107 lines
3.3 KiB
JavaScript
107 lines
3.3 KiB
JavaScript
'use strict';
|
|
|
|
const bhttp = require('bhttp');
|
|
const cheerio = require('cheerio');
|
|
const moment = require('moment');
|
|
|
|
const knex = require('../knex');
|
|
const { matchTags } = require('../tags');
|
|
|
|
/* eslint-disable newline-per-chained-call */
|
|
function scrape(html, site) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
const sceneElements = $('.widget-release-card').toArray();
|
|
|
|
return sceneElements.map((element) => {
|
|
const sceneLinkElement = $(element).find('.title a');
|
|
|
|
const title = sceneLinkElement.text().trim();
|
|
const url = `https://www.mofos.com${sceneLinkElement.attr('href')}`;
|
|
const entryId = url.split('/').slice(-2, -1)[0];
|
|
|
|
const date = moment.utc($(element).find('.date-added').text(), 'MMM DD, YYYY').toDate();
|
|
const actors = $(element).find('.girls-name a').map((actorIndex, actorElement) => $(actorElement).attr('title').replace(/\s+/g, ' ')).toArray();
|
|
|
|
const stars = Number($(element).find('.rating').text().slice(0, -1).trim()) / 20;
|
|
|
|
return {
|
|
url,
|
|
entryId,
|
|
title,
|
|
actors,
|
|
date,
|
|
rating: {
|
|
stars,
|
|
},
|
|
site,
|
|
};
|
|
});
|
|
}
|
|
|
|
async function scrapeScene(html, url, site) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
const sceneElement = $('.video-info');
|
|
|
|
const entryId = url.split('/').slice(-2, -1)[0];
|
|
const title = sceneElement.find('.title').text();
|
|
const description = sceneElement.find('.desc').text();
|
|
const actors = sceneElement.find('.girls-site-box a.model-name').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray();
|
|
|
|
const siteElement = sceneElement.find('.site-name');
|
|
const sitename = siteElement.text().trim();
|
|
const siteId = sitename.replace(/\s+/g, '').toLowerCase();
|
|
const siteUrl = siteElement.attr('href').split('/').slice(0, 4).join('/');
|
|
|
|
const stars = Number(sceneElement.find('.rating-box .rating').text().slice(0, -1).trim()) / 20;
|
|
|
|
const rawTags = sceneElement.find('.categories a').map((tagIndex, tagElement) => $(tagElement).text().trim()).toArray();
|
|
|
|
const [channelSite, tags] = await Promise.all([
|
|
knex('sites')
|
|
.where({ slug: siteId })
|
|
.orWhere({ url: `https://www.mofos.com${siteUrl}` })
|
|
.orWhere({ name: sitename })
|
|
.first(),
|
|
matchTags(rawTags),
|
|
]);
|
|
|
|
return {
|
|
url,
|
|
entryId,
|
|
title,
|
|
description,
|
|
actors,
|
|
tags,
|
|
rating: {
|
|
stars,
|
|
},
|
|
site: channelSite || site,
|
|
};
|
|
}
|
|
|
|
async function fetchLatest(site, page = 1) {
|
|
const res = page > 1
|
|
? await bhttp.get(`${site.url}/all-models/all-categories/alltime/bydate/${page}/`)
|
|
: await bhttp.get(`${site.url}/all-models/all-categories/alltime/bydate/`); // explicit page 1 redirects to homepage
|
|
|
|
return scrape(res.body.toString(), site);
|
|
}
|
|
|
|
async function fetchUpcoming(site) {
|
|
const res = await bhttp.get(`${site.url}/all-models/all-categories/upcoming/bydate/`);
|
|
|
|
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,
|
|
};
|