traxxx/src/scrapers/innofsin.js

101 lines
2.4 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
const slugify = require('../utils/slugify');
function scrapeAll(scenes) {
return scenes.map(({ query }) => {
const release = {};
release.url = query.url('a');
release.entryId = new URL(release.url).pathname.replaceAll('/', '');
release.title = query.content('h2 a');
release.date = query.date('.date', 'MMM DD, YYYY');
release.actors = query.content('.modelname').split(',').filter(Boolean);
release.poster = query.img('.imgc img');
release.teaser = query.video('.vidhover source');
console.log(release);
return release;
});
}
function scrapeScene({ query }, { url }) {
const release = {};
release.entryId = new URL(url).pathname.replaceAll('/', '');
release.title = query.content('.vid-box h1');
release.description = query.content('//div[contains(@class, "vid-box")]/*[contains(text(), "About")]/following-sibling::p[2]');
release.actors = query.all('.vid-box a[href*="/pornstars"]').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}));
release.poster = query.poster('.vid-play video');
release.photos = query.imgs('.img-sample img').map((source) => [
source.replace(/-\d+x\d+/, ''), // usually same size, but try anyway
source,
]);
release.tags = query.contents('.vid-box [itemprop="keywords"] a');
release.trailer = query.video('.vid-play source');
console.log(release);
return release;
}
function scrapeProfile({ query }, actorName, entity, include) {
const profile = {};
profile.description = query.cnt('.bio-text');
profile.birthPlace = query.cnt('.birth-place span');
profile.avatar = query.img('.actor-photo img');
if (include.releases) {
return scrapeAll(unprint.initAll(query.all('.scene')));
}
console.log(profile);
return profile;
}
async function fetchLatest(channel, page = 1) {
const url = `${channel.url}/videos/page/${page}/`;
const res = await unprint.get(url, { selectAll: '.vid-listing li' });
if (res.ok) {
return scrapeAll(res.context, channel);
}
return res.status;
}
async function fetchProfile({ name: actorName }, entity, include) {
const url = `${entity.url}/actors/${slugify(actorName, '_')}`;
const res = await unprint.get(url);
if (res.ok) {
return scrapeProfile(res.item, actorName, entity, include);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene: {
scraper: scrapeScene,
unprint: true,
},
};