traxxx/src/scrapers/xempire.js

87 lines
2.3 KiB
JavaScript
Raw Normal View History

'use strict';
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
function scrape(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const scenesElements = $('.sceneInfo').toArray();
return scenesElements.map((element) => {
const sceneLinkElement = $(element).find('.sceneTitle a');
const url = `${site.url}${sceneLinkElement.attr('href')}`;
const title = sceneLinkElement.attr('title');
const date = moment
.utc($(element).find('.sceneDate').text(), 'MM-DD-YYYY')
.toDate();
const actors = $(element).find('.sceneActors a')
.map((actorIndex, actorElement) => $(actorElement).attr('title'))
.toArray();
const [likes, dislikes] = $(element).find('.value')
.toArray()
.map(value => Number($(value).text()));
return {
url,
title,
actors,
date,
rating: {
likes,
dislikes,
},
site,
};
});
}
function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const title = $('h1.title').text();
const date = moment.utc($('.updatedDate').text(), 'MM-DD-YYYY').toDate();
const actors = $('.sceneColActors a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const description = $('.sceneDesc').text().replace(/Video Description:/g, ' ').trim();
const stars = $('.currentRating').text().split('/')[0] / 2;
return {
url,
title,
date,
actors,
description,
rating: {
stars,
},
site,
};
}
async function fetchLatest(site) {
const res = await bhttp.get(`${site.url}/en/videos`);
return scrape(res.body.toString(), site);
}
async function fetchUpcoming(site) {
const res = await bhttp.get(`${site.url}/en/videos/AllCategories/0/1/upcoming`);
return scrape(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,
fetchUpcoming,
fetchScene,
};