traxxx/src/scrapers/missax.js

92 lines
2.6 KiB
JavaScript

'use strict';
const unprint = require('unprint');
const slugify = require('../utils/slugify');
function scrapeAll(scenes) {
return scenes.map(({ query }) => {
const release = {};
release.url = query.url('.photo-thumb_body > a');
release.entryId = new URL(release.url).pathname.match(/\/trailers\/([\w-]+).html/)[1].toLowerCase();
release.title = query.content('.thumb-title');
release.actors = query.all('.model-name a').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}));
release.poster = Array.from({ length: 4 }, (item, index) => query.img('img.mainThumb', { attribute: `src0_${4 - index}x` }));
return release;
});
}
function scrapeScene({ query, html }, { url, entity }) {
const release = {};
release.entryId = new URL(url).pathname.match(/\/trailers\/([\w-]+).html/)[1].toLowerCase();
release.title = query.content('.raiting-section__title');
release.description = query.contents('p.text, p.text ~ p').map((p) => p.trim()).join('\n\n');
release.date = query.date('.dvd-scenes__data', 'MM/DD/YYYY');
release.duration = query.duration('.dvd-scenes__data');
release.actors = query.all('.dvd-scenes__data a[href*="/models"]').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}));
release.tags = query.contents('.dvd-scenes__data a[href*="/categories"]');
release.poster = Array.from({ length: 4 }, (item, index) => query.img('img.update_thumb', { attribute: `src0_${4 - index}x` }));
release.trailer = unprint.prefixUrl(html.match(/src="(\/trailers\/\w+\.mp4)"/)?.[1], entity.url);
return release;
}
function scrapeProfile({ query }) {
const profile = {};
profile.description = query.contents('p.text, p.text ~ p')?.map((p) => p.trim()).join('\n\n');
profile.avatar = Array.from({ length: 3 }, (item, index) => query.img('.bio-img img', { attribute: `src0_${3 - index}x` }));
return profile;
}
async function fetchLatest(channel, page = 1) {
const url = `${channel.url}/tour/categories/movies_${page}_d.html`;
const res = await unprint.get(url, { selectAll: '.tab-content .video-thumb' });
if (res.ok) {
return scrapeAll(res.context, channel);
}
return res.status;
}
async function fetchProfile({ name }, entity) {
const res = await unprint.get(`${entity.url}/tour/models/${name.replace(/\s+/, '')}.html`);
if (res.ok) {
return scrapeProfile(res.context, entity);
}
const slugRes = await unprint.get(`${entity.url}/tour/models/${slugify(name, '-')}.html`);
if (slugRes.ok) {
return scrapeProfile(slugRes.context, entity);
}
return slugRes.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene,
};