traxxx/src/scrapers/realitykings.js

117 lines
3.2 KiB
JavaScript

'use strict';
/* eslint-disable newline-per-chained-call */
const Promise = require('bluebird');
const bhttp = require('bhttp');
const { CookieJar } = Promise.promisifyAll(require('tough-cookie'));
const cheerio = require('cheerio');
const moment = require('moment');
const { matchTags } = require('../tags');
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.card.card--release').toArray();
console.log(sceneElements);
return sceneElements.map((element) => {
const sceneLinkElement = $(element).find('.card-info__title a');
const title = sceneLinkElement.attr('title');
const url = `${site.url}${sceneLinkElement.attr('href')}`;
const entryId = url.split('/').slice(-3)[0];
const date = moment.utc($(element).find('.card-info__meta-date').text(), 'MMMM DD, YYYY').toDate();
const actors = $(element).find('.card-info__cast a').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray();
console.log(date, actors, title);
return {
url,
entryId,
title,
actors,
date,
rating: null,
site,
};
});
}
async function scrapeScene(data, url, site) {
const {
id: entryId,
title,
description,
} = data;
const date = new Date(data.dateReleased);
const actors = data.actors
.sort(({ gender: genderA }, { gender: genderB }) => {
if (genderA === 'female' && genderB === 'male') return -1;
if (genderA === 'male' && genderB === 'female') return 1;
return 0;
})
.map(actor => actor.name);
const { likes, dislikes } = data.stats;
const duration = data.videos.mediabook.length;
console.log(data);
const rawTags = data.tags.map(tag => tag.name);
const tags = await matchTags(rawTags);
return {
url,
entryId,
title,
description,
actors,
date,
duration,
tags,
rating: {
likes,
dislikes,
},
site,
};
}
async function fetchLatest(site, page = 1) {
const res = await bhttp.get(`https://www.realitykings.com/tour/videos/${site.name.replace(/\s+/g, '-').toLowerCase()}/all-categories/all-time/recent/${page}`);
return scrapeLatest(res.body.toString(), site);
}
async function fetchScene(url, site) {
if (site.isFallback || (site.parameters && site.parameters.altLayout)) {
throw new Error('Cannot fetch scene details from this resource');
}
const entryId = url.split('/').slice(-1)[0];
const cookieJar = new CookieJar();
const session = bhttp.session({ cookieJar });
await session.get(url);
const cookies = await cookieJar.getCookieStringAsync(url);
const instanceToken = cookies.split(';')[0].split('=')[1];
const res = await session.get(`https://site-api.project1service.com/v2/releases/${entryId}`, {
headers: {
Instance: instanceToken,
},
});
return scrapeScene(res.body.result.parent || res.body.result, url, site);
}
module.exports = {
fetchLatest,
fetchScene,
};