traxxx/src/scrapers/spizoo.js

'use strict';

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

const slugify = require('../utils/slugify');

function getEntryId(url) {
	return slugify(new URL(url).pathname.match(/\/updates\/(.*)\.html/)?.[1], '-');
}

function scrapeAll(scenes) {
	return scenes.map(({ query }) => {
		const release = {};

		release.url = query.url('a');
		release.entryId = getEntryId(release.url);

		release.title = query.content('.title-label a, .thumb-title a, .p-7, .text h3');
		release.date = query.date('.date-label', 'MM/DD/YYYY');

		release.actors = query.all(['.update_models a', '.tour_update_models a', '.pornstar-label span']).map((el) => ({
			name: unprint.query.content(el),
			url: unprint.query.url(el, null),
		}));

		release.poster = query.img('a img');
		release.teaser = query.video('.leVideo source');

		return release;
	});
}

function scrapeScene({ query }, { url, entity }) {
	const release = {};

	release.entryId = getEntryId(url);
	release.title = query.content(['#media-holder .title', '.content-holder h1', '#scene h1', 'h2.titular', 'title'])?.replace(/\s+-$/, '');

	release.date = query.date('#sceneInfo .date, #trailer-data .date', 'YYYY-MM-DD');
	release.duration = query.duration('#sceneInfo .data-others, #trailer-data', /\d+:\d+/);

	release.description = query.content('#sceneInfo .description, #trailer-data > div:first-child p');

	release.actors = query.all('#sceneInfo .data-others a[href*="/models"], #trailer-data a[href*="/models"]').map((el) => ({
		name: unprint.query.attribute(el, null, 'title'),
		url: unprint.query.url(el, null, { origin: entity.url }),
	}));

	release.tags = query.contents('.categories-holder a, #sceneInfo a[href*="/categories"], #trailer-data a[href*="/categories"]').map((tag) => tag.trim());

	const poster = query.img(['#video-holder .update_thumb', '#video-holder .thumb', '#noMore .update_thumb', '#hpromo .update_thumb', '.trailer-thumb']) || query.poster('#trailervideo');
	const posterPathname = poster && new URL(poster)?.pathname;

	release.poster = [poster, poster?.replace(/imgw=\w+/, 'imgw=680')];

	release.photos = query.imgs('.photos-holder img')
		.filter((src) => new URL(src).pathname !== posterPathname)
		.map((src) => [
			src.replace(/imgw=\d+/, 'imgw=1284'),
			src,
		]);

	release.trailer = query.video('#trailervideo source[type="video/mp4"], #FulsSizeVideo source[type="video/mp4"]'); // sic
	release.teaser = query.video('#trailer-video source[src*="/videothumbs"]');

	return release;
}

function scrapeProfile({ query }) {
	const profile = {};
	const bioKeys = query.contents('.statsText b');
	const bioValues = query.text('.statsText', { join: false });

	const bio = bioKeys.reduce((acc, key, index) => ({
		...acc,
		[slugify(key, '_')]: bioValues[index],
	}), {});

	profile.description = query.contents('.descriptionText');

	profile.avatar = [
		query.img('.model-bio-pic img', { attribute: 'src0_2x' }),
		query.img('.model-bio-pic img', { attribute: 'src0_3x' }), // unnecessarily big
		query.img('.model-bio-pic img', { attribute: 'src0_1x' }),
	];

	profile.height = Number(bio.height?.match(/(\d+)\s?cm/i)?.[1]);
	profile.dateOfBirth = unprint.extractDate(bio.date_of_birth, 'MMMM D, YYYY');

	profile.measurements = bio.measurements;
	profile.butt = bio.ass_type;
	profile.pussy = bio.pussy_type;

	profile.ethnicity = bio.ethnicity;
	profile.hairColor = bio.hair_color;
	profile.eyes = bio.eye_color;
	profile.nationality = bio.nationality;

	if (/tattoo/i.test(bio.body_art)) {
		profile.hasTattoos = true;
	}

	if (/piercing/i.test(bio.body_art)) {
		profile.hasPiercings = true;
	}

	return profile;
}

async function fetchLatest(channel, page) {
	const res = await unprint.get(`${channel.url}${format(channel.parameters?.latest || '/categories/movies_{page}_d.html', { page })}`, {
		selectAll: '.thumb-big, .thumb-video, .thumbnail, .thumbnail-popular, .full-thumbnail',
	});

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

	return res.status;
}

async function fetchProfile(actor, channel) {
	if (actor.url) {
		const res = await unprint.get(actor.url);

		if (res.ok) {
			return scrapeProfile(res.context);
		}
	}

	const resA = await unprint.get(`${channel.url}/models/${slugify(actor.name)}.html`);

	if (resA.ok) {
		return scrapeProfile(resA.context, channel);
	}

	const resB = await unprint.get(`${channel.url}/models/${slugify(actor.name, '')}.html`);

	if (resB.ok) {
		return scrapeProfile(resB.context, channel);
	}

	return resB.status;
}

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