traxxx/src/scrapers/kink.js

216 lines
5.8 KiB
JavaScript
Raw Normal View History

'use strict';
const unprint = require('unprint');
const qu = require('../utils/qu');
const http = require('../utils/http');
const slugify = require('../utils/slugify');
function scrapeAll(scenes) {
return scenes.map(({ query }) => {
const release = {};
const href = query.url('.shoot-link');
release.url = `https://www.kink.com${href}`;
release.shootId = href.split('/').slice(-1)[0];
release.entryId = release.shootId;
release.title = query.content('.shoot-thumb-title a', true);
release.date = query.date('.date', 'MMM DD, YYYY');
release.actors = query.all('.shoot-thumb-models a').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null, { origin: 'https://www.kink.com' }),
}));
release.rating = query.number('.thumb-ratings') / 10;
release.poster = query.img('.adimage');
release.photos = query.imgs('.rollover .roll-image', { attribute: 'data-imagesrc' }).map((photo) => [
photo.replace('410/', '830/'),
photo,
]);
release.duration = query.dur('.video span');
return release;
});
}
function scrapeScene({ query }, url) {
const release = { url };
release.shootId = new URL(url).pathname.split('/')[2];
release.entryId = release.shootId;
release.title = query.attribute('.shoot-title .favorite-button', 'data-title') || query.content('.shoot-title');
release.description = query.content('.description-text');
release.date = query.date('.shoot-date', 'MMMM DD, YYYY');
release.actors = query.elements('.names a').map((actorEl) => ({
name: unprint.query.content(actorEl).replace(/,\s*/, ''),
url: unprint.query.url(actorEl, null, { origin: 'https://www.kink.com' }),
}));
release.director = query.content('.director-name');
release.photos = query.imgs('.gallery .thumb img, #gallerySlider .gallery-img', 'data-image-file');
release.poster = query.poster();
release.tags = query.contents('.tag-list a[href*="/tag"]').map((tag) => tag.replace(/,\s*/, ''));
const trailer = query.attribute('.player span[data-type="trailer-src"]', 'data-url');
if (trailer) {
release.trailer = [
{
src: trailer.replace('480p', '1080p'),
quality: 1080,
},
{
src: trailer.replace('480p', '720p'),
quality: 720,
},
{
src: trailer,
quality: 480,
},
{
src: trailer.replace('480p', '360p'),
quality: 360,
},
];
}
release.channel = slugify(query.url('.shoot-logo a')?.split('/').slice(-1)[0], '');
console.log(release);
return release;
}
async function fetchActorReleases(actorUrl, page = 1, accReleases = []) {
const res = await qu.get(`${actorUrl}?page=${page}`);
if (res.ok) {
const releases = scrapeAll(qu.initAll(res.item.el, '.shoot-list .shoot'));
const hasNextPage = res.item.query.exists('.paginated-nav li:last-child:not(.disabled)');
if (hasNextPage) {
return fetchActorReleases(actorUrl, page + 1, accReleases.concat(releases));
}
return accReleases.concat(releases);
}
return accReleases;
}
async function scrapeProfile({ query }, actorUrl, include) {
const profile = {};
profile.description = query.q('.bio #expand-text', true);
const tags = query.all('.bio-tags a', true);
if (tags.includes('brunette') || tags.includes('brunet')) profile.hairColor = 'brown';
if (tags.includes('blonde') || tags.includes('blond')) profile.hairColor = 'blonde';
if (tags.includes('black hair')) profile.hairColor = 'black';
if (tags.includes('redhead')) profile.hairColor = 'red';
if (tags.includes('natural boobs')) profile.naturalBoobs = true;
if (tags.includes('fake boobs')) profile.naturalBoobs = false;
if (tags.includes('white')) profile.ethnicity = 'white';
if (tags.includes('latin')) profile.ethnicity = 'latin';
if (tags.includes('Black')) profile.ethnicity = 'black';
if (tags.includes('pierced nipples')) profile.hasPiercings = true;
if (tags.includes('tattoo')) profile.hasTattoos = true;
if (tags.includes('foreskin')) profile.hasForeskin = true;
if ((tags.includes('big dick') || tags.includes('foreskin'))
&& (tags.includes('fake boobs') || tags.includes('big tits'))) profile.gender = 'transsexual';
profile.avatar = query.img('.bio-slider-img, .bio-img:not([src*="Missing"])');
profile.social = query.urls('a.social-link');
if (include.releases) {
profile.releases = await fetchActorReleases(actorUrl);
}
return profile;
}
async function fetchLatest(site, page = 1) {
const { tab } = await http.getBrowserSession('kink', { headless: false });
const res = await tab.goto(`https://www.kink.com/search?type=shoots&channelIds=${site.slug}&sort=published&page=${page}`);
const status = res.status();
if (status === 200) {
const html = await tab.content();
const items = unprint.initAll(html, '.results .shoot-card');
const scenes = scrapeAll(items, site);
await tab.close();
return scenes;
}
return status;
}
async function fetchScene(url, channel) {
const { tab } = await http.getBrowserSession('kink');
const res = await tab.goto(url);
const status = res.status();
if (status === 200) {
const html = await tab.content();
const item = unprint.init(html);
const scene = scrapeScene(item, url, channel);
await tab.close();
return scene;
}
return status;
}
async function fetchProfile({ name: actorName }, entity, include) {
const searchRes = await qu.getAll(`https://kink.com/search?type=performers&q=${actorName}`, '.model');
if (searchRes.ok) {
const actorItem = searchRes.items.find((item) => item.query.exists(`.model-link img[alt="${actorName}"]`));
if (actorItem) {
const actorPath = actorItem.query.url('.model-link');
const actorUrl = `https://kink.com${actorPath}`;
const actorRes = await qu.get(actorUrl);
if (actorRes.ok) {
return scrapeProfile(actorRes.item, actorUrl, include);
}
return actorRes.status;
}
return null;
}
return searchRes.status;
}
module.exports = {
// beforeNetwork,
fetchLatest,
fetchScene,
fetchProfile,
};