feat: detect soundtrack based on the release title

Author: gustavkj

harmonizer/release_types.test.ts

@@ -42,6 +42,51 @@ describe('release types', () => {
 			['should detect type before comment', 'Enter Suicidal Angels - EP (Remastered 2021)', new Set(['EP'])],
 			['should detect EP suffix', 'Zero Distance EP', new Set(['EP'])],
 			['should detect demo type', 'Parasite Inc. (Demo)', new Set(['Demo'])],
+			// Soundtrack releases
+			...([
+				// Titles with original/official <medium> soundtrack
+				'The Lord of the Rings: The Return of the King (Original Motion Picture Soundtrack)',
+				'The Bodyguard - Original Soundtrack Album',
+				'Plants Vs. Zombies (Original Video Game Soundtrack)',
+				'Stardew Valley (Original Game Soundtrack)',
+				'L.A. Noire Official Soundtrack',
+				'Tarzan (Deutscher Original Film-Soundtrack)',
+				'Die Eiskönigin Völlig Unverfroren (Deutscher Original Film Soundtrack)',
+				// Soundtrack from the ... <medium>
+				'KPop Demon Hunters (Soundtrack from the Netflix Film)',
+				'The Witcher: Season 2 (Soundtrack from the Netflix Original Series)',
+				'The White Lotus (Soundtrack from the HBO® Original Limited Series)',
+				'Inception (Music from the Motion Picture)',
+				// Releases referring to score instead of soundtrack
+				'Fantastic Mr. Fox - Additional Music From The Original Score By Alexandre Desplat - The Abbey Road Mixes',
+				'Scott Pilgrim Vs. The World (Original Score Composed by Nigel Godrich)',
+				'F1® The Movie (Original Score By Hans Zimmer)',
+				'EUPHORIA SEASON 2 OFFICIAL SCORE (FROM THE HBO ORIGINAL SERIES)',
+				'The Bible (Official Score Soundtrack)',
+				'The Good Wife (The Official TV Score)',
+				// Releases with OST abbreviation
+				'O.S.T. Das Boot',
+				'Alvin & The Chipmunks / OST',
+				// German release titles
+				'Get Up (Der Original Soundtrack zum Kinofilm)',
+				'Ein Mädchen namens Willow - Soundtrack zum Film',
+				'Das Boot (Soundtrack zur TV Serie, zweite Staffel)',
+				// Swedish release titles
+				'Fucking Åmål - Musiken från filmen',
+				'Fejkpatient (Musik från TV-serien)',
+				'Kärlek Fårever (Soundtrack till Netflix-filmen)',
+				// Norwegian release titles
+				'Kvitebjørn (Musikken fra filmen)',
+				'Døden på Oslo S (Musikken fra teaterforestillingen)',
+				// Musical releases
+				'The Lion King: Original Broadway Cast Recording',
+			].map((
+				title,
+			): FunctionSpec<typeof guessTypesFromTitle>[number] => [
+				'should detect soundtrack type',
+				title,
+				new Set(['Soundtrack']),
+			])),
 		];
 
 		describe('exact case match', () => {

harmonizer/release_types.ts

@@ -11,23 +11,60 @@ export function guessTypesForRelease(release: HarmonyRelease): Iterable<ReleaseG
 	return types;
 }
 
-const detectTypesPatterns = [
+const releaseGroupTypeMatchers: Array<{ type?: ReleaseGroupType; pattern: RegExp }> = [
 	// Commonly used for Bandcamp releases
-	/\s\((EP|Single|Live|Demo)\)(?:\s\(.*?\))?$/i,
+	{ pattern: /\s\((EP|Single|Live|Demo)\)(?:\s\(.*?\))?$/i },
 	// iTunes singles and EPs
-	/\s- (EP|Single|Live)(?:\s\(.*?\))?$/i,
+	{ pattern: /\s- (EP|Single|Live)(?:\s\(.*?\))?$/i },
 	// Generic "EP" suffix
-	/\s(EP)(?:\s\(.*?\))?$/i,
+	{ pattern: /\s(EP)(?:\s\(.*?\))?$/i },
+	// Common soundtrack title: "Official/Original <Medium> Soundtrack" and "Original Score"
+	{
+		type: 'Soundtrack',
+		pattern:
+			/(?:Original\s|Official\s)(?:(?:(?:Video\s)?Game|Motion Picture|Film|Movie|Television|TV|(?:(?:TV|Television)[\s-]?)?(?:Mini[\s-]?)?Series?|Musical)[\s-])?(?:Soundtrack|Score)/i,
+	},
+	// Common soundtrack title: "Soundtrack from the <Medium>", should also match "Soundtrack from the <Streaming service> <Medium>"
+	{
+		type: 'Soundtrack',
+		pattern:
+			/(?:Soundtrack|Score|Music)\s(?:(?:from|to) the)\s(?:.+[\s-])?(?:(?:Video\s)?Game|Motion Picture|Film|Movie|(?:(?:TV|Television)[\s-]?)?(?:Mini[\s-]?)?Series|Musical)/i,
+	},
+	// Common soundtrack title. Starting or ending with O.S.T. or OST (with or without wrapping parenthesis). Note: it's case sensitive.
+	{
+		type: 'Soundtrack',
+		pattern: /(?:^(?:\(O\.S\.T\.\)|O\.S\.T\.|OST|\(OST\))\s.+|.+\s(?:\(O\.S\.T\.\)|O\.S\.T\.|OST|\(OST\))$)/,
+	},
+	// Common musical soundtrack release titles
+	{ type: 'Soundtrack', pattern: /Original (?:.+\s)?Cast Recording/i },
+	// Common German soundtrack release titles
+	{
+		type: 'Soundtrack',
+		pattern: /(?:Soundtrack|Musik)\s(?:zum|zur)\s(?:.+[\s-])?(?:(?:Kino)?Film|Theaterstück|(?:TV[\s-]?)?Serie)/i,
+	},
+	// Common Swedish soundtrack release titles
+	{
+		type: 'Soundtrack',
+		pattern:
+			/(?:Soundtrack|Musik(?:en)?)\s(?:från|till|ur)\s(?:.+[\s-])?(?:Film(?:en)?|(?:TV[\s-]?)?(?:Mini[\s-]?)?Serien?|Musikalen)/i,
+	},
+	// Common Norwegian soundtrack release titles
+	{ type: 'Soundtrack', pattern: /Musikk(?:en)? (?:fra) (?:Filmen|TV[\s-]serien|(?:teater)?forestillingen)/i },
 ];
 
 /** Guesses a release type from a title. */
 export function guessTypesFromTitle(title: string): Set<ReleaseGroupType> {
 	const types = new Set<ReleaseGroupType>();
-	detectTypesPatterns.forEach((pattern) => {
-		const match = title.match(pattern);
-		if (match) {
-			types.add(capitalizeReleaseType(match[1]));
+	releaseGroupTypeMatchers.forEach((matcher) => {
+		const match = title.match(matcher.pattern);
+		if (!match) {
+			return;
 		}
+		const type = match[1] || matcher.type;
+		if (!type) {
+			return;
+		}
+		types.add(capitalizeReleaseType(type));
 	});
 	return types;
 }