Changed qu's HTML element detection. Passing base actor instead of actorName to profile scrapers.

seeds/04_media.js

@@ -719,7 +719,7 @@ const tagPhotos = [
 	['latina', 2, 'Alexis Love for Penthouse'],
 	['mff', 0, 'Madison Ivy, Adriana Chechik and Keiran Lee in "Day With A Pornstar" for Brazzers'],
 	['mfm', 6, 'Honey Gold in "Slut Puppies 12" for Jules Jordan'],
-	['natural-boobs', 0, 'Autumn Falls in "Manuel Ferrara\'s Ripe 7" for Jules Jordan'],
+	['natural-boobs', 0, 'Valentina Nappi in "Hypnotic Curves" for LesbianX'],
 	['oil', 1, 'Kissa Sins in "Oil Overload 14" for JulesJordan'],
 	['oil', 3, 'Vina Sky for Lubed'],
 	['oil', 0, 'Jada Stevens in "Jada Stevens Anal Ass Oiled Up For James Deen\'s Cock" for Jules Jordan'],

src/actors.js

@@ -114,12 +114,15 @@ function getAverage(items) {
 
 function toBaseActors(actorsOrNames, release) {
 	return actorsOrNames.map((actorOrName) => {
-		const name = capitalize(actorOrName.name || actorOrName);
+		const [baseName, entryId] = (actorOrName.name || actorOrName).split(':');
+
+		const name = capitalize(baseName);
 		const slug = slugify(name);
 
 		const baseActor = {
 			name,
 			slug,
+			entryId: entryId || null,
 			entity: release?.site?.network || release?.entity?.parent || release?.entity || null,
 		};
 
@@ -213,6 +216,7 @@ function curateActorEntry(baseActor, batchId) {
 		name: baseActor.name,
 		slug: baseActor.slug,
 		entity_id: null,
+		entry_id: baseActor.entry_id,
 		batch_id: batchId,
 	};
 }
@@ -538,7 +542,7 @@ async function scrapeProfiles(actor, sources, entitiesBySlug, existingProfilesBy
 
 					logger.verbose(`Searching profile for '${actor.name}' on '${label}'`);
 
-					const profile = await scraper.fetchProfile(actor.name, context, include);
+					const profile = await scraper.fetchProfile(actor, context, include);
 
 					if (!profile || typeof profile === 'number') { // scraper returns HTTP code on request failure
 						logger.verbose(`Profile for '${actor.name}' not available on ${label}, scraper returned ${profile}`);
@@ -587,7 +591,7 @@ async function scrapeActors(actorNames) {
 			.leftJoin('entities as parents', 'parents.id', 'entities.parent_id')
 			.orderBy('entities.type'),
 		knex('actors')
-			.select(['id', 'name', 'slug'])
+			.select(['id', 'name', 'slug', 'entry_id'])
 			.modify((queryBuilder) => {
 				if (actorNames.length > 0) {
 					queryBuilder.whereIn('slug', baseActors.map(baseActor => baseActor.slug));
@@ -598,12 +602,22 @@ async function scrapeActors(actorNames) {
 
 	const entitiesBySlug = entities.reduce((acc, entity) => ({ ...acc, [entity.slug]: entity }), {});
 
-	const existingActorEntriesBySlug = existingActorEntries.reduce((acc, actorEntry) => ({ ...acc, [actorEntry.slug]: actorEntry }), {});
-	const newBaseActors = baseActors.filter(baseActor => !existingActorEntriesBySlug[baseActor.slug]);
+	const existingActorEntriesBySlugAndEntryId = existingActorEntries.reduce((acc, actorEntry) => ({
+		...acc,
+		[actorEntry.slug]: {
+			...acc[actorEntry.slug],
+			[actorEntry.entryId || null]: actorEntry,
+		},
+	}), {});
+
+	const newBaseActors = baseActors.filter(baseActor => !existingActorEntriesBySlugAndEntryId[baseActor.slug][baseActor.entryId]);
 
 	const [batchId] = newBaseActors.length > 0 ? await knex('batches').insert({ comment: null }).returning('id') : [null];
 	const curatedActorEntries = batchId && curateActorEntries(newBaseActors, batchId);
-	const newActorEntries = batchId && await knex('actors').insert(curatedActorEntries).returning(['id', 'name', 'slug']);
+
+	const newActorEntries = batchId && await knex('actors')
+		.insert(curatedActorEntries)
+		.returning(['id', 'name', 'slug', 'entry_id']);
 
 	const actors = existingActorEntries.concat(Array.isArray(newActorEntries) ? newActorEntries : []);
 

src/scrapers/brazzers.js

@@ -69,10 +69,12 @@ async function scrapeScene({ query, html }, url, _site) {
 		],
 	}), {});
 
-	release.actors = query.all('.related-model a').map(actorEl => ({
-		name: query.q(actorEl, null, 'title'),
-		avatar: actorImagesByActorId[query.url(actorEl, null).match(/\/view\/id\/(\d+)/)?.[1]],
-	}));
+	release.actors = query.all('.related-model a').map((actorEl) => {
+		const name = query.q(actorEl, null, 'title');
+		const avatar = actorImagesByActorId[query.url(actorEl, null).match(/\/view\/id\/(\d+)/)?.[1]];
+
+		return { name, avatar };
+	});
 
 	release.likes = query.number('.label-rating .like');
 	release.dislikes = query.number('.label-rating .dislike');

src/scrapers/fcuk.js

@@ -1,8 +1,8 @@
 'use strict';
 
 const qu = require('../utils/qu');
-
-// TODO: profile scraping
+const slugify = require('../utils/slugify');
+const { feetInchesToCm } = require('../utils/convert');
 
 function scrapeLatestBlog(scenes, channel) {
 	return scenes.map(({ query }) => {
@@ -106,6 +106,33 @@ function scrapeScene({ query, html }, url, channel) {
 	return release;
 }
 
+function scrapeProfile({ query }, entity) {
+	const profile = {};
+
+	const bio = query.cnts('.info p').reduce((acc, info) => {
+		const [key, value] = info.match(/(\w+):\s*(.*)/).slice(1);
+
+		return { ...acc, [slugify(key, '_')]: value };
+	}, {});
+
+	profile.age = Number(bio.age);
+	profile.height = feetInchesToCm(bio.height);
+	profile.eyes = bio.eyes || bio.eyecolor;
+
+	if (bio.figure || bio.measurements) {
+		const [bust, cup, waist, hip] = (bio.figure || bio.measurements)?.match(/(\d+)(\w+)-(\d+)-(\d+)/).slice(1);
+
+		profile.bust = Number(bust);
+		profile.cup = cup;
+		profile.waist = Number(waist);
+		profile.hip = Number(hip);
+	}
+
+	profile.avatar = query.img('img.main-preview', 'src', { origin: entity.url });
+
+	return profile;
+}
+
 async function fetchLatestBlog(channel, page) {
 	const url = `${channel.url}/free/updates/videos/${(page - 1) * 10}`;
 	const res = await qu.getAll(url, '.videos');
@@ -148,8 +175,32 @@ async function fetchScene(url, channel) {
 	return res.status;
 }
 
+async function fetchProfile(baseActor, entity) {
+	const modelsRes = await qu.getAll(`${entity.url}/free/girls.php?alpha=${baseActor.name.slice(0, 1)}`, '.model');
+
+	console.log(baseActor);
+
+	if (modelsRes.ok) {
+		const models = modelsRes.items.filter(({ query }) => query.cnt('strong') === baseActor.name);
+
+		return Promise.all(models.map(async (model) => {
+			const modelUrl = model.query.url('a', 'href', { origin: entity.url });
+			const modelRes = await qu.get(modelUrl);
+
+			if (modelRes.ok) {
+				return scrapeProfile(modelRes.item, entity);
+			}
+
+			return modelRes.status;
+		}));
+	}
+
+	return modelsRes.status;
+}
+
 module.exports = {
 	fetchLatest,
 	fetchScene,
 	fetchUpcoming,
+	fetchProfile,
 };

src/scrapers/scrapers.js

@@ -176,6 +176,7 @@ module.exports = {
 		evilangel,
 		eyeontheguy: hush,
 		fakehub,
+		exploitedcollegegirls: fcuk,
 		forbondage: porndoe,
 		freeones,
 		gangbangcreampie: aziani,

src/utils/qu.js

@@ -326,9 +326,9 @@ function init(element, window) {
 	const quContextFuncs = Object.entries(quFuncs) // dynamically attach methods with context
 		.reduce((acc, [key, func]) => ({
 			...acc,
-			[key]: (...args) => (args[0] instanceof globalWindow.HTMLElement // allow for different context
-				? func(...args)
-				: func(element, ...args)),
+			[key]: (...args) => (args[0].nodeType === undefined // allow for different context
+				? func(element, ...args)
+				: func(...args)),
 		}), {});
 
 	return {