.test.ts

@@ -0,0 +1,35 @@
+import { DISTRO_KID_PATTERN } from '@/harmonizer/release_label.ts';
+import { cleanupBogusReleaseLabels } from '@/harmonizer/release_label.ts';
+import type { Label } from '@/harmonizer/types.ts';
+import { noLabel } from '@/musicbrainz/special_entities.ts';
+import { describe, it } from '@std/testing/bdd';
+import { assert } from 'std/assert/assert.ts';
+import { assertEquals } from 'std/assert/assert_equals.ts';
+
+describe('cleanupBogusReleaseLabels', () => {
+	it('replaces DistroKid with [no label]', () => {
+		const label: Label = {
+			name: 'DistroKid',
+			catalogNumber: '12345',
+			externalIds: [],
+		};
+		cleanupBogusReleaseLabels([label]);
+		assertEquals(label, {
+			...noLabel,
+			catalogNumber: '12345',
+		});
+	});
+
+	const distroKidPlaceholders = [
+		'Distro Kid', // Tidal
+		'123456 Records DK',
+		'1234567 Records DK',
+		'1234567 Records DK2',
+	];
+
+	for (const label of distroKidPlaceholders) {
+		it(`detects "${label}" as DistroKid placeholder`, () => {
+			assert(DISTRO_KID_PATTERN.test(label));
+		});
+	}
+});

harmonizer/release_label.ts

@@ -0,0 +1,17 @@
+import { Label } from '@/harmonizer/types.ts';
+import { noLabel } from '@/musicbrainz/special_entities.ts';
+
+/** Placeholder label names that are used by DistroKid. */
+export const DISTRO_KID_PATTERN = /^(Distro ?Kid|\d+ Records DK\d*)$/;
+
+/** Tries to clean up common cases of release labels which are not considered imprints by MusicBrainz. */
+export function cleanupBogusReleaseLabels(labels: Label[]) {
+	for (const label of labels) {
+		if (label.name && DISTRO_KID_PATTERN.test(label.name)) {
+			// DistroKid (https://musicbrainz.org/label/4108147d-f37e-4151-a3e9-d92f0074f1eb) is a distributor
+			label.name = noLabel.name;
+			label.mbid = noLabel.mbid;
+			delete label.externalIds;
+		}
+	}
+}

lookup.ts

@@ -1,5 +1,6 @@
 import { detectLanguageAndScript } from '@/harmonizer/language_script.ts';
 import { mergeRelease } from '@/harmonizer/merge.ts';
+import { cleanupBogusReleaseLabels } from '@/harmonizer/release_label.ts';
 import { defaultProviderPreferences, providers } from '@/providers/mod.ts';
 import { FeatureQuality } from '@/providers/features.ts';
 import { LookupError, ProviderError } from '@/utils/errors.ts';
@@ -300,7 +301,12 @@ export class CombinedReleaseLookup {
 		const release = mergeRelease(releaseMap, providerPreferences);
 		// Prepend error and warning messages of the combined lookup.
 		release.info.messages.unshift(...this.messages);
+
+		// Provider-independent post-processing of the merged release.
 		detectLanguageAndScript(release);
+		if (release.labels) {
+			cleanupBogusReleaseLabels(release.labels);
+		}
 
 		return release;
 	}