traxxx/src/scrapers/mikeadriano.js

260 lines
8.2 KiB
JavaScript

'use strict';
/* eslint-disable newline-per-chained-call */
const { JSDOM } = require('jsdom');
const cheerio = require('cheerio');
const moment = require('moment');
// const bhttp = require('bhttp');
const { get } = require('../utils/http');
const descriptionTags = {
'anal cream pie': 'anal creampie',
'ass to mouth': 'ass to mouth',
'cream pie in her ass': 'anal creampie',
'eats ass': 'ass eating',
facial: 'facial',
gaped: 'gaping',
gapes: 'gaping',
gape: 'gaping',
'rectal cream pie': 'anal creampie',
rimming: 'ass eating',
};
function deriveTagsFromDescription(description) {
const matches = (description || '').toLowerCase().match(new RegExp(Object.keys(descriptionTags).join('|'), 'g'));
return matches
? matches.map(match => descriptionTags[match])
: [];
}
async function scrapeLatestA(html, site) {
const { document } = new JSDOM(html).window;
const sceneElements = document.querySelectorAll('.content-item-large, .content-item');
return Promise.all(Array.from(sceneElements, async (element) => {
const $ = cheerio.load(element.innerHTML, { normalizeWhitespace: true });
const titleElement = element.querySelector('h3.title a');
const title = titleElement.textContent;
const url = titleElement.href;
const entryId = url.split('/').slice(-2)[0];
const descriptionElement = element.querySelector('.desc');
const description = descriptionElement && descriptionElement.textContent.trim();
const date = moment(element.querySelector('.date, time').textContent, 'Do MMM YYYY').toDate();
const actors = Array.from(element.querySelectorAll('h4.models a'), actorElement => actorElement.textContent);
const durationString = element.querySelector('.total-time').textContent.trim();
// timestamp is sometimes 00:00, sometimes 0:00:00
const duration = durationString.split(':').length === 3
? moment.duration(durationString).asSeconds()
: moment.duration(`00:${durationString}`).asSeconds();
const ratingElement = element.querySelector('.rating');
const stars = ratingElement && ratingElement.dataset.rating;
const [poster, ...primaryPhotos] = Array.from(element.querySelectorAll('img'), imageElement => imageElement.src);
const secondaryPhotos = $('.thumb-top, .thumb-bottom')
.map((photoIndex, photoElement) => $(photoElement).css()['background-image'])
.toArray()
.map(photoUrl => photoUrl.slice(photoUrl.indexOf('http'), photoUrl.indexOf('.jpg') + 4));
const photos = [...primaryPhotos, ...secondaryPhotos];
const tags = deriveTagsFromDescription(description);
const scene = {
url,
entryId,
title,
description,
actors,
director: 'Mike Adriano',
date,
duration,
tags,
poster,
photos,
rating: {
stars,
},
site,
};
return scene;
}));
}
async function scrapeLatestB(html) {
const { document } = new JSDOM(html).window;
const sceneElements = document.querySelectorAll('.content-border');
return Promise.all(Array.from(sceneElements, async (element) => {
const $ = cheerio.load(element.innerHTML, { normalizeWhitespace: true });
const release = {
director: 'Mike Adriano',
};
const titleElement = element.querySelector('.content-title-wrap a');
release.title = titleElement.title || titleElement.textContent.trim();
release.url = titleElement.href;
release.entryId = release.url.split('/').slice(-2)[0];
release.description = element.querySelector('.content-description').textContent.trim();
release.date = (moment(element.querySelector('.mobile-date').textContent, 'MM/DD/YYYY')
|| moment(element.querySelector('.date').textContent, 'Do MMM YYYY')).toDate();
release.actors = Array.from(element.querySelectorAll('.content-models a'), actorElement => actorElement.textContent);
const durationString = element.querySelector('.total-time').textContent.trim();
// timestamp is somethines 00:00, sometimes 0:00:00
release.duration = durationString.split(':').length === 3
? moment.duration(durationString).asSeconds()
: moment.duration(`00:${durationString}`).asSeconds();
const [poster, ...primaryPhotos] = Array.from(element.querySelectorAll('a img'), imageElement => imageElement.src);
const secondaryPhotos = $('.thumb-mouseover')
.map((photoIndex, photoElement) => $(photoElement).css()['background-image'])
.toArray()
.map(photoUrl => photoUrl.slice(photoUrl.indexOf('http'), photoUrl.indexOf('.jpg') + 4));
release.poster = poster;
release.photos = [...primaryPhotos, ...secondaryPhotos];
release.tags = deriveTagsFromDescription(release.description);
return release;
}));
}
async function scrapeSceneA(html, url) {
const { document } = new JSDOM(html).window;
const element = document.querySelector('.content-page-info');
const release = {
url,
director: 'Mike Adriano',
};
release.entryId = url.split('/').slice(-2)[0];
release.title = element.querySelector('.title').textContent.trim();
release.description = element.querySelector('.desc').textContent.trim();
release.date = moment(element.querySelector('.post-date').textContent.trim(), 'Do MMM YYYY').toDate();
release.actors = Array.from(element.querySelectorAll('.models a'), actorElement => actorElement.textContent);
const durationString = element.querySelector('.total-time').textContent.trim();
// timestamp is sometimes 00:00, sometimes 0:00:00
release.duration = durationString.split(':').length === 3
? moment.duration(durationString).asSeconds()
: moment.duration(`00:${durationString}`).asSeconds();
const { poster } = document.querySelector('.content-page-header video');
const { src, type } = document.querySelector('.content-page-header source');
release.poster = poster;
release.trailer = { src, type };
release.tags = deriveTagsFromDescription(release.description);
return release;
}
async function scrapeSceneB(html, url, site) {
const { document } = new JSDOM(html).window;
const element = document.querySelector('.content-page-info');
const entryId = url.split('/').slice(-2)[0];
const title = element.querySelector('.title').textContent.trim();
const description = element.querySelector('.desc').textContent.trim();
const date = moment(element.querySelector('.date').textContent.trim(), 'Do MMM YYYY').toDate();
const actors = Array.from(element.querySelectorAll('.models a'), actorElement => actorElement.textContent);
const durationString = element.querySelector('.total-time').textContent.trim();
// timestamp is somethines 00:00, sometimes 0:00:00
const duration = durationString.split(':').length === 3
? moment.duration(durationString).asSeconds()
: moment.duration(`00:${durationString}`).asSeconds();
const { poster } = document.querySelector('.content-page-header-inner video');
const { src, type } = document.querySelector('.content-page-header-inner source');
const tags = deriveTagsFromDescription(description);
const scene = {
url,
entryId,
title,
description,
actors,
director: 'Mike Adriano',
date,
duration,
tags,
poster,
trailer: {
src,
type,
},
site,
};
return scene;
}
async function fetchLatest(site, page = 1) {
const { host } = new URL(site.url);
const url = `https://tour.${host}/videos?page=${page}`;
const res = await get(url);
if (res.code === 200) {
if (host === 'trueanal.com' || host === 'swallowed.com') {
return scrapeLatestA(res.html, site);
}
return scrapeLatestB(res.html, site);
}
return res.code;
}
async function fetchScene(url, site) {
const { host } = new URL(site.url);
const res = await get(url);
if (res.code === 200) {
if (host === 'trueanal.com' || host === 'swallowed.com') {
return scrapeSceneA(res.body.toString(), url, site);
}
return scrapeSceneB(res.body.toString(), url, site);
}
return res.code;
}
/* API protected
async function fetchProfile(actorName, context , site) {
const session = bhttp.session();
await session.get(`https://tour.${site.slug}.com`);
const url = `https://tour.${site.slug}.com/search-preview`;
const res = await session.post(url, { q: actorName }, {
headers: {
'user-agent': 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.163 Safari/537.36',
origin: `https://tour.${site.slug}.com`,
},
});
console.log(res.body.toString());
}
*/
module.exports = {
fetchLatest,
// fetchProfile,
fetchScene,
};