traxxx/src/scrapers/naughtyamerica.js

159 lines
4.7 KiB
JavaScript
Raw Normal View History

2019-04-08 01:07:16 +00:00
'use strict';
/* eslint-disable newline-per-chained-call */
const cheerio = require('cheerio');
const moment = require('moment');
2020-08-26 21:59:29 +00:00
const http = require('../utils/http');
const slugify = require('../utils/slugify');
const qu = require('../utils/q');
2019-04-08 01:07:16 +00:00
function titleExtractor(pathname) {
const components = pathname.split('/')[2].split('-');
const entryId = components.slice(-1)[0];
2019-04-08 01:07:16 +00:00
const title = components.slice(0, -1).reduce((accTitle, word, index) => `${accTitle}${index > 0 ? ' ' : ''}${word.slice(0, 1).toUpperCase()}${word.slice(1)}`, '');
2019-04-08 01:07:16 +00:00
return { title, entryId };
2019-04-08 01:07:16 +00:00
}
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.site-list .scene-item').toArray();
return sceneElements.map((item) => {
const element = $(item);
const sceneLinkElement = element.find('a').first();
const { protocol, hostname, pathname } = new URL(sceneLinkElement.attr('href'));
const url = `${protocol}//${hostname}${pathname}`;
const { title, entryId } = titleExtractor(pathname);
const date = moment.utc(element.find('.entry-date').text(), 'MMM D, YYYY').toDate();
const actors = element.find('.contain-actors a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const duration = Number(element.find('.scene-runtime').text().slice(0, -4)) * 60;
const posterString = sceneLinkElement.find('img[data-srcset]').attr('data-srcset') || sceneLinkElement.find('img[data-src]').attr('data-src');
const poster = `https:${posterString.match(/[\w/.]+$/)[0]}`;
return {
url,
entryId,
title,
actors,
date,
duration,
poster,
rating: null,
site,
};
});
2019-04-08 01:07:16 +00:00
}
function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElement = $('.scene-info');
const { protocol, hostname, pathname } = new URL(url);
const originalUrl = `${protocol}//${hostname}${pathname}`;
const entryId = originalUrl.split('-').slice(-1)[0];
2020-05-26 02:11:29 +00:00
const title = sceneElement.find('h1.scene-title').text();
const description = sceneElement.find('.synopsis').contents().slice(2).text().replace(/[\s\n]+/g, ' ').trim();
2020-05-26 02:11:29 +00:00
const date = moment.utc(sceneElement.find('span.entry-date').text()?.match(/\w+ \d{1,2}, \d{4}/), 'MMM D, YYYY').toDate();
const actors = $('.performer-list a, h1 a.scene-title').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const duration = Number(sceneElement.find('.duration-ratings .duration').text().slice(10, -4)) * 60;
2020-05-26 02:11:29 +00:00
const posterPath = $('video, dl8-video').attr('poster') || $('img.start-card').attr('src');
const poster = posterPath && `https:${posterPath}`;
const photos = $('.contain-scene-images.desktop-only a').map((index, el) => $(el).attr('href')).toArray().filter(Boolean).map(photo => `https:${photo}`);
const trailerEl = $('source');
const trailerSrc = trailerEl.attr('src');
const trailerType = trailerEl.attr('type');
const siteName = sceneElement.find('a.site-title').text();
const channel = siteName.replace(/[\s']+/g, '').toLowerCase();
const tags = $('.categories a.cat-tag').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
return {
url,
entryId,
title,
description,
actors,
date,
duration,
tags,
photos,
poster,
2020-05-26 02:11:29 +00:00
trailer: trailerSrc ? {
src: trailerSrc,
type: trailerType,
2020-05-26 02:11:29 +00:00
} : null,
rating: null,
site,
channel,
};
2019-04-08 01:07:16 +00:00
}
async function fetchActorReleases(url) {
const res = await qu.get(url);
return res.ok
? res.item.query.urls('.contain-block:not(.live-scenes) .scene-item > a:first-child') // live scenes repeat on all pages
: [];
}
async function scrapeProfile(html) {
const { query } = qu.extract(html);
const profile = {};
profile.description = query.q('.bio_about_text', true);
const avatar = query.q('img.performer-pic', 'src');
if (avatar) profile.avatar = `https:${avatar}`;
const releases = query.urls('.scene-item > a:first-child');
const otherPages = query.urls('.pagination a:not([rel=next]):not([rel=prev])');
const olderReleases = await Promise.all(otherPages.map(async page => fetchActorReleases(page)));
profile.releases = releases.concat(olderReleases.flat());
return profile;
}
2019-04-08 01:07:16 +00:00
async function fetchLatest(site, page = 1) {
2020-08-26 21:59:29 +00:00
const res = await http.get(`${site.url}?page=${page}`);
2019-04-08 01:07:16 +00:00
return scrapeLatest(res.body.toString(), site);
2019-04-08 01:07:16 +00:00
}
async function fetchScene(url, site) {
2020-08-26 21:59:29 +00:00
const res = await http.get(url);
2019-04-08 01:07:16 +00:00
return scrapeScene(res.body.toString(), url, site);
2019-04-08 01:07:16 +00:00
}
async function fetchProfile({ name: actorName }) {
const actorSlug = slugify(actorName);
2020-08-26 21:59:29 +00:00
const res = await http.get(`https://www.naughtyamerica.com/pornstar/${actorSlug}`);
if (res.statusCode === 200) {
return scrapeProfile(res.body.toString());
}
return null;
}
2019-04-08 01:07:16 +00:00
module.exports = {
fetchLatest,
fetchScene,
fetchProfile,
2019-04-08 01:07:16 +00:00
};