Refactored Kink scraper to use unprint browser. Improved socials handling in actors module.

src/actors.js

@@ -698,7 +698,61 @@ async function scrapeProfiles(actor, sources, entitiesBySlug, existingProfilesBy
 	return profiles.filter(Boolean);
 }
 
+function curateSocials(socials, platformsByHostname) {
+	return socials
+		.map((social) => {
+			if (social.url) {
+				return social.url;
+			}
+
+			if (social.handle && social.platform) {
+				return social;
+			}
+
+			if (typeof social === 'string') {
+				return {
+					url: social,
+				};
+			}
+
+			return null;
+		})
+		.filter(Boolean)
+		.map((social) => {
+			if (social.handle && social.platform && /[\w-]+/.test(social.handle) && /[a-z]+/i.test(social.platform)) {
+				return {
+					platform: social.platform.toLowerCase(),
+					handle: social.handle,
+				};
+			}
+
+			if (social.url) {
+				const { hostname, pathname } = new URL(social.url);
+				const platform = platformsByHostname[hostname];
+
+				if (platform) {
+					const handle = pathname.match(new RegExp(platform.pathname.replace('{handle}', '([\\w-]+)')))?.[1];
+
+					if (handle) {
+						return {
+							platform: platform.platform,
+							handle,
+						};
+					}
+				}
+
+				return {
+					url: social.url,
+				};
+			}
+
+			throw new Error('Invalid social');
+		})
+		.filter(Boolean);
+}
+
 async function associateSocials(profiles) {
+	const { platformsByHostname } = await actorsCommon;
 	const profileEntries = await knex('actors_profiles').whereIn(['actor_id', 'entity_id'], profiles.map((profile) => [profile.actorId, profile.entity.id]));
 
 	const profileEntriesByActorIdAndEntityId = profileEntries.reduce((acc, profileEntry) => {
@@ -725,11 +779,12 @@ async function associateSocials(profiles) {
 		}
 
 		await knex('actors_socials')
-			.insert(profile.social.map((url) => ({
-				url,
-				platform: new URL(url).hostname.match(/([\w-]+)?\.(\w+)$/)?.[1],
+			.insert(curateSocials(profile.social, platformsByHostname).map((social) => ({
+				platform: social.platform,
+				handle: social.handle,
+				url: social.url,
 				actor_id: profile.actorId,
-				profile_id: profileId,
+				// profile_id: profileId,
 			})))
 			.onConflict()
 			.ignore();