Added Fame Digital. Added actor release scraping to DDF Network. Improved q and Gamma scraper.

This commit is contained in:
2020-02-06 23:15:28 +01:00
parent db14eaa5f9
commit 6e1de52a40
42 changed files with 752 additions and 168 deletions

View File

@@ -1,101 +1,81 @@
'use strict';
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const { JSDOM } = require('jsdom');
const moment = require('moment');
const { d, ex, exa, get } = require('../utils/q');
const slugify = require('../utils/slugify');
/* eslint-disable newline-per-chained-call */
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.card.m-1').toArray();
function scrapeAll(html, site, origin) {
return exa(html, '.card.m-1:not(.pornstar-card)').map(({ q, qa, qd }) => {
const release = {};
return sceneElements.map((element) => {
const sceneLinkElement = $(element).find('a').first();
const title = sceneLinkElement.attr('title');
const url = `${site.url}${sceneLinkElement.attr('href')}`;
const entryId = url.split('/').slice(-1)[0];
release.title = q('a', 'title');
release.url = `${site?.url || origin || 'https://ddfnetwork.com'}${q('a', 'href')}`;
[release.entryId] = release.url.split('/').slice(-1);
const date = moment.utc($(element).find('small[datetime]').attr('datetime'), 'YYYY-MM-DD HH:mm:ss').toDate();
const actors = $(element).find('.card-subtitle a').map((actorIndex, actorElement) => $(actorElement).text().trim())
.toArray()
.filter(actor => actor);
release.date = qd('small[datetime]', 'YYYY-MM-DD HH:mm:ss', null, 'datetime');
release.actors = qa('.card-subtitle a', true).filter(Boolean);
const duration = parseInt($(element).find('.card-info div:nth-child(2) .card-text').text(), 10) * 60;
const duration = parseInt(q('.card-info div:nth-child(2) .card-text', true), 10) * 60;
if (duration) release.duration = duration;
const poster = sceneLinkElement.find('img').attr('data-src');
release.poster = q('img').dataset.src;
return {
url,
entryId,
title,
actors,
date,
duration,
poster,
rating: null,
site,
};
return release;
});
}
async function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
async function scrapeScene(html, url, _site) {
const { q, qa, qd, qm, qp, qus } = ex(html);
const release = {};
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
[release.entryId] = url.split('/').slice(-1);
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) => $(actorElement).attr('title')).toArray();
release.title = qm('itemprop=name');
release.description = q('.descr-box p', true);
release.date = qd('meta[itemprop=uploadDate]', 'YYYY-MM-DD', null, 'content')
|| qd('.title-border:nth-child(2) p', 'MM.DD.YYYY');
const likes = Number($('.info-panel.likes .likes').text());
const duration = Number($('.info-panel.duration .duration').text().slice(0, -4)) * 60;
release.actors = qa('.pornstar-card > a', 'title');
release.tags = qa('.tags-tab .tags a', true);
const tags = $('.tags-tab .tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
release.duration = parseInt(q('.icon-video-red + span', true), 10) * 60;
release.likes = Number(q('.icon-like-red + span', true));
const poster = $('#video').attr('poster');
const photos = $('.photo-slider-guest .card a').map((photoIndex, photoElement) => $(photoElement).attr('href')).toArray();
release.poster = qp();
release.photos = qus('.photo-slider-guest .card a');
const trailer540 = $('source[res="540"]').attr('src');
const trailer720 = $('source[res="720"]').attr('src');
release.trailer = qa('source[type="video/mp4"]').map(trailer => ({
src: trailer.src,
quality: Number(trailer.attributes.res.value),
}));
return {
url,
entryId,
title,
description,
actors,
date,
duration,
tags,
poster,
photos,
trailer: [
{
src: trailer720,
quality: 720,
},
{
src: trailer540,
quality: 540,
},
],
rating: {
likes,
},
site,
};
return release;
}
async function fetchActorReleases(urls) {
// DDF Network and DDF Network Stream list all scenes, exclude
const sources = urls.filter(url => !/ddfnetwork/.test(url));
const releases = await Promise.all(sources.map(async (url) => {
const { html } = await get(url);
return scrapeAll(html, null, new URL(url).origin);
}));
// DDF cross-releases scenes between sites, filter duplicates by entryId
return Object.values(releases
.flat()
.sort((releaseA, releaseB) => releaseB.date - releaseA.date) // sort by date so earliest scene remains
.reduce((acc, release) => ({ ...acc, [release.entryId]: release }), {}));
}
async function scrapeProfile(html, _url, actorName) {
const { document } = new JSDOM(html).window;
const { q, qa, qus } = ex(html);
const keys = Array.from(document.querySelectorAll('.about-title'), el => el.textContent.trim().replace(':', ''));
const values = Array.from(document.querySelectorAll('.about-info'), (el) => {
const keys = qa('.about-title', true).map(key => slugify(key, { delimiter: '_' }));
const values = qa('.about-info').map((el) => {
if (el.children.length > 0) {
return Array.from(el.children, child => child.textContent.trim()).join(', ');
}
@@ -104,9 +84,7 @@ async function scrapeProfile(html, _url, actorName) {
});
const bio = keys.reduce((acc, key, index) => {
if (values[index] === '-') {
return acc;
}
if (values[index] === '-') return acc;
return {
...acc,
@@ -114,45 +92,49 @@ async function scrapeProfile(html, _url, actorName) {
};
}, {});
const descriptionEl = document.querySelector('.description-box');
const avatarEl = document.querySelector('.pornstar-details .card-img-top');
const profile = {
name: actorName,
};
profile.birthdate = moment.utc(bio.Birthday, 'MMMM DD, YYYY').toDate();
if (bio.Nationality) profile.nationality = bio.Nationality;
profile.description = q('.description-box', true);
profile.birthdate = d(bio.birthday, 'MMMM DD, YYYY');
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.nationality) profile.nationality = bio.nationality;
if (bio.Height) profile.height = Number(bio.Height.match(/\d{2,}/)[0]);
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['Tit Style'] && bio['Tit Style'].match('Enhanced')) profile.naturalBoobs = false;
if (bio['Tit Style'] && bio['Tit Style'].match('Natural')) profile.naturalBoobs = true;
if (bio.height) profile.height = Number(bio.height.match(/\d{2,}/)[0]);
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.tit_style && /Enhanced/.test(bio.tit_style)) profile.naturalBoobs = false;
if (bio.tit_style && /Natural/.test(bio.tit_style)) profile.naturalBoobs = 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.body_art && /Tattoo/.test(bio.body_art)) profile.hasTattoos = true;
if (bio.body_art && /Piercing/.test(bio.body_art)) profile.hasPiercings = true;
if (bio['Shoe size']) profile.shoes = Number(bio['Shoe size'].split('|')[1]);
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 (descriptionEl) profile.description = descriptionEl.textContent.trim();
if (bio.shoe_size) profile.shoes = Number(bio.shoe_size.split('|')[1]);
const avatarEl = q('.pornstar-details .card-img-top');
if (avatarEl && avatarEl.dataset.src.match('^//')) profile.avatar = `https:${avatarEl.dataset.src}`;
profile.releases = await fetchActorReleases(qus('.find-me-tab li a'));
return profile;
}
async function fetchLatest(site, page = 1) {
const url = `https://ddfnetwork.com/videos/search/latest/ever/${new URL(site.url).hostname}/-/${page}`;
const url = site.parameters?.native
? `${site.url}/videos/search/latest/ever/allsite/-/${page}`
: `https://ddfnetwork.com/videos/search/latest/ever/${new URL(site.url).hostname}/-/${page}`;
console.log(url);
const res = await bhttp.get(url);
return scrapeLatest(res.body.toString(), site);
return scrapeAll(res.body.toString(), site);
}
async function fetchScene(url, site) {