traxxx/src/scrapers/teenmegaworld.js

119 lines
3.3 KiB
JavaScript
Raw Normal View History

2021-09-17 01:30:49 +00:00
'use strict';
const qu = require('../utils/q');
const slugify = require('../utils/slugify');
function scrapeAll(scenes, channel) {
return scenes.map(({ query }) => {
const release = {};
const network = channel.type === 'network' ? channel : channel.parent;
release.url = query.url('a.title');
release.entryId = new URL(release.url).pathname.match(/\/trailers\/(.+).html/)?.[1].toLowerCase();
release.title = query.cnt('.title');
release.date = query.date('time', 'MMMM D, YYYY');
release.actors = query.all('.actors a').map(el => ({
name: query.cnt(el),
url: query.url(el, null),
}));
[release.poster, ...release.photos] = Object
.entries(query.el('.images img').dataset)
.filter(([key]) => /src/.test(key))
.map(([, value]) => qu.prefixUrl(value, network.url));
release.likes = query.number('.rating-on-thumb');
const siteId = query.url('.site a', 'href', { origin: network.url, object: true })?.searchParams.get('site[]');
if (siteId) {
release.channel = network.children.find(child => child.parameters.siteId.toString() === siteId)?.slug;
}
return release;
});
}
function scrapeScene({ query }, url, channel) {
const release = {};
const network = channel.type === 'network' ? channel : channel.parent;
release.entryId = new URL(url).pathname.match(/\/trailers\/(.+).html/)[1].toLowerCase();
release.title = query.cnt('.title-line h1');
release.description = query.cnt('.shadow .text');
release.date = query.date('.title-line .date', 'MMMM D, YYYY');
release.duration = query.number('.dur') * 60;
release.actors = query.all('.site a[href*="/models"]').map(el => ({
name: query.cnt(el),
url: query.url(el, null),
}));
release.tags = query.cnts('.tag-list a');
release.poster = query.poster();
release.trailer = query.video();
release.likes = query.number('.rating-data .value');
const siteId = query.url('.site a[href*="site[]"]', 'href', { origin: network.url, object: true })?.searchParams.get('site[]');
if (siteId) {
release.channel = network.children.find(child => child.parameters.siteId.toString() === siteId)?.slug;
}
return release;
}
function scrapeProfile({ query, el }, url, entity) {
const profile = {};
const bio = query.all('.stats li').reduce((acc, bioEl) => ({
...acc,
[slugify(query.cnt(bioEl, '.title'))]: query.cnt(bioEl, '.value'),
}), {});
profile.url = url;
profile.description = query.cnt('.bio .text');
profile.hairColor = bio.hair;
profile.eyes = bio.eyes;
profile.avatar = query.img('.model-info .photo img', 'data-src', { origin: entity.url });
profile.scenes = scrapeAll(qu.initAll(el, '.video-list .video[data-type="vids"]'), entity);
return profile;
}
async function fetchLatest(channel, page = 1) {
const url = `${channel.type === 'network' ? channel.url : channel.parent.url}/search.php?site[]=${channel.parameters.siteId}&page=${page}`;
const res = await qu.getAll(url, '.video-list .video');
if (res.ok) {
return scrapeAll(res.items, channel);
}
return res.status;
}
async function fetchProfile({ name: actorName }, entity, include) {
const url = `${entity.url}/models/${slugify(actorName)}.html`;
const res = await qu.get(url);
if (res.ok) {
return scrapeProfile(res.item, url, entity, include);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene,
};