traxxx/src/scrapers/teenmegaworld.js

126 lines
3.7 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
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('.thumb__title-link');
release.entryId = new URL(release.url).pathname.match(/\/trailers\/(.+).html/)?.[1].toLowerCase();
release.title = query.content('.thumb__title-link');
release.date = query.date('time', unprint.dateConstants.DATETIME_LOCAL_SECONDS, { attribute: 'datetime' }) || query.date('time', 'MMMM D, YYYY');
release.actors = query.all('.actor__link').map((el) => ({
name: unprint.query.content(el),
url: unprint.query.url(el, null),
}));
release.poster = query.img('.thumb__picture img');
[release.poster, ...release.photos] = Object
.entries(query.el('.thumb__image').dataset)
.filter(([key]) => /^src/.test(key))
.map(([, value]) => [value.replace('-1x', '-2x'), value].map((path) => unprint.prefixUrl(path, network.url)));
release.teaser = query.video('.thumb__video', { origin: network.url });
release.likes = query.number('.thumb__rating-value');
const channelSlug = slugify(query.content('.thumb__detail__site-link'), '');
if (channelSlug) {
release.channel = network.children.find((child) => child.slug === channelSlug)?.slug;
}
return release;
});
}
function scrapeScene({ query }, context) {
const release = {};
release.entryId = new URL(context.url).pathname.match(/\/trailers\/(.+).html/)[1].toLowerCase();
release.title = query.content('.video-title');
release.description = query.content('.video-description-text');
release.date = query.date('.video-info-date', 'MMMM D, YYYY');
release.duration = query.number('.video-info-time') * 60;
release.actors = query.all('.video-actor-link').map((el) => ({
name: unprint.query.content(el),
url: unprint.query.url(el, null),
}));
release.tags = query.contents('.video-tag-link');
const poster = query.img('.video-cover__image');
if (!poster.includes('error')) {
release.poster = poster;
}
release.trailer = query.video();
release.likes = query.number('.video-actions-button [data-count]');
release.channel = slugify(query.content('.video-site-link'), '');
return release;
}
function scrapeProfile({ query }, url, entity) {
const profile = {};
const bio = query.all('.model-profile-information-element').reduce((acc, bioEl) => ({
...acc,
[slugify(unprint.query.content(bioEl, '.model-profile-information-key'))]: unprint.query.text(bioEl),
}), {});
profile.url = url;
profile.description = query.content('.model-profile-about');
profile.hairColor = bio.hair;
profile.eyes = bio.eyes;
profile.avatar = query.img('.model-profile-image-picture source', { origin: entity.url, attribute: 'srcset' }) || query.img('.model-profile-image-picture img', { origin: entity.url });
profile.scenes = scrapeAll(unprint.initAll(query.all('.video-list .thumb')), 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=${page}`;
const res = await unprint.get(url, { selectAll: '.video-list .thumb' });
if (res.ok) {
return scrapeAll(res.context, channel);
}
return res.status;
}
async function fetchProfile(actor, entity, include) {
const url = actor.url || `${entity.url}/models/${slugify(actor.name)}.html`;
const res = await unprint.get(url);
if (res.ok) {
return scrapeProfile(res.context, url, entity, include);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene: {
scraper: scrapeScene,
unprint: true,
},
};