traxxx/src/scrapers/fullpornnetwork.js

121 lines
3.7 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
const slugify = require('../utils/slugify');
const { convert } = require('../utils/convert');
const tryUrls = require('../utils/try-urls');
function scrapeAll(scenes, channel) {
return scenes.map(({ query }) => {
const release = {};
release.url = query.url('a');
release.entryId = slugify(new URL(release.url).pathname.match(/trailers\/(.*)/)[1]);
release.title = query.content('h2 a');
release.duration = query.duration('.video-data');
release.actors = query.all('a[href*="models/"], a[href*="sets.php"]').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null, { origin: channel.url }),
}));
const poster = query.img('.thumbnail img');
const teaser = query.video('.thumbnail img', { attribute: 'data-vid' }); // not a mistake, video source is on img tag
if (!poster?.includes('blur')) { // seemingly global SFW
release.poster = poster;
}
if (!teaser?.includes('blur')) { // seemingly global SFW
release.teaser = teaser;
}
return release;
});
}
async function fetchLatest(channel, page = 1) {
const section = channel.pathname?.match(/\/channels\/([\w-]+)/)?.[1] || 'movies';
const url = `${channel.origin}/porn-categories/${section}/?page=${page}&sort=most-recent`; // parameter order matters for some reason!
const res = await unprint.get(url, { selectAll: '.content div[data-setid]' });
if (res.ok) {
return scrapeAll(res.context, channel);
}
return res.status;
}
function scrapeScene({ query }, { url, entity }) {
const release = {};
release.entryId = slugify(new URL(url).pathname.match(/trailers\/(.*)/)[1]);
release.title = query.content('h1.title_bar');
release.description = query.content('.description-text, #description');
release.date = query.date('//label[contains(text(), \'Date\')]/following-sibling::p[1]', 'YYYY-MM-DD')
|| query.date('//label[contains(text(), \'Date Added\')]/following-sibling::text()[1]', 'YYYY-MM-DD');
release.actors = query.all('#preview a[href*="/models"]').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null, { origin: entity.origin }),
}));
release.tags = query.contents('#preview a[href*="categories/"]');
release.poster = query.poster('#preview video');
release.trailer = query.video('#preview video source');
return release;
}
function scrapeProfile({ query }, { url }) {
const profile = { url };
const bio = Object.fromEntries(query.all('.model-details > div').map((bioEl) => [
slugify(unprint.query.content(bioEl, 'h2'), '_'),
unprint.query.text(bioEl),
]));
profile.avatar = [
query.img('.model_bio_thumb', { attribute: 'src0_3x' }),
query.img('.model_bio_thumb', { attribute: 'src0_2x' }),
query.img('.model_bio_thumb', { attribute: 'src0_1x' }),
query.img('.model_bio_thumb'),
].filter(Boolean);
profile.description = [query.content('.model-bio-text, #performer-description'), bio.funfact].filter(Boolean).join(' ');
profile.aliases = bio.alias?.split(/[,\n]/).map((alias) => alias.trim());
profile.age = parseInt(bio.age, 10) || null;
profile.dateOfBirth = unprint.extractDate(bio.age, 'MM/DD/YYYY');
profile.measurements = bio.measurements;
profile.height = Number(bio.height?.match(/(\d+)\s*cm/)?.[1]) || convert(bio.height, 'cm');
return profile;
}
async function fetchProfile({ name: actorName, url: actorUrl }, { entity, include }) {
const { res, url } = await tryUrls([
actorUrl,
`${entity.url}/models/${slugify(actorName, '')}.html`,
`${entity.url}/models/${slugify(actorName, '-')}.html`,
]);
if (res.ok) {
return scrapeProfile(res.context, { entity, include, url });
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene,
};