traxxx/src/scrapers/teamskeet.js

'use strict';

const format = require('template-format');

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;
	}

	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) {
	const release = {};

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

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

	release.actors = scene.models?.map((model) => model.modelName) || [];
	release.actors = scene.models?.map((model) => ({
		name: model.modelName,
		avatar: `https://images.mylfcdn.net/tsv4/model/profiles/${slugify(model.modelName, '_')}.jpg`,
		url: `${channel.url}/models/www.mylf.com/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 = { stream: `https://videodelivery.net/${scene.video}/manifest/video.mpd` };
	}

	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) {
	return scenes.map(({ _source: scene }) => scrapeScene(scene, channel));
}

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

	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.scenes = actor.movies?.map((scene) => scrapeScene(scene, entity));

	return profile;
}

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

	if (res.ok) {
		return scrapeAll(res.body.hits.hits, channel);
	}

	return res.status;
}

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

	const sceneSlug = new URL(url).pathname.match(/\/([\w-]+$)/)[1];
	const res = await http.get(`${parameters.videos}/${sceneSlug}`);

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

	return res.status;
}

async function fetchProfile(baseActor, { entity, parameters }) {
	const url = format(parameters.profiles, { slug: baseActor.slug });
	const res = await qu.get(url);

	if (res.ok && res.body) {
		return scrapeProfile(res.body._source, entity);
	}

	return res.status;
}

module.exports = {
	fetchLatest,
	fetchScene,
	fetchProfile,
};