filter url properties by mediaType

packages/backend/src/core/activitypub/ApUtilityService.ts

@@ -45,25 +45,36 @@ export class ApUtilityService {
 	}
 
 	/**
-	 * Searches a list of URLs or Links for the first one matching a given target URL's host authority.
+	 * Finds the "best" URL for a given AP object.
-	 * Returns null if none match.
+	 * The list of URLs is first filtered via findSameAuthorityUrl, then further filtered based on mediaType, and finally sorted to select the best one.
-	 * @param targetUrl URL with the target host authority
+	 * @throws {IdentifiableError} if object does not have an ID
-	 * @param searchUrls URL, Link, or array to search for matching URLs
+	 * @returns the best URL, or null if none were found
 	 */
-	public findSameAuthorityUrl(targetUrl: string, searchUrls: string | IObject | undefined | (string | IObject)[]): string | null {
+	public findBestObjectUrl(object: IObject): string | null {
+		const targetUrl = getApId(object);
 		const targetAuthority = this.utilityService.punyHostPSLDomain(targetUrl);
 
-		const match = toArray(searchUrls)
+		const rawUrls = toArray(object.url);
-			.map(raw => getOneApHrefNullable(raw))
+		const acceptableUrls = rawUrls
-			.find(url => {
+			.map(raw => ({
+				url: getOneApHrefNullable(raw),
+				type: typeof(raw) === 'object'
+					? raw.mediaType?.toLowerCase()
+					: undefined,
+			}))
+			.filter(({ url, type }) => {
 				if (!url) return false;
 				if (!this.checkHttps(url)) return false;
+				if (!isAcceptableUrlType(type)) return false;
 
 				const urlAuthority = this.utilityService.punyHostPSLDomain(url);
 				return urlAuthority === targetAuthority;
+			})
+			.sort((a, b) => {
+				return rankUrlType(a.type) - rankUrlType(b.type);
 			});
 
-		return match ?? null;
+		return acceptableUrls[0]?.url ?? null;
 	}
 
 	/**
@@ -78,3 +89,20 @@ export class ApUtilityService {
 		return url.startsWith('https://') || (url.startsWith('http://') && isNonProd);
 	}
 }
+
+function isAcceptableUrlType(type: string | undefined): boolean {
+	if (!type) return true;
+	if (type.startsWith('text/')) return true;
+	if (type.startsWith('application/ld+json')) return true;
+	if (type.startsWith('application/activity+json')) return true;
+	return false;
+}
+
+function rankUrlType(type: string | undefined): number {
+	if (!type) return 2;
+	if (type === 'text/html') return 0;
+	if (type.startsWith('text/')) return 1;
+	if (type.startsWith('application/ld+json')) return 3;
+	if (type.startsWith('application/activity+json')) return 4;
+	return 5;
+}

packages/backend/src/core/activitypub/models/ApNoteService.ts

@@ -180,7 +180,7 @@ export class ApNoteService {
 			throw new UnrecoverableError(`unexpected schema of note.id ${note.id} in ${entryUri}`);
 		}
 
-		const url = this.apUtilityService.findSameAuthorityUrl(note.id, note.url);
+		const url = this.apUtilityService.findBestObjectUrl(note);
 
 		this.logger.info(`Creating the Note: ${note.id}`);
 
@@ -395,7 +395,7 @@ export class ApNoteService {
 			throw new UnrecoverableError(`unexpected schema of note.id ${note.id} in ${noteUri}`);
 		}
 
-		const url = this.apUtilityService.findSameAuthorityUrl(note.id, note.url);
+		const url = this.apUtilityService.findBestObjectUrl(note);
 
 		this.logger.info(`Creating the Note: ${note.id}`);
 

packages/backend/src/core/activitypub/models/ApPersonService.ts

@@ -351,8 +351,7 @@ export class ApPersonService implements OnModuleInit {
 			throw new UnrecoverableError(`Refusing to create person without id: ${uri}`);
 		}
 
-		const personId = getApId(person);
+		const url = this.apUtilityService.findBestObjectUrl(person);
-		const url = this.apUtilityService.findSameAuthorityUrl(personId, person.url);
 
 		// Create user
 		let user: MiRemoteUser | null = null;
@@ -561,8 +560,8 @@ export class ApPersonService implements OnModuleInit {
 		if (person.id == null) {
 			throw new UnrecoverableError(`Refusing to update person without id: ${uri}`);
 		}
-		const personId = getApId(person);
+
-		const url = this.apUtilityService.findSameAuthorityUrl(personId, person.url);
+		const url = this.apUtilityService.findBestObjectUrl(person);
 
 		const updates = {
 			lastFetchedAt: new Date(),

packages/backend/test/unit/core/activitypub/ApUtilityService.ts

@@ -95,83 +95,257 @@ describe(ApUtilityService, () => {
 		});
 	});
 
-	describe('findSameAuthorityUrl', () => {
+	describe('findBestObjectUrl', () => {
 		it('should return null when input is undefined', () => {
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com', undefined);
+			const object = {
+				id: 'https://example.com',
+				url: undefined,
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBeNull();
 		});
 
 		it('should return null when input is empty array', () => {
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com', []);
+			const object = {
+				id: 'https://example.com',
+				url: [] as string[],
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBeNull();
 		});
 
 		it('should return return url if string input matches', () => {
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', 'https://example.com/2');
+			const object = {
+				id: 'https://example.com/1',
+				url: 'https://example.com/2',
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
 		it('should return return url if object input matches', () => {
-			const input = {
+			const object = {
-				href: 'https://example.com/2',
+				id: 'https://example.com/1',
+				url: {
+					href: 'https://example.com/2',
+				} as IObject,
 			} as IObject;
 
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', input);
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
 		it('should return return url if string[] input matches', () => {
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', ['https://example.com/2']);
+			const object = {
+				id: 'https://example.com/1',
+				url: ['https://example.com/2'],
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
 		it('should return return url if object[] input matches', () => {
-			const input = {
+			const object = {
-				href: 'https://example.com/2',
+				id: 'https://example.com/1',
+				url: [{
+					href: 'https://example.com/2',
+				} as IObject],
 			} as IObject;
 
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', [input]);
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
 		it('should skip invalid entries', () => {
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', [{} as IObject, 'https://example.com/2']);
+			const object = {
+				id: 'https://example.com/1',
+				url: [{} as IObject, 'https://example.com/2'],
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
-		it('should return first match', () => {
+		it('should allow empty mediaType', () => {
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', ['https://example.com/2', 'https://example.com/3']);
+			const object = {
+				id: 'https://example.com/1',
+				url: {
+					href: 'https://example.com/2',
+				} as IObject,
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
+
+			expect(result).toBe('https://example.com/2');
+		});
+
+		it('should allow text/html mediaType', () => {
+			const object = {
+				id: 'https://example.com/1',
+				url: {
+					href: 'https://example.com/2',
+					mediaType: 'text/html',
+				} as IObject,
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
+
+			expect(result).toBe('https://example.com/2');
+		});
+
+		it('should allow other text/ mediaTypes', () => {
+			const object = {
+				id: 'https://example.com/1',
+				url: {
+					href: 'https://example.com/2',
+					mediaType: 'text/imaginary',
+				} as IObject,
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
+
+			expect(result).toBe('https://example.com/2');
+		});
+
+		it('should allow application/ld+json mediaType', () => {
+			const object = {
+				id: 'https://example.com/1',
+				url: {
+					href: 'https://example.com/2',
+					mediaType: 'application/ld+json;profile=https://www.w3.org/ns/activitystreams',
+				} as IObject,
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
+
+			expect(result).toBe('https://example.com/2');
+		});
+
+		it('should allow application/activity+json mediaType', () => {
+			const object = {
+				id: 'https://example.com/1',
+				url: {
+					href: 'https://example.com/2',
+					mediaType: 'application/activity+json',
+				} as IObject,
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
+
+			expect(result).toBe('https://example.com/2');
+		});
+
+		it('should reject other mediaTypes', () => {
+			const object = {
+				id: 'https://example.com/1',
+				url: [
+					{
+						href: 'https://example.com/2',
+						mediaType: 'application/json',
+					} as IObject,
+					{
+						href: 'https://example.com/3',
+						mediaType: 'image/jpeg',
+					} as IObject,
+				],
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
+
+			expect(result).toBeNull();
+		});
+
+		it('should return best match', () => {
+			const object = {
+				id: 'https://example.com/1',
+				url: [
+					'https://example.com/2',
+					{
+						href: 'https://example.com/3',
+					} as IObject,
+					{
+						href: 'https://example.com/4',
+						mediaType: 'text/html',
+					} as IObject,
+					{
+						href: 'https://example.com/5',
+						mediaType: 'text/plain',
+					} as IObject,
+					{
+						href: 'https://example.com/6',
+						mediaType: 'application/ld+json',
+					} as IObject,
+					{
+						href: 'https://example.com/7',
+						mediaType: 'application/activity+json',
+					} as IObject,
+					{
+						href: 'https://example.com/8',
+						mediaType: 'image/jpeg',
+					} as IObject,
+				],
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
+
+			expect(result).toBe('https://example.com/4');
+		});
+
+		it('should return first match in case of ties', () => {
+			const object = {
+				id: 'https://example.com/1',
+				url: ['https://example.com/2', 'https://example.com/3'],
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
 		it('should skip invalid scheme', () => {
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', ['file://example.com/1', 'https://example.com/2']);
+			const object = {
+				id: 'https://example.com/1',
+				url: ['file://example.com/1', 'https://example.com/2'],
+			} as IObject;
+
+			const result = serviceUnderTest.findBestObjectUrl(object);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
 		it('should skip HTTP in production', () => {
+			// noinspection HttpUrlsUsage
+			const object = {
+				id: 'https://example.com/1',
+				url: ['http://example.com/1', 'https://example.com/2'],
+			} as IObject;
 			env.NODE_ENV = 'production';
 
-			// noinspection HttpUrlsUsage
+			const result = serviceUnderTest.findBestObjectUrl(object);
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', ['http://example.com/1', 'https://example.com/2']);
 
 			expect(result).toBe('https://example.com/2');
 		});
 
 		it('should allow HTTP in non-prod', () => {
+			// noinspection HttpUrlsUsage
+			const object = {
+				id: 'https://example.com/1',
+				url: ['http://example.com/1', 'https://example.com/2'],
+			} as IObject;
 			env.NODE_ENV = 'test';
 
-			// noinspection HttpUrlsUsage
+			const result = serviceUnderTest.findBestObjectUrl(object);
-			const result = serviceUnderTest.findSameAuthorityUrl('https://example.com/1', ['http://example.com/1', 'https://example.com/2']);
 
 			// noinspection HttpUrlsUsage
 			expect(result).toBe('http://example.com/1');