2019-03-18 03:46:53 +00:00
|
|
|
'use strict';
|
|
|
|
|
|
|
|
const bhttp = require('bhttp');
|
|
|
|
const cheerio = require('cheerio');
|
|
|
|
const moment = require('moment');
|
|
|
|
|
|
|
|
function scrapeLatest(html, site) {
|
|
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
const scenesElements = $('.update_details').toArray();
|
|
|
|
|
|
|
|
return scenesElements.map((element) => {
|
|
|
|
const sceneLinkElement = $(element).children('a').eq(1);
|
|
|
|
const url = sceneLinkElement.attr('href');
|
|
|
|
const title = sceneLinkElement.text();
|
|
|
|
|
|
|
|
const date = moment
|
|
|
|
.utc($(element).find('.update_date').text(), 'MM/DD/YYYY')
|
|
|
|
.toDate();
|
|
|
|
|
|
|
|
const actors = $(element).find('.update_models a')
|
|
|
|
.map((actorIndex, actorElement) => $(actorElement).text())
|
|
|
|
.toArray();
|
|
|
|
|
|
|
|
return {
|
|
|
|
url,
|
|
|
|
title,
|
|
|
|
actors,
|
|
|
|
date,
|
|
|
|
rating: null,
|
|
|
|
site,
|
|
|
|
};
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
function scrapeUpcoming(html, site) {
|
|
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
const scenesElements = $('#coming_soon_carousel').find('.table').toArray();
|
|
|
|
|
|
|
|
return scenesElements.map((element) => {
|
|
|
|
const details = $(element).find('.update_details_comingsoon')
|
|
|
|
.eq(1)
|
|
|
|
.children()
|
|
|
|
.remove();
|
|
|
|
|
|
|
|
const title = details
|
|
|
|
.end()
|
|
|
|
.text()
|
|
|
|
.trim();
|
|
|
|
|
|
|
|
const actors = details
|
|
|
|
.text()
|
|
|
|
.trim()
|
|
|
|
.split(', ');
|
|
|
|
|
|
|
|
const date = moment
|
|
|
|
.utc($(element).find('.update_date_comingsoon').text().slice(7), 'MM/DD/YYYY')
|
|
|
|
.toDate();
|
|
|
|
|
|
|
|
return {
|
|
|
|
url: null,
|
|
|
|
title,
|
|
|
|
actors,
|
|
|
|
date,
|
|
|
|
rating: null,
|
|
|
|
site,
|
|
|
|
};
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
2019-03-23 21:48:39 +00:00
|
|
|
function scrapeScene(html, url, site) {
|
|
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
|
|
|
|
const title = $('.title_bar_hilite').text();
|
|
|
|
const date = moment
|
|
|
|
.utc($('.update_date').text(), 'MM/DD/YYYY')
|
|
|
|
.toDate();
|
|
|
|
|
|
|
|
const actors = $('.update_description + .update_models a')
|
|
|
|
.map((_actorIndex, actorElement) => $(actorElement).text())
|
|
|
|
.toArray();
|
|
|
|
|
|
|
|
const description = $('.update_description').text().trim();
|
|
|
|
|
|
|
|
const stars = Number($('.avg_rating').text().trim().replace(/[\s|Avg Rating:]/g, ''));
|
|
|
|
|
|
|
|
return {
|
|
|
|
url,
|
|
|
|
title,
|
|
|
|
date,
|
|
|
|
actors,
|
|
|
|
description,
|
|
|
|
rating: {
|
|
|
|
stars,
|
|
|
|
},
|
|
|
|
site,
|
|
|
|
};
|
|
|
|
}
|
|
|
|
|
|
|
|
async function fetchLatest(site) {
|
|
|
|
const res = await bhttp.get(`${site.url}/trial/categories/movies_1_d.html`);
|
|
|
|
|
|
|
|
return scrapeLatest(res.body.toString(), site);
|
|
|
|
}
|
|
|
|
|
|
|
|
async function fetchUpcoming(site) {
|
|
|
|
const res = await bhttp.get(`${site.url}/trial/index.php`);
|
|
|
|
|
|
|
|
return scrapeUpcoming(res.body.toString(), site);
|
|
|
|
}
|
2019-03-18 03:46:53 +00:00
|
|
|
|
2019-03-23 21:48:39 +00:00
|
|
|
async function fetchScene(url, site) {
|
|
|
|
const res = await bhttp.get(url);
|
2019-03-18 03:46:53 +00:00
|
|
|
|
2019-03-23 21:48:39 +00:00
|
|
|
return scrapeScene(res.body.toString(), url, site);
|
2019-03-18 03:46:53 +00:00
|
|
|
}
|
|
|
|
|
2019-03-23 21:48:39 +00:00
|
|
|
module.exports = {
|
|
|
|
fetchLatest,
|
|
|
|
fetchUpcoming,
|
|
|
|
fetchScene,
|
|
|
|
};
|