DebaucheryLibrarian/traxxx
2
1
Fork 1
Code Issues 39 Pull Requests 1 Releases Wiki Activity

Switched to tabs. Adding missing actor entries when scraping actors, with batch ID.

Browse Source
This commit is contained in:
ThePendulum
2020-05-14 04:26:05 +02:00
parent f1eb29c713
commit 11eb66f834
178 changed files with 16594 additions and 16929 deletions
Download Patch File Download Diff File Expand all files Collapse all files

102
src/scrapers/whalemember.js
View File

@@ -5,86 +5,86 @@ const { JSDOM } = require('jsdom');
const moment = require('moment');
function scrapeLatest(html, site) {
const { document } = new JSDOM(html).window;
const { origin } = new URL(site.url);
const { document } = new JSDOM(html).window;
const { origin } = new URL(site.url);
const videos = Array.from(document.querySelectorAll('.video-releases-list')).slice(-1)[0];
const videos = Array.from(document.querySelectorAll('.video-releases-list')).slice(-1)[0];
return Array.from(videos.querySelectorAll('.card'), (scene) => {
const release = { site };
return Array.from(videos.querySelectorAll('.card'), (scene) => {
const release = { site };
release.url = `${origin}${scene.querySelector(':scope > a').href}`;
release.entryId = scene.dataset.videoId;
release.title = scene.querySelector('.card-title').textContent;
release.date = moment.utc(scene.dataset.date, 'MMMM DD, YYYY').toDate();
release.actors = Array.from(scene.querySelectorAll('.actors a'), el => el.textContent);
release.url = `${origin}${scene.querySelector(':scope > a').href}`;
release.entryId = scene.dataset.videoId;
release.title = scene.querySelector('.card-title').textContent;
release.date = moment.utc(scene.dataset.date, 'MMMM DD, YYYY').toDate();
release.actors = Array.from(scene.querySelectorAll('.actors a'), el => el.textContent);
release.poster = `https:${scene.querySelector('.single-image').src}`;
release.photos = Array.from(scene.querySelectorAll('.rollover-thumbs img'), el => `https:${el.dataset.src}`);
release.poster = `https:${scene.querySelector('.single-image').src}`;
release.photos = Array.from(scene.querySelectorAll('.rollover-thumbs img'), el => `https:${el.dataset.src}`);
const trailerEl = scene.querySelector('source');
if (trailerEl) release.trailer = { src: trailerEl.dataset.src };
const trailerEl = scene.querySelector('source');
if (trailerEl) release.trailer = { src: trailerEl.dataset.src };
return release;
});
return release;
});
}
function scrapeScene(html, site, url) {
const { document } = new JSDOM(html).window;
const release = { site };
const { document } = new JSDOM(html).window;
const release = { site };
const scene = document.querySelector('#t2019-2col');
const scene = document.querySelector('#t2019-2col');
release.url = url;
release.title = scene.querySelector('.t2019-stitle').textContent.trim();
release.description = scene.querySelector('#t2019-description').textContent.trim();
release.actors = Array.from(scene.querySelectorAll('#t2019-models a'), el => el.textContent);
release.url = url;
release.title = scene.querySelector('.t2019-stitle').textContent.trim();
release.description = scene.querySelector('#t2019-description').textContent.trim();
release.actors = Array.from(scene.querySelectorAll('#t2019-models a'), el => el.textContent);
const durationEls = Array.from(scene.querySelectorAll('#t2019-stime span'));
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;
}
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;
}
release.photos = Array.from(scene.querySelectorAll('#t2019-main .t2019-thumbs img'), el => `https:${el.src}`);
release.photos = Array.from(scene.querySelectorAll('#t2019-main .t2019-thumbs img'), el => `https:${el.src}`);
const posterEl = scene.querySelector('#no-player-image');
const videoEl = scene.querySelector('video');
const posterEl = scene.querySelector('#no-player-image');
const videoEl = scene.querySelector('video');
if (posterEl) release.poster = `https:${posterEl.src}`;
else if (videoEl) release.poster = `https:${videoEl.poster}`;
if (posterEl) release.poster = `https:${posterEl.src}`;
else if (videoEl) release.poster = `https:${videoEl.poster}`;
const trailerEl = scene.querySelector('#t2019-video source');
if (trailerEl) release.trailer = { src: trailerEl.src };
const trailerEl = scene.querySelector('#t2019-video source');
if (trailerEl) release.trailer = { src: trailerEl.src };
return release;
return release;
}
async function fetchLatest(site, page = 1) {
const url = `${site.url}?page=${page}`;
const res = await bhttp.get(url);
const url = `${site.url}?page=${page}`;
const res = await bhttp.get(url);
if (res.statusCode === 200) {
return scrapeLatest(res.body.toString(), site);
}
if (res.statusCode === 200) {
return scrapeLatest(res.body.toString(), site);
}
return [];
return [];
}
async function fetchScene(url, site) {
const res = await bhttp.get(url);
const res = await bhttp.get(url);
if (res.statusCode === 200) {
return scrapeScene(res.body.toString(), site, url);
}
if (res.statusCode === 200) {
return scrapeScene(res.body.toString(), site, url);
}
return null;
return null;
}
module.exports = {
fetchLatest,
fetchScene,
fetchLatest,
fetchScene,
};