traxxx/src/scrapers/littlecapricedreams.js

'use strict';

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

function matchChannel(release, channel) {
	const series = channel.children || channel.parent.children;

	const serieNames = series.reduce((acc, serie) => ({
		...acc,
		[serie.name]: serie,
		[serie.slug]: serie,
	}), {
		vr: 'littlecapricevr',
	});

	const serieName = release.title.match(new RegExp(Object.keys(serieNames).join('|'), 'i'))?.[0];
	const serie = serieName && serieNames[slugify(serieName, '')];

	console.log(release.title);
	console.log(serieName);

	if (serie) {
		return {
			channel: serie.slug,
			title: release.title.replace(new RegExp(`${serieName}[\\s:–-]*`), ''),
		};
	}

	return null;
}

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

		release.url = query.url('a');
		release.entryId = query.q(el, null, 'id')?.match(/post-(\d+)/)?.[1];

		release.title = query.cnt('.meta h3');
		release.date = query.date('.meta .post-meta', 'MMMM D, YYYY');

		release.poster = query.img('img');

		return {
			...release,
			...matchChannel(release, channel),
		};
	});
}

async function fetchPhotos(url) {
	if (url) {
		const res = await qu.get(url, '.et_post_gallery');

		if (res.ok) {
			return res.item.query.urls('a').map(imgUrl => ({
				src: imgUrl,
				referer: url,
			}));
		}
	}

	return null;
}

async function scrapeScene({ query }, url, channel, include) {
	const release = {};

	const script = query.cnt('script.yoast-schema-graph');
	const data = script && JSON.parse(script);

	release.entryId = query.q('article.project', 'id')?.match(/post-(\d+)/)?.[1];

	release.title = query.cnt('.vid_title');
	release.description = query.cnt('.vid_desc p');

	release.date = query.date('.vid_date', 'MMMM D, YYYY');
	release.duration = query.dur('.vid_length');

	release.actors = query.all('.vid_infos a[href*="author/"]').map(actorEl => ({
		name: query.cnt(actorEl),
		url: query.url(actorEl, null),
	}));

	release.tags = query.cnts('.vid_infos a[rel="tag"]');

	const posterData = data['@graph']?.find(item => item['@type'] === 'ImageObject');

	const poster = posterData?.url
		|| query.q('meta[property="og:image"]', 'content')
		|| query.q('meta[name="twitter:image"]', 'content');

	release.poster = {
		src: poster,
		referer: url,
	};

	release.stars = Math.min(Number(query.q('.post-ratings-image', 'title')?.match(/average:\s*(\d\.\d+)/)?.[1]), 5) || null; // rating out of 5, yet sometimes 5.07?

	if (include.photos) {
		release.photos = await fetchPhotos(query.url('.vid_buttons a[href*="project/"]'));
	}

	return {
		...release,
		...matchChannel(release, channel),
	};
}

async function fetchLatest(channel) {
	// no apparent pagination, all updates on one page
	// using channels in part because main overview contains indistinguishable photo albums
	const res = await qu.getAll(channel.url, '.project');

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

	return res.status;
}

async function fetchScene(url, channel, baseRelease, include) {
	const res = await qu.get(url);

	if (res.ok) {
		return scrapeScene(res.item, url, channel, include);
	}

	return res.status;
}

module.exports = {
	fetchLatest,
	fetchScene,
};