traxxx/src/scrapers/teamskeet.js

'use strict';

const qu = require('../utils/qu');
const http = require('../utils/http');
const slugify = require('../utils/slugify');
const { lbsToKg, feetInchesToCm } = require('../utils/convert');

function getChannelSlug(channelName, entity) {
	if (entity.type === 'channel') {
		return entity.slug;
	}

	if (!channelName) {
		return null;
	}

	const channelSlug = slugify(channelName, '', { removePunctuation: true });
	const channel = entity.children.find((child) => new RegExp(channelSlug).test(child.slug));

	return channel?.slug || null;
}

function scrapeScene(scene, channel, parameters) {
	const release = {};

	release.entryId = scene.id;
	release.url = `${channel.type === 'network' || channel.parameters?.layout === 'organic' ? channel.url : channel.parent.url}/movies/${release.entryId}`;

	release.title = scene.title;
	release.description = scene.description;
	release.date = qu.extractDate(scene.publishedDate);

	release.actors = scene.models?.map((model) => model.modelName) || [];
	release.actors = scene.models?.map((model) => ({
		name: model.modelName,
		avatar: parameters.avatars && `${parameters.avatars}/${slugify(model.modelName, '_')}.jpg`,
		url: `${channel.url}/models/${model.modelId}`,
	}));

	release.poster = [
		// scene.img.replace('med.jpg', 'hi.jpg'), // this image is not always from the same scene! for example on Petite Teens 18
		scene.img,
	];

	release.teaser = scene.videoTrailer;

	if (scene.video) {
		release.trailer = `https://cloudflarestream.com/${scene.video}/manifest/video.mpd?parentOrigin=${encodeURIComponent(channel.url)}`;
	}

	release.tags = scene.tags;

	release.likes = scene.stats?.likeCount;
	release.dislikes = scene.stats?.dislikeCount;

	release.channel = getChannelSlug(scene.site?.name || scene.site?.nickName, channel);

	return release;
}

function scrapeAll(scenes, channel, parameters) {
	return scenes.map((scene) => scrapeScene(scene, channel, parameters));
}

function scrapeProfile(actor, entity, parameters) {
	const profile = {};

	profile.url = `${entity.url}/models/${actor.id}`;

	profile.description = actor.modelBio;

	if (actor.bio.about && !/\band\b/.test(actor.bio.about)) {
		const bio = actor.bio.about.split(/\n/).filter(Boolean).reduce((acc, item) => {
			const [key, value] = item.match(/(.+): (.+)/).slice(1);

			return { ...acc, [slugify(key, '_')]: value.trim() };
		}, {});

		// birthdate seems never/rarely correct

		if (bio.measurements) {
			profile.measurements = bio.measurements;
		} else {
			const breastSize = actor.bio.breastSize?.match(/(\d+)(\w+)/)?.slice(1) || actor.bio.about.match(/Measurements: (\d+)(\w+)/)?.slice(1);

			if (breastSize) {
				[profile.bust, profile.cup] = breastSize;
			}
		}

		profile.birthPlace = bio.birth_location;
		profile.nationality = bio.nationality;
		profile.ethnicity = bio.ethnicity;
		profile.hairColor = bio.hair_color;

		const piercings = actor.bio.about.match(/Piercings: (\w+)/i)?.[1];
		const tattoos = actor.bio.about.match(/Tattoos: (\w+)/i)?.[1];

		if (/yes|various/i.test(piercings)) profile.hasPiercings = true;
		else if (/no/i.test(piercings)) profile.hasPiercings = false;
		else if (bio.piercings) {
			profile.hasPiercings = true;
			profile.piercings = piercings;
		}

		if (/yes|various/i.test(tattoos)) profile.hasTattoos = true;
		else if (/no/i.test(tattoos)) profile.hasTattoos = false;
		else if (bio.tattoos) {
			profile.hasTattoos = true;
			profile.tattoos = tattoos;
		}
	}

	if (actor.bio.heightFeet && actor.bio.heightInches) {
		profile.height = feetInchesToCm(actor.bio.heightFeet, actor.bio.heightInches);
	}

	if (actor.bio.weight) {
		profile.weight = lbsToKg(actor.bio.weight);
	}

	profile.avatar = actor.img;
	profile.banner = actor.cover;
	profile.scenes = actor.movies?.map((scene) => scrapeScene(scene, entity, parameters));

	return profile;
}

async function fetchLatest(channel, page = 1, { parameters }) {
	const res = await http.get(`https://store2.psmcdn.net/${parameters.endpoint}-videoscontent/_search?q=site.seo.seoSlug:"${parameters.id}"&sort=publishedDate:desc&size=30&from=${(page - 1) * 30}`);

	if (res.ok) {
		return scrapeAll(res.body.hits.hits.map(({ _source: scene }) => scene), channel, parameters);
	}

	return res.status;
}

async function fetchLatestOrganic(channel, page, context) {
	const res = await http.get(`https://store.psmcdn.net/${context.parameters.endpoint}/newestMovies/items.json?orderBy="$key"&startAt="${context.cursor || 'aaaaaaaa'}"&limitToFirst=100`);

	if (res.ok) {
		const scenes = scrapeAll(Object.values(res.body), channel, context.parameters);

		return {
			// cursor implies page > 1 and first scene is last scene on previous page,
			// it probably won't trip up the pagination logic, but avoid the duplicate anyway
			scenes: context.cursor ? scenes.slice(1) : scenes,
			context: {
				cursor: Object.keys(res.body).at(-1), // official page seems to derive cursor from last scene, too
			},
		};
	}

	return res.status;
}

async function fetchScene(url, channel, baseScene, { parameters }) {
	if (parameters.layout !== 'organic' && baseScene?.entryId) {
		// overview and deep data is the same in elastic API, don't hit server unnecessarily
		return baseScene;
	}

	const sceneSlug = new URL(url).pathname.match(/\/([\w-]+$)/)[1];

	const res = await http.get(parameters.layout === 'organic'
		? `https://store.psmcdn.net/${parameters.endpoint}/moviesContent/${sceneSlug}.json`
		: `https://store2.psmcdn.net/${parameters.endpoint}-videoscontent/_doc/${sceneSlug}`);

	if (res.ok && res.body.found) {
		return scrapeScene(res.body._source, channel, parameters);
	}

	if (res.ok && parameters.layout === 'organic' && res.body.id) {
		return scrapeScene(res.body, channel, parameters);
	}

	return res.status;
}

async function fetchProfile(baseActor, { entity, parameters }) {
	// const url = format(parameters.profiles, { slug: baseActor.slug });
	const url = parameters.layout === 'organic'
		? `https://store.psmcdn.net/${parameters.endpoint}/modelsContent/${baseActor.slug}.json`
		: `https://store2.psmcdn.net/${parameters.endpoint}-modelscontent/_doc/${baseActor.slug}`;

	const res = await qu.get(url);

	if (res.ok && res.body) {
		return scrapeProfile(parameters.layout === 'organic' ? res.body : res.body._source || res.body, entity, parameters);
	}

	return res.status;
}

module.exports = {
	fetchLatest,
	fetchScene,
	fetchProfile,
	organic: {
		fetchLatest: fetchLatestOrganic,
		fetchScene,
	},
};