From 85c73bad775bef7567bf44f1f145001f3fb6a439 Mon Sep 17 00:00:00 2001
From: DebaucheryLibrarian <moonloop.adult@protonmail.com>
Date: Mon, 30 May 2022 00:05:33 +0200
Subject: [PATCH] Improved MindGeek actor scraper.

---
 assets/components/actors/tile.vue |  2 +-
 src/scrapers/mindgeek.js          | 69 +++++++++++++------------------
 2 files changed, 30 insertions(+), 41 deletions(-)

diff --git a/assets/components/actors/tile.vue b/assets/components/actors/tile.vue
index 7e1b70a7..7caa7eff 100644
--- a/assets/components/actors/tile.vue
+++ b/assets/components/actors/tile.vue
@@ -22,7 +22,7 @@
 					class="favicon"
 				>
 					<img
-						:src="`/img/logos/${actor.entity.slug}/favicon_dark.png`"
+						:src="`/img/logos/${actor.entity.slug}/favicon_light.png`"
 						class="favicon-icon"
 					>
 				</RouterLink>
diff --git a/src/scrapers/mindgeek.js b/src/scrapers/mindgeek.js
index 6ca0a319..72e61a20 100644
--- a/src/scrapers/mindgeek.js
+++ b/src/scrapers/mindgeek.js
@@ -242,29 +242,22 @@ async function getSession(site, parameters, url) {
 	throw new Error(`Failed to acquire MindGeek session (${res.statusCode})`);
 }
 
-function scrapeProfile(data, html, releases = [], networkName) {
-	const { query } = qu.extract(html);
-
+function scrapeProfile(data, releases = [], networkName) {
 	const profile = {
 		description: data.bio,
-		aliases: data.aliases,
+		aliases: data.aliases.filter(Boolean),
 	};
 
 	profile.gender = data.gender === 'other' ? 'transsexual' : data.gender;
+	profile.measurements = data.measurements;
 
-	if (data.measurements) {
-		const [bust, waist, hip] = data.measurements.split('-');
+	profile.dateOfBirth = qu.parseDate(data.birthday);
+	profile.birthPlace = data.birthPlace;
+	profile.height = inchesToCm(data.height);
+	profile.weight = lbsToKg(data.weight);
 
-		if (profile.gender === 'female') {
-			if (bust) profile.bust = bust.toUpperCase();
-			if (waist) profile.waist = waist;
-			if (hip) profile.hip = hip;
-		}
-	}
-
-	if (data.birthPlace) profile.birthPlace = data.birthPlace;
-	if (data.height) profile.height = inchesToCm(data.height);
-	if (data.weight) profile.weight = lbsToKg(data.weight);
+	profile.hairColor = data.tags.find((tag) => /hair color/i.test(tag.category))?.name;
+	profile.ethnicity = data.tags.find((tag) => /ethnicity/i.test(tag.category))?.name;
 
 	if (data.images.card_main_rect?.[0]) {
 		profile.avatar = data.images.card_main_rect[0].xl?.url
@@ -274,9 +267,6 @@ function scrapeProfile(data, html, releases = [], networkName) {
             || data.images.card_main_rect[0].xs?.url;
 	}
 
-	const birthdate = query.all('li').find((el) => /Date of Birth/.test(el.textContent));
-	if (birthdate) profile.birthdate = query.date(birthdate, 'span', 'MMMM Do, YYYY');
-
 	if (data.tags.some((tag) => /boob type/i.test(tag.category) && /natural tits/i.test(tag.name))) {
 		profile.naturalBoobs = true;
 	}
@@ -285,6 +275,14 @@ function scrapeProfile(data, html, releases = [], networkName) {
 		profile.naturalBoobs = false;
 	}
 
+	if (data.tags.some((tag) => /body art/i.test(tag.category) && /tattoo/i.test(tag.name))) {
+		profile.hasTattoos = true;
+	}
+
+	if (data.tags.some((tag) => /body art/i.test(tag.category) && /piercing/i.test(tag.name))) {
+		profile.hasPiercings = true;
+	}
+
 	profile.releases = releases.map((release) => scrapeRelease(release, null, null, networkName));
 
 	return profile;
@@ -377,7 +375,7 @@ async function fetchRelease(url, site, baseScene, options) {
 	return null;
 }
 
-async function fetchProfile({ name: actorName, slug: actorSlug }, { entity, parameters }) {
+async function fetchProfile({ name: actorName }, { entity, parameters }, include) {
 	// const url = `https://www.${networkOrNetworkSlug.slug || networkOrNetworkSlug}.com`;
 	const { session, instanceToken } = await getSession(entity, parameters);
 
@@ -395,31 +393,22 @@ async function fetchProfile({ name: actorName, slug: actorSlug }, { entity, para
 		const actorData = res.body.result.find((actor) => actor.name.toLowerCase() === actorName.toLowerCase());
 
 		if (actorData) {
-			const actorUrl = `https://www.${entity.slug}.com/${entity.parameters?.actorPath || 'model'}/${actorData.id}/${actorSlug}`;
 			const actorReleasesUrl = `https://site-api.project1service.com/v2/releases?actorId=${actorData.id}&limit=100&offset=0&orderBy=-dateReleased&type=scene`;
 
-			const [actorRes, actorReleasesRes] = await Promise.all([
-				http.get(actorUrl, {
-					interval: parameters.interval,
-					concurrency: parameters.concurrency,
-				}),
-				http.get(actorReleasesUrl, {
-					session,
-					interval: parameters.interval,
-					concurrency: parameters.concurrency,
-					headers: {
-						Instance: instanceToken,
-					},
-				}),
-			]);
+			const actorReleasesRes = include.includeActorScenes && await http.get(actorReleasesUrl, {
+				session,
+				interval: parameters.interval,
+				concurrency: parameters.concurrency,
+				headers: {
+					Instance: instanceToken,
+				},
+			});
 
-			if (actorRes.statusCode === 200 && actorReleasesRes.statusCode === 200 && actorReleasesRes.body.result) {
-				return scrapeProfile(actorData, actorRes.body.toString(), actorReleasesRes.body.result, entity.slug);
+			if (actorReleasesRes.statusCode === 200 && actorReleasesRes.body.result) {
+				return scrapeProfile(actorData, actorReleasesRes.body.result, entity.slug);
 			}
 
-			if (actorRes.statusCode === 200) {
-				return scrapeProfile(actorData, actorRes.body.toString(), null, entity.slug);
-			}
+			return scrapeProfile(actorData, [], entity.slug);
 		}
 	}