traxxx/src/scrapers/pascalssubsluts.js

113 lines
2.8 KiB
JavaScript
Raw Normal View History

2020-09-14 00:40:27 +00:00
'use strict';
const qu = require('../utils/q');
const slugify = require('../utils/slugify');
2020-09-14 00:40:27 +00:00
const capitalize = require('../utils/capitalize');
const { feetInchesToCm } = require('../utils/convert');
2020-09-14 00:40:27 +00:00
function scrapeAll(months, channel, year) {
return months.map(({ query: queryMonth, el }) => {
const month = queryMonth.cnt('h3');
const scenes = queryMonth.all(el.nextElementSibling, 'li:nth-child(2n)');
return qu.initAll(scenes).map(({ query }) => {
const release = {};
release.url = query.url('a.video-pop-up', 'href', { origin: `${channel.url}/submissive` });
release.entryId = new URL(release.url).searchParams.get('id');
release.title = query.cnt('.updates-item-title h4');
release.date = qu.parseDate(`${month} ${year}`, 'MMMM/YYYY');
release.datePrecision = 'month';
release.actors = [{
name: capitalize(query.q('a.video-pop-up', 'data-modelname'), { uncapitalize: true }),
url: query.url('a.video-pop-up', 'data-modellink', { origin: `${channel.url}/submissive` }),
}].filter(actor => /lockdown/i.test(actor.name));
2020-09-14 00:40:27 +00:00
release.poster = query.img('img');
return release;
});
}).flat();
}
function scrapeScene({ html }, url) {
const release = {};
release.entryId = new URL(url).searchParams.get('id');
release.trailer = html.match(/file: '(.*)'/)[1];
return release;
}
function scrapeProfile({ query }) {
2020-09-14 00:40:27 +00:00
const profile = {};
const bio = query.all('.about-desc li').reduce((acc, el) => {
const key = query.cnt(el, 'strong');
const value = query.text(el);
2020-09-14 00:40:27 +00:00
return {
...acc,
[slugify(key, '_')]: value,
};
}, {});
2020-09-14 00:40:27 +00:00
profile.nationality = bio.nationality;
profile.height = feetInchesToCm(bio.height);
profile.age = bio.age;
profile.hair = bio.hair;
profile.description = query.cnt('.twocolumns');
profile.avatar = query.img('#individual-description img');
// no dates or links available
// profile.releases = scrapeAll(qu.initAll(query.all('.individal-video-item'))); // sic
2020-09-14 00:40:27 +00:00
return profile;
}
async function fetchLatest(channel, page = 1) {
const year = new Date().getFullYear() - (page - 1);
const url = `${channel.url}/submissive/updates.php?y=${year}`;
const res = await qu.getAll(url, '.month');
if (res.ok) {
return scrapeAll(res.items, channel, year);
}
return res.status;
}
async function fetchScene(url, channel) {
const res = await qu.get(url);
if (res.ok) {
return scrapeScene(res.item, url, channel);
}
return res.status;
}
async function fetchProfile({ name: actorName }, entity, include) {
const url = `${entity.url}/submissive/subslut-${slugify(actorName)}.php`;
const res = await qu.get(url, null, null, {
followRedirects: false,
});
2020-09-14 00:40:27 +00:00
if (res.ok && res.status === 200) {
2020-09-14 00:40:27 +00:00
return scrapeProfile(res.item, actorName, entity, include);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchScene,
fetchProfile,
2020-09-14 00:40:27 +00:00
};