traxxx/src/scrapers/ddfnetwork.js

207 lines
6.6 KiB
JavaScript

'use strict';
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const { JSDOM } = require('jsdom');
const moment = require('moment');
const knex = require('../knex');
/* eslint-disable newline-per-chained-call */
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.card.m-1').toArray();
return sceneElements.map((element) => {
const sceneLinkElement = $(element).find('a').first();
const title = sceneLinkElement.attr('title');
const url = `${site.url}${sceneLinkElement.attr('href')}`;
const entryId = url.split('/').slice(-1)[0];
const date = moment.utc($(element).find('small[datetime]').attr('datetime'), 'YYYY-MM-DD HH:mm:ss').toDate();
const actors = $(element).find('.card-subtitle a').map((actorIndex, actorElement) => $(actorElement).text().trim())
.toArray()
.filter(actor => actor);
const duration = parseInt($(element).find('.card-info div:nth-child(2) .card-text').text(), 10) * 60;
const poster = sceneLinkElement.find('img').attr('data-src');
return {
url,
entryId,
title,
actors,
date,
duration,
poster,
rating: null,
site,
};
});
}
async function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const entryId = url.split('/').slice(-1)[0];
const title = $('meta[itemprop="name"]').attr('content');
const description = $('.descr-box p').text(); // meta tags don't contain full description
const dateProp = $('meta[itemprop="uploadDate"]').attr('content');
const date = dateProp
? moment.utc($('meta[itemprop="uploadDate"]').attr('content'), 'YYYY-MM-DD').toDate()
: moment.utc($('.title-border:nth-child(2) p').text(), 'MM.DD.YYYY').toDate();
const actors = $('.pornstar-card > a').map((actorIndex, actorElement) => $(actorElement).attr('title')).toArray();
const likes = Number($('.info-panel.likes .likes').text());
const duration = Number($('.info-panel.duration .duration').text().slice(0, -4)) * 60;
const tags = $('.tags-tab .tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
const poster = $('#video').attr('poster');
const photos = $('.photo-slider-guest .card a').map((photoIndex, photoElement) => $(photoElement).attr('href')).toArray();
const trailer540 = $('source[res="540"]').attr('src');
const trailer720 = $('source[res="720"]').attr('src');
return {
url,
entryId,
title,
description,
actors,
date,
duration,
tags,
poster,
photos,
trailer: [
{
src: trailer720,
quality: 720,
},
{
src: trailer540,
quality: 540,
},
],
rating: {
likes,
},
site,
};
}
async function scrapeProfile(html, _url, actorName) {
const { document } = new JSDOM(html).window;
const keys = Array.from(document.querySelectorAll('.about-title'), el => el.textContent.trim().replace(':', ''));
const values = Array.from(document.querySelectorAll('.about-info'), (el) => {
if (el.children.length > 0) {
return Array.from(el.children, child => child.textContent.trim()).join(', ');
}
return el.textContent.trim();
});
const bio = keys.reduce((acc, key, index) => {
if (values[index] === '-') {
return acc;
}
return {
...acc,
[key]: values[index],
};
}, {});
const descriptionEl = document.querySelector('.description-box');
const avatarEl = document.querySelector('.pornstar-details .card-img-top');
const country = await knex('countries')
.where('nationality', 'ilike', `%${bio.Nationality}%`)
.orderBy('priority', 'desc')
.first();
const profile = {
name: actorName,
};
profile.birthdate = moment.utc(bio.Birthday, 'MMMM DD, YYYY').toDate();
if (country) profile.birthPlace = country.name;
if (bio['Bra size']) [profile.bust] = bio['Bra size'].match(/\d+\w+/);
if (bio.Waist) profile.waist = Number(bio.Waist.match(/\d+/)[0]);
if (bio.Hips) profile.hip = Number(bio.Hips.match(/\d+/)[0]);
if (bio.Height) profile.height = Number(bio.Height.match(/\d{2,}/)[0]);
if (bio['Tit Style'] && bio['Tit Style'].match('Enhanced')) profile.naturalBoobs = false;
if (bio['Tit Style'] && bio['Tit Style'].match('Natural')) profile.naturalBoobs = true;
if (bio['Body Art'] && bio['Body Art'].match('Tattoo')) profile.hasTattoos = true;
if (bio['Body Art'] && bio['Body Art'].match('Piercing')) profile.hasPiercings = true;
if (bio['Hair Style']) profile.hair = bio['Hair Style'].split(',')[0].trim().toLowerCase();
if (bio['Eye Color']) profile.eyes = bio['Eye Color'].match(/\w+/)[0].toLowerCase();
if (bio['Shoe size']) profile.shoes = Number(bio['Shoe size'].split('|')[1]);
if (descriptionEl) profile.description = descriptionEl.textContent.trim();
if (avatarEl && avatarEl.dataset.src.match('^//')) profile.avatar = `https:${avatarEl.dataset.src}`;
return profile;
}
async function fetchLatest(site, page = 1) {
const res = await bhttp.get(`https://ddfnetwork.com/videos/search/latest/ever/${new URL(site.url).hostname}/-/${page}`);
return scrapeLatest(res.body.toString(), site);
}
async function fetchScene(url, site) {
const res = await bhttp.get(`https://ddfnetwork.com${new URL(url).pathname}`);
return scrapeScene(res.body.toString(), url, site);
}
async function fetchProfile(actorName) {
const resSearch = await bhttp.post('https://ddfnetwork.com/search/ajax',
{
type: 'hints',
word: actorName,
},
{
decodeJSON: true,
headers: {
'x-requested-with': 'XMLHttpRequest',
},
});
if (resSearch.statusCode !== 200 || Array.isArray(resSearch.body.list)) {
return null;
}
if (!resSearch.body.list.pornstarsName || resSearch.body.list.pornstarsName.length === 0) {
return null;
}
const [actor] = resSearch.body.list.pornstarsName;
const url = `https://ddfnetwork.com${actor.href}`;
const resActor = await bhttp.get(url);
if (resActor.statusCode !== 200) {
return null;
}
return scrapeProfile(resActor.body.toString(), url, actorName);
}
module.exports = {
fetchLatest,
fetchProfile,
fetchScene,
};