2 Commits

Author SHA1 Message Date
7abe7bb103 Update src/scrapers/newsensations.js 2024-08-07 18:29:44 +00:00
f0c20d879f Fix Newsensations scene description QuerySelector 2024-07-17 19:40:22 +00:00
2 changed files with 16 additions and 5 deletions

View File

@@ -60,7 +60,7 @@ function scrapeSceneBlock({ query }) {
const release = {}; const release = {};
release.title = query.content('.indScene h2, .indScene h1'); release.title = query.content('.indScene h2, .indScene h1');
release.description = query.content('.description'); release.description = query.content('.description h2').replace(" - INTERACTIVE TOY READY!", "");
release.date = query.date('.sceneDateP span', 'MM/DD/YYYY'); release.date = query.date('.sceneDateP span', 'MM/DD/YYYY');
release.actors = query.all('.sceneTextLink .tour_update_models a').map((actorEl) => ({ release.actors = query.all('.sceneTextLink .tour_update_models a').map((actorEl) => ({

View File

@@ -58,13 +58,24 @@ function scrapeScene(html, site, url) {
const { document } = new JSDOM(html).window; const { document } = new JSDOM(html).window;
const release = { site }; const release = { site };
const scene = document.querySelector('#t2019-2col');
release.url = url; release.url = url;
release.title = document.querySelector('.scene-info .text-2xl').textContent.trim(); release.title = scene.querySelector('.t2019-stitle').textContent.trim();
release.description = document.querySelector('.scene-info .space-x-4 span').textContent.trim(); release.description = scene.querySelector('#t2019-description').textContent.trim();
release.actors = Array.from(document.querySelectorAll('.scene-info .link-list-with-commas a'), (el) => el.textContent); release.actors = Array.from(scene.querySelectorAll('#t2019-models a'), (el) => el.textContent);
const durationEls = Array.from(scene.querySelectorAll('#t2019-stime span'));
if (durationEls.length > 1) {
release.date = moment.utc(durationEls[0].textContent, 'MMMM DD, YYYY').toDate();
release.duration = Number(durationEls[1].textContent.match(/\d+/)[0]) * 60;
} else {
release.duration = Number(durationEls[0].textContent.match(/\d+/)[0]) * 60;
}
// unreliable CDN // unreliable CDN
release.photos = Array.from(scene.querySelectorAll('.lg:hidden img'), (el) => ({ release.photos = Array.from(scene.querySelectorAll('#t2019-main .t2019-thumbs img'), (el) => ({
src: (/^http/.test(el.src) ? el.src : `https:${el.src}`), src: (/^http/.test(el.src) ? el.src : `https:${el.src}`),
referer: site.url, referer: site.url,
attempts: 5, attempts: 5,