traxxx/src/scrapers/archangel.js

'use strict';

// ALSO USED BY THE FLOURISH

const unprint = require('unprint');

const slugify = require('../utils/slugify');
const { feetInchesToCm } = require('../utils/convert');

const placeholder = /images\/p\d+\.jpe?g/i;

function getEntryId(release) {
	return slugify(new URL(release.url).pathname.match(/\/([\w-]+)\.html/)?.[1]
		|| [unprint.formatDate(release.date, 'YYYY-MM-DD'), release.title, ...release.actors]);
}

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

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

		release.title = query.content('a span');

		release.date = query.date('.timeDate', 'YYYY-MM-DD');
		release.duration = query.duration('.timeDate');

		release.actors = query.all('a[href*="models/"], a[href*="sets.php"]').map((actorEl) => ({
			name: unprint.query.content(actorEl),
			url: unprint.query.url(actorEl, null),
		}));

		const poster = query.img('img.mainThumb');
		const previewCount = query.number('img.mainThumb', { attribute: 'cnt' });

		if (poster && !placeholder.test(poster)) {
			const posterFallbacks = [
				poster.replace('-1x', '-3x'),
				poster.replace('-1x', '-2x'),
				poster.replace('-1x', '-4x'),
				poster,
			];

			release.poster = posterFallbacks;
		}

		if (previewCount) {
			release.photos = Array.from(
				{ length: previewCount - 1 },
				(value, index) => [3, 2, 4, 1].map((scale) => unprint.prefixUrl(query.img('img.mainThumb', { attribute: `src${index + 1}_${scale}x` }))).filter(Boolean), // 4x is unnecessarily big and possibly upscaled
			).filter(Boolean);
		}

		release.photoCount = query.number('.timeDate', { match: /(\d+) photos/i, matchIndex: 1 });

		release.entryId = getEntryId(release);

		return release;
	});
}

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

	release.title = query.content('.title h2');
	release.description = query.content('.description p');

	release.date = query.date('.info p', 'MMMM D, YYYY');
	release.duration = query.duration('.info p');

	release.actors = query.all('.info a[href*="models/"], .info a[href*="sets.php"]').map((actorEl) => ({
		name: unprint.query.content(actorEl),
		url: unprint.query.url(actorEl, null),
	}));

	const poster = unprint.prefixUrl(query.img('.update_thumb') || html.match(/poster="(.*\.jpg)"/)?.[1], entity.url);

	if (poster && !placeholder.test(poster)) {
		const posterFallbacks = [
			poster.replace('-1x', '-3x'),
			poster.replace('-1x', '-2x'),
			poster.replace('-1x', '-4x'),
			poster,
		];

		// scene page poster usually different from overview page, don't replace
		if (baseRelease?.poster && baseRelease.poster !== poster) {
			release.photos = baseRelease.photos
				? [posterFallbacks, ...baseRelease.photos]
				: [posterFallbacks];
		} else {
			release.poster = posterFallbacks;
		}
	}

	const trailer = html.match(/src="(.*\.mp4)"/)?.[1];

	if (trailer) {
		release.trailer = unprint.prefixUrl(encodeURI(trailer), entity.url);
	}

	release.tags = query.contents('.info .tags a');

	release.photoCount = query.number('.info', { match: /(\d+) photos/i, matchIndex: 1 });
	release.entryId = getEntryId(release);

	return release;
}

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

	release.title = query.content('.title h2');
	release.description = query.content('.aboutArea p');

	release.covers = [[
		query.img('.update_thumb', { attribute: 'src0_2x', origin: entity.url }),
		query.img('.update_thumb', { attribute: 'src0_1x', origin: entity.url }),
		query.img('.update_thumb', { attribute: 'src0', origin: entity.url }),
		// usually upscaled
		query.img('.update_thumb', { attribute: 'src0_4x', origin: entity.url }),
		query.img('.update_thumb', { attribute: 'src0_3x', origin: entity.url }),
	].filter(Boolean)];

	release.entryId = getEntryId(release);

	release.scenes = scrapeAll(unprint.initAll(element, '.item-video'));

	return release;
}

function scrapeProfile({ query, element }, { url, entity }) {
	const profile = { url };

	const bio = Object.fromEntries(query.all('.stats li')
		.map((row) => [
			slugify(unprint.query.content(row, '.data-name, span'), '_'),
			unprint.query.text(row),
		])
		.filter(([key, value]) => key && value));

	profile.description = query.content('.aboutArea p');

	profile.birthPlace = bio.place_of_birth;
	profile.dateOfBirth = unprint.extractDate(bio.age, 'MMMM D, YYYY');

	profile.height = Number(bio.height?.match(/(\d+)\s*cm/)?.[1]) || (/\d fe*t \d+ inch/i.test(bio.height) && feetInchesToCm(bio.height)) || null;
	profile.measurements = bio.measurements;

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

	profile.avatar = [
		query.img('.model_bio_thumb', { attribute: 'src0_4x', origin: entity.url }),
		query.img('.model_bio_thumb', { attribute: 'src0_3x', origin: entity.url }),
		query.img('.model_bio_thumb', { attribute: 'src0_2x', origin: entity.url }),
		query.img('.model_bio_thumb', { attribute: 'src0_1x', origin: entity.url }),
		query.img('.model_bio_thumb', { attribute: 'src0', origin: entity.url }),
	].filter((avatar) => avatar && !placeholder.test(avatar));

	profile.scenes = scrapeAll(unprint.initAll(element, '.item-video'));

	return profile;
}

async function fetchLatest(channel, page = 1, context) {
	const url = `${channel.url}${context.parameters.path || ''}/categories/movies_${page}_d.html`;
	const res = await unprint.get(url, { selectAll: '.item-video' });

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

	return res.status;
}

async function fetchProfile({ name: actorName, url: actorUrl }, { entity, include, parameters }) {
	const res = await [
		actorUrl,
		`${entity.url}${parameters.path || ''}/models/${slugify(actorName, '-')}.html`,
		`${entity.url}${parameters.path || ''}/models/${slugify(actorName, '')}.html`,
	].reduce(async (chain, url) => {
		const prevRes = await chain;

		if (prevRes.ok || !url) {
			return prevRes;
		}

		const actorRes = await unprint.get(url);

		if (actorRes.ok) {
			return {
				...actorRes,
				url,
			};
		}

		return prevRes;
	}, Promise.resolve({ ok: false, status: null }));

	if (res.ok) {
		return scrapeProfile(res.context, { entity, include, url: res.url });
	}

	return res.status;
}

module.exports = {
	fetchLatest,
	fetchProfile,
	scrapeScene: {
		scraper: scrapeScene,
		unprint: true,
	},
	scrapeMovie: {
		scraper: scrapeMovie,
		unprint: true,
	},
};