traxxx/src/scrapers/private.js

102 lines
3.0 KiB
JavaScript

'use strict';
/* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
const knex = require('../knex');
const { matchTags } = require('../tags');
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.content-wrapper.container .scene').toArray();
return sceneElements.map((element) => {
const sceneLinkElement = $(element).find('a[data-track="TITLE_LINK"]');
const url = sceneLinkElement.attr('href');
const title = sceneLinkElement.text();
const shootId = url.split('/').slice(-1)[0];
const date = moment.utc($(element).find('.scene-date'), 'MM/DD/YYYY').toDate();
const actors = $(element).find('a[data-track="PORNSTAR_LINK"]').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const likes = Number($(element).find('.scene-votes').text());
return {
url,
shootId,
title,
actors,
date,
rating: {
likes,
},
site,
};
});
}
async function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const shootId = url.split('/').slice(-1)[0];
const title = $('.video-wrapper meta[itemprop="name"]').attr('content');
const date = moment.utc($('.video-wrapper meta[itemprop="uploadDate"]').attr('content'), 'MM/DD/YYYY').toDate();
const actors = $('.content-wrapper .scene-models-list a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const description = $('.video-wrapper meta[itemprop="description"]').attr('content');
const [minutes, seconds] = $('.video-wrapper meta[itemprop="duration"]').attr('content').match(/\d+/g);
const duration = Number(minutes) * 60 + Number(seconds);
const likes = Number($('.content-desc #social-actions #likes').text());
const siteElement = $('.content-wrapper .logos-sites a');
const siteUrl = siteElement.attr('href').slice(0, -1);
const siteName = siteElement.text();
const rawTags = $('.content-desc .scene-tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
const [tags, channelSite] = await Promise.all([
matchTags(rawTags),
knex('sites')
.where({ url: siteUrl })
.orWhere({ name: siteName })
.first(),
]);
return {
url,
shootId,
title,
date,
actors,
description,
duration,
tags,
rating: {
likes,
},
site: channelSite || site,
};
}
async function fetchLatest(site, page = 1) {
const res = await bhttp.get(`${site.url}/${page}/`);
return scrapeLatest(res.body.toString(), site);
}
async function fetchScene(url, site) {
const res = await bhttp.get(url);
return scrapeScene(res.body.toString(), url, site);
}
module.exports = {
fetchLatest,
fetchScene,
};