traxxx/src/scrapers/dogfart.js

136 lines
4.5 KiB
JavaScript

'use strict';
const slugify = require('../utils/slugify');
const qu = require('../utils/qu');
async function getPhotos(albumUrl, channel) {
const res = await qu.get(albumUrl);
if (!res.ok) {
return [];
}
const lastPhotoPage = res.item.query.urls('.pics-container .preview-image-container a').at(-1);
const lastPhotoIndex = parseInt(lastPhotoPage.match(/\d+.jpg/)[0], 10);
const photoUrls = Array.from({ length: lastPhotoIndex }, (value, index) => {
const pageUrl = `${channel.url}${lastPhotoPage.replace(/\d+.jpg/, `${(index + 1).toString().padStart(3, '0')}.jpg`)}`;
const networkPageUrl = `https://dogfartnetwork.com${lastPhotoPage.replace('tourx', 'tour').replace(/\d+.jpg/, `${(index + 1).toString().padStart(3, '0')}.jpg`)}`;
const extract = ({ query }) => query.img('.scenes-module img');
return [
{
url: pageUrl,
extract,
},
{
url: networkPageUrl,
extract,
},
];
});
return photoUrls;
}
function scrapeLatest(scenes, site, filter = true) {
return scenes.reduce((acc, { query }) => {
const release = {};
const siteUrl = query.cnt('.recent-details-title .help-block, .model-details-title .site-name');
release.url = query.url('.thumbnail, .preview-image-container > a', 'href', { origin: site.url });
release.entryId = `${site.slug}_${new URL(release.url).pathname.split('/')[4]}`;
release.title = query.cnt('.scene-title');
// release.actors = release.title.split(/[,&]|\band\b/).map((actor) => actor.replace(/BTS/i, '').trim()); // the titles don't always list the actors, e.g. BarbCummings.com
// release.poster = `https:${element.querySelector('img').src}`;
release.poster = query.img();
release.teaser = query.video('.thumbnail, .preview-thumbnail', 'data-preview_clip_url');
release.channel = siteUrl?.match(/(.*).com/)?.[1].toLowerCase();
if (filter && siteUrl && `www.${siteUrl.toLowerCase()}` !== new URL(site.url).host) {
// different dogfart site
return { ...acc, unextracted: [...acc.unextracted, release] };
}
return { ...acc, scenes: [...acc.scenes, release] };
}, {
scenes: [],
unextracted: [],
});
}
async function scrapeScene({ query }, url, channel, baseScene, parameters) {
const release = {};
const { origin, pathname } = new URL(url);
release.channel = channel.type === 'channel' ? channel.slug : query.cnt('.site-name').split('.')[0].toLowerCase();
release.entryId = `${release.channel}_${pathname.split('/').slice(-2)[0]}`;
release.title = query.cnt('.description-title') || query.text('.scene-title');
release.actors = query.all('.more-scenes a, .starring-list a').map((actorEl) => ({
name: query.cnt(actorEl),
url: query.url(actorEl, null, 'href', { origin: channel.url }),
}));
release.description = query.meta('meta[itemprop="description"]') || query.cnt('.description, [itemprop="description"]')?.replace(/[ \t\n]{2,}/g, ' ').replace('...read more', '').trim();
release.date = query.date('meta[itemprop="uploadDate"]', null, null, 'content');
release.duration = query.duration('.extra-info p:nth-child(2), .run-time-container');
release.tags = query.exists('.scene-details .categories a') ? query.cnts('.scene-details .categories a') : query.text('.categories')?.split(/,\s+/);
const trailer = query.video('.html5-video', 'data-trailer');
const lastPhotosUrl = query.urls('.pagination a').at(-1);
release.poster = query.poster('.html5-video', 'data-poster') || query.img('.trailer-image');
if (trailer && !trailer?.includes('join')) {
release.trailer = trailer;
}
if (lastPhotosUrl && parameters.includePhotos) {
release.photos = await getPhotos(`${origin}${pathname}${lastPhotosUrl}`, channel, url);
}
release.stars = Number(((query.number('span[itemprop="average"], span[itemprop="ratingValue"]') || query.number('canvas[data-score]', null, 'data-score')) / 2).toFixed(2));
return release;
}
async function fetchLatest(channel, page = 1, { parameters }) {
const res = await qu.getAll(parameters.latest ? `${parameters.latest}/?p=${page}` : `${channel.url}/tour/scenes/?p=${page}`, '.recent-updates, .preview-image-container');
if (res.ok) {
return scrapeLatest(res.items, channel);
}
return res.status;
}
async function fetchProfile(baseActor, entity) {
const slug = slugify(baseActor.name, '+');
const url = `https://www.dogfartnetwork.com/tour/girls/${slug}/`;
const res = await qu.getAll(url, '.recent-updates');
if (res.ok) {
const { scenes } = scrapeLatest(res.items, entity, false);
// no bio available
return { scenes };
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene,
};