traxxx/src/scrapers/jayrock.js

106 lines
2.6 KiB
JavaScript
Raw Normal View History

2020-01-10 21:10:11 +00:00
'use strict';
const qu = require('../utils/qu');
2020-01-10 21:10:11 +00:00
function scrapeLatest(items, channel) {
return items.map(({ query }) => {
const release = {};
2020-01-10 21:10:11 +00:00
release.url = query.url('h5 a', null, { origin: channel.url });
release.entryId = new URL(release.url).pathname.match(/\/(\d+)/)[1];
2020-01-10 21:10:11 +00:00
release.title = query.cnt('h5 a');
2020-01-10 21:10:11 +00:00
[release.poster, ...release.photos] = query.imgs('.screenshot').map(src => [
// unnecessarily large
// src.replace(/\/\d+/, 3840),
// src.replace(/\/\d+/, '/2000'),
src.replace(/\/\d+/, '/1500'),
src.replace(/\/\d+/, '/1000'),
src,
]);
2020-01-10 21:10:11 +00:00
return release;
});
2020-01-10 21:10:11 +00:00
}
function scrapeScene({ query, html }, url, channel) {
const release = {};
release.entryId = new URL(url).pathname.match(/\/(\d+)/)[1];
release.title = query.cnt('h1.description');
release.actors = query
.all('.video-performer')
.map((actorEl) => {
const actorUrl = query.url(actorEl, 'a', 'href', { origin: channel.url });
const entryId = new URL(url).pathname.match(/\/(\d+)/)?.[1];
const avatar = query.img(actorEl, 'img:not([data-bgsrc*="not-available"])', 'data-bgsrc');
return {
name: query.cnt(actorEl, '.video-performer-name'),
gender: 'female',
avatar: avatar && [
avatar.replace(/\/actor\/(\d+)/, '/actor/500'),
avatar,
],
url: actorUrl,
entryId,
};
})
.concat({ name: 'Jay Rock', gender: 'male' });
release.date = query.date('.release-date:first-child', 'MMM DD, YYYY', /\w+ \d{1,2}, \d{4}/);
release.duration = query.number('.release-date:last-child') * 60;
release.studio = query.cnt('.studio span:nth-child(2)');
release.director = query.text('.director');
release.tags = query.cnts('.tags a');
const poster = html.match(/url\((https.+\.jpg)\)/)?.[1];
const photos = query.imgs('#moreScreenshots img');
[release.poster, ...release.photos] = [poster]
.concat(photos)
.filter(Boolean)
.map(src => [
src.replace(/\/(\d+)\/\d+/, '/$1/1500'),
src.replace(/\/(\d+)\/\d+/, '/$1/1000'),
src,
]);
const videoId = html.match(/item: (\d+)/)?.[1];
if (videoId) {
release.trailer = { stream: `https://trailer.adultempire.com/hls/trailer/${videoId}/master.m3u8` };
}
2020-01-10 21:10:11 +00:00
return release;
2020-01-10 21:10:11 +00:00
}
async function fetchLatest(channel, page = 1) {
const res = await qu.getAll(`https://jayspov.net/jays-pov-updates.html?view=list&page=${page}`, '.item-grid-list-view > .grid-item');
2020-01-10 21:10:11 +00:00
if (res.ok) {
return scrapeLatest(res.items, channel);
}
2020-01-10 21:10:11 +00:00
return res.status;
2020-01-10 21:10:11 +00:00
}
async function fetchScene(url, channel) {
const res = await qu.get(url);
2020-01-10 21:10:11 +00:00
if (res.ok) {
return scrapeScene(res.item, url, channel);
}
2020-01-10 21:10:11 +00:00
return res.status;
2020-01-10 21:10:11 +00:00
}
module.exports = {
fetchLatest,
fetchScene,
2020-01-10 21:10:11 +00:00
};