DebaucheryLibrarian
/
traxxx
2
1
Fork 1
Code Issues 37 Pull Requests 1 Releases Wiki Activity

Added Porn World to replace DDF scraper.

This commit is contained in:
DebaucheryLibrarian 2024-09-01 00:00:40 +02:00
parent f09a94cc6a
commit 668221dafd
5 changed files with 134 additions and 207 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
seeds/02_sites.js
View File

@ -2934,7 +2934,13 @@ const sites = [
description: "Do you love blowjob scenes? Welcome to the place with the best outdoors cock-eating techniques you've ever seen. The randiest females sucking cocks right there.", description: "Do you love blowjob scenes? Welcome to the place with the best outdoors cock-eating techniques you've ever seen. The randiest females sucking cocks right there.",
parent: 'cumlouder', parent: 'cumlouder',
}, },
// DDF NETWORK // PORN WORLD / DDF NETWORK
{
slug: 'pornworld',
name: 'Porn World',
url: 'https://pornworld.com',
parent: 'pornworld',
},
{ {
slug: 'ddfbusty', slug: 'ddfbusty',
name: 'DDF Busty', name: 'DDF Busty',

196
src/scrapers/ddfnetwork.js
View File

@ -1,196 +0,0 @@
'use strict';
const qu = require('../utils/qu');
const slugify = require('../utils/slugify');
const http = require('../utils/http');
function scrapeAll(scenes, site, origin) {
return scenes.map(({ query }) => {
const release = {};
release.title = query.q('.card-title a, .videoContent h4 a', 'title');
release.url = `${site?.url || origin || 'https://pornworld.com'}${query.q('a', 'href')}`;
release.entryId = release.url.match(/\/(\d+)$/)[1];
release.date = query.date('small[datetime]', 'YYYY-MM-DD HH:mm:ss', null, 'datetime');
release.actors = query.all('.card-subtitle a, .featuring a', true).filter(Boolean);
release.description = query.q('h4 + p', true);
const duration = parseInt(query.q('.card-info div:nth-child(2) .card-text', true), 10) * 60;
if (duration) release.duration = duration;
else release.duration = query.dur('.time');
release.poster = query.img();
return release;
});
}
async function scrapeScene({ query }, url, channel) {
const release = {};
release.entryId = url.match(/\/(\d+)$/)[1];
release.title = query.meta('itemprop=name') || query.cnt('.video-title h1, .about-text .story-title, .video-specs h1') || query.cnt('h3');
release.description = query.cnt('.descr-box p') || query.cnt('.about-text p:not(.story-title)') || query.text('.description p');
release.date = query.date('meta[itemprop=uploadDate]', 'YYYY-MM-DD', null, 'content')
|| query.date('.actors time', 'MMMM DD, YYYY')
|| query.date('.title-border:nth-child(2) p', 'MM.DD.YYYY')
|| query.date('.length', 'MMMM DD, YYYY', /\w+ \d{2}, \d{4}/);
if (query.exists('.pornstar-card > a')) release.actors = query.all('.pornstar-card > a', 'title');
else if (query.exists('.actors a')) release.actors = query.cnts('.actors a');
if (query.exists('.tags-tab')) release.tags = query.cnts('.tags-tab .tags a');
else if (query.exists('.tags-box')) release.tags = query.cnts('.tags-box .tags li');
release.duration = parseInt(query.cnt('.icon-video-red + span'), 10) * 60 || query.dur('.length') || null;
release.likes = Number(query.cnt('.icon-like-red + span')) || null;
release.poster = query.poster() || query.poster('dl8-video') || query.img('#videoBlock img');
release.photos = query.urls('.photo-slider-guest .card a');
release.trailer = query.all('source[type="video/mp4"]').map((trailer) => ({
src: trailer.src,
quality: Number(trailer.attributes.res?.value || trailer.attributes.quality?.value.slice(0, -1)) || null,
vr: channel.tags?.some((tag) => tag.slug === 'vr'),
}));
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 res = await qu.getAll(url, '.card.m-1:not(.pornstar-card)');
return res.ok ? scrapeAll(res.items, null, new URL(url).origin) : null;
}));
// 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({ query }, _url, actorName) {
const keys = query.all('.about-title', true).map((key) => slugify(key, '_'));
const values = query.all('.about-info').map((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 profile = {
name: actorName,
};
profile.description = query.q('.description-box', true);
profile.birthdate = qu.extractDate(bio.birthday, 'MMMM DD, YYYY');
if (bio.nationality) profile.nationality = bio.nationality;
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 && /Enhanced/.test(bio.tit_style)) profile.naturalBoobs = false;
if (bio.tit_style && /Natural/.test(bio.tit_style)) profile.naturalBoobs = true;
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.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]);
const avatarEl = query.q('.pornstar-details .card-img-top');
if (avatarEl && avatarEl.dataset.src.match('^//')) profile.avatar = `https:${avatarEl.dataset.src}`;
profile.releases = await fetchActorReleases(query.urls('.find-me-tab li a'));
return profile;
}
async function fetchLatest(channel, page = 1) {
/* ddfnetwork.com redirects to pornworld.com
const url = site.parameters?.native
? `${site.url}/videos/search/latest/ever/allsite/-/${page}`
: `https://ddfnetwork.com/videos/search/latest/ever/${new URL(site.url).hostname}/-/${page}`;
*/
const url = channel.parameters?.latest || `${channel.url}/videos/search/latest/ever/allsite/-/${page}`;
const res = await qu.getAll(url, '.card.m-1:not(.pornstar-card), .allVideos .videoBlock');
if (res.ok) {
return scrapeAll(res.items, channel);
}
return res.status;
}
async function fetchScene(url, site) {
// DDF's main site moved to Porn World
// const res = await http.get(`https://ddfnetwork.com${new URL(url).pathname}`);
const res = await qu.get(url, '.content, #content, .taspVideoPage');
return res.ok ? scrapeScene(res.item, url, site) : res.status;
}
async function fetchProfile({ name: actorName }) {
const resSearch = await http.post('https://ddfnetwork.com/search/ajax',
{
type: 'hints',
word: actorName,
},
{
decodeJSON: false,
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 http.get(url);
if (resActor.statusCode !== 200) {
return null;
}
return scrapeProfile(resActor.body.toString(), url, actorName);
}
module.exports = {
fetchLatest,
fetchProfile,
fetchScene,
};

113
src/scrapers/pornworld.js Executable file
View File

@ -0,0 +1,113 @@
'use strict';
const unprint = require('unprint');
function scrapeAll(scenes) {
return scenes.map(({ query }) => {
const release = {};
release.url = query.url('.card-title a');
release.entryId = new URL(release.url).pathname.match(/\/watch\/([\d-]+)/)[1];
release.title = query.content('.card-title a');
release.date = query.date('.release-date', 'YYYY MMMM, DD', { match: /\d{4} \w+, \d{1,2}/i });
release.duration = query.duration('.video-duration');
release.actors = query.all('.starring a').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}));
const poster = query.img('.thumbnail-pic .card-img');
if (poster) {
const { origin, pathname } = new URL(poster);
release.poster = [
`${origin}${pathname}`,
poster,
];
}
release.trailer = query.video('.thumbnail-pic', { attribute: 'data-video-src' }); // actually the full trailer with audio, not a teaser
return release;
});
}
function scrapeScene({ query }, { url }) {
const release = {};
release.entryId = new URL(url).pathname.match(/\/watch\/([\d-]+)/)[1];
release.title = query.content('.scene__title');
release.description = query.text('//p[span[contains(text(), "Description:")]]');
release.date = query.date('//p[strong[contains(text(), "Publication date:")]]/span', 'YYYY, MMMM D', { match: /\d{4}, \w+ \d{1,2}/i });
release.duration = query.duration('//p[i[contains(@class, "bi-clock-fill")]]');
release.actors = query.all('p a[href*="model/"]').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}));
release.tags = query.contents('p a[href*="?tags"]');
release.poster = query.poster('.video-player');
release.trailer = query.video('.video-player source');
return release;
}
function scrapeProfile({ query }) {
const profile = {};
profile.nationality = query.content('//h3[contains(text(), "Nationality:")]/span') || null;
profile.age = query.number('//h3[contains(text(), "Age:")]/span');
profile.avatar = query.img();
return profile;
}
async function fetchLatest(channel, page = 1) {
const url = `${channel.url}/videos?page=${page}`;
const res = await unprint.get(url, { selectAll: '.card.scene' });
if (res.ok) {
return scrapeAll(res.context, channel);
}
return res.status;
}
async function fetchProfile({ name: actorName }, entity) {
const searchUrl = `${entity.url}/models?name=${actorName}&sort=popularity`;
const searchRes = await unprint.get(searchUrl);
if (searchRes.ok) {
const actorEl = searchRes.context.query.all('.pagination-items .model a').find((resultEl) => unprint.query.attribute(resultEl, null, 'title') === actorName);
const actorUrl = unprint.query.url(actorEl, null);
if (actorUrl) {
const res = await unprint.get(actorUrl, { select: '.model-detail-card' });
if (res.ok) {
return scrapeProfile(res.context, actorName, entity);
}
return res.status;
}
return null;
}
return searchRes.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene,
};

14
src/scrapers/scrapers.js
View File

@ -15,7 +15,6 @@ const cherrypimps = require('./cherrypimps');
const cliffmedia = require('./cliffmedia'); const cliffmedia = require('./cliffmedia');
const cumlouder = require('./cumlouder'); const cumlouder = require('./cumlouder');
const czechav = require('./czechav'); const czechav = require('./czechav');
const ddfnetwork = require('./ddfnetwork');
const modelmedia = require('./modelmedia'); const modelmedia = require('./modelmedia');
const dorcel = require('./dorcel'); const dorcel = require('./dorcel');
const fabulouscash = require('./fabulouscash'); const fabulouscash = require('./fabulouscash');
@ -48,16 +47,17 @@ const missax = require('./missax');
const naughtyamerica = require('./naughtyamerica'); const naughtyamerica = require('./naughtyamerica');
const newsensations = require('./newsensations'); const newsensations = require('./newsensations');
const nubiles = require('./nubiles'); const nubiles = require('./nubiles');
const pascalssubsluts = require('./pascalssubsluts');
const perfectgonzo = require('./perfectgonzo'); const perfectgonzo = require('./perfectgonzo');
const pervcity = require('./pervcity'); const pervcity = require('./pervcity');
const pierrewoodman = require('./pierrewoodman');
const pinkyxxx = require('./pinkyxxx');
const porndoe = require('./porndoe'); const porndoe = require('./porndoe');
const porncz = require('./porncz'); const porncz = require('./porncz');
const pornhub = require('./pornhub'); const pornhub = require('./pornhub');
const pascalssubsluts = require('./pascalssubsluts'); // reserved keyword const pornworld = require('./pornworld');
const pierrewoodman = require('./pierrewoodman');
const pinkyxxx = require('./pinkyxxx');
const privateNetwork = require('./private'); // reserved keyword const privateNetwork = require('./private'); // reserved keyword
const purgatoryx = require('./purgatoryx'); // reserved keyword const purgatoryx = require('./purgatoryx');
const radical = require('./radical'); const radical = require('./radical');
const rickysroom = require('./rickysroom'); const rickysroom = require('./rickysroom');
const score = require('./score'); const score = require('./score');
@ -101,7 +101,7 @@ const scrapers = {
cliffmedia, cliffmedia,
cumlouder, cumlouder,
czechav, czechav,
pornworld: ddfnetwork, pornworld,
delphine: modelmedia, delphine: modelmedia,
dorcel, dorcel,
elegantangel: adultempire, elegantangel: adultempire,
@ -292,7 +292,7 @@ const scrapers = {
pimpxxx: cherrypimps, pimpxxx: cherrypimps,
porncz, porncz,
pornhub, pornhub,
pornworld: ddfnetwork, pornworld,
povperverts: fullpornnetwork, povperverts: fullpornnetwork,
povpornstars: hush, povpornstars: hush,
private: privateNetwork, private: privateNetwork,

10
src/scrapers/template.js
View File

@ -12,13 +12,17 @@ function scrapeAll(scenes) {
release.entryId = new URL(release.url).pathname.match(/\/scene\/(\d+)/)[1]; release.entryId = new URL(release.url).pathname.match(/\/scene\/(\d+)/)[1];
release.title = query.content('.title a'); release.title = query.content('.title a');
release.description = query.content('.description');
release.date = query.date('.date', 'MMM DD, YYYY'); release.date = query.date('.date', 'MMM DD, YYYY');
release.actors = query.contents('.models a.model'); release.duration = query.duration('.duration');
release.actors = query.all('.models a.model').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}));
release.poster = query.img('img.poster'); release.poster = query.img('img.poster');
release.teaser = { src: query.video('.teaser video') }; release.teaser = query.video('.teaser video');
release.stars = query.number('.rating'); release.stars = query.number('.rating');
release.likes = query.number('.likes'); release.likes = query.number('.likes');