traxxx/src/scrapers/insex.js

173 lines
5.0 KiB
JavaScript

'use strict';
const qu = require('../utils/qu');
const http = require('../utils/http');
const slugify = require('../utils/slugify');
function scrapeLatest(scenes, site) {
return scenes.map(({ query }) => {
// if (q('.articleTitleText')) return scrapeFirstLatest(ctx(el), site);
const release = {};
const titleEl = query.q('.galleryTitleText, .articleTitleText');
const [title, ...actors] = titleEl.textContent.split('|');
const date = query.date('.articlePostDateText td', 'MMM D, YYYY');
const url = query.url(titleEl, 'a');
[release.entryId] = url.split('/').slice(-2);
release.url = `${site.url}${url}`;
if (date) {
release.title = title.trim();
release.date = date;
} else {
// title should contain date instead, not applicable in brief mode
release.title = title.slice(title.indexOf(':') + 1).trim();
release.date = qu.ed(title.slice(0, title.indexOf(':')), 'MMM D, YYYY');
}
release.actors = actors.map((actor) => actor.trim());
const description = query.q('.articleCopyText', true);
if (description) release.description = description.slice(0, description.lastIndexOf('('));
const duration = query.dur('.articleCopyText a:nth-child(2)');
if (duration) release.duration = duration;
release.likes = parseInt(query.q('.articlePostDateText td:nth-child(3)', true), 10);
const cover = query.img('a img');
release.covers = [[
cover.replace('_thumbnail', ''),
cover,
]];
return release;
});
}
function scrapeLatestAlt(scenes, site) {
return scenes.map(({ query }) => {
const release = {};
release.url = query.url('figure a', 'href', { origin: site.parameters.latest });
release.title = query.cnt('.has-text-weight-bold');
release.date = query.date('span.tag', 'YYYY-MM-DD');
release.actors = query.cnts('a.tag');
const cover = query.img('.image img');
release.poster = cover.replace('poster_noplay', 'trailer_noplay');
release.covers = [cover];
release.entryId = `${qu.formatDate(release.date, 'YYYY-MM-DD')}-${slugify(release.title)}`;
return release;
});
}
function scrapeScene({ query }, site) {
const release = {};
const titleEl = query.q('.articleTitleText');
const [title, ...actors] = titleEl.textContent.split('|');
const url = query.url(titleEl, 'a');
[release.entryId] = url.split('/').slice(-2);
release.url = `${site.url}${url}`;
release.title = title.trim();
release.description = query.q('.articleCopyText', true);
release.actors = actors.map((actor) => actor.trim());
release.date = query.date('.articlePostDateText', 'MMMM D, YYYY');
release.duration = query.dur('.articlePostDateText a:nth-child(2)');
const [cover, ...photos] = query.imgs('img[src*="images"]');
release.covers = [cover];
release.photos = photos;
release.poster = query.poster();
const trailer = query.trailer();
if (trailer) release.trailer = { src: trailer };
return release;
}
async function scrapeSceneAlt({ query }, url, channel, session) {
const release = {};
release.title = query.cnt('.columns div.is-size-5');
release.description = query.cnt('.has-background-black-ter > div:nth-child(4)');
release.date = query.date('.has-text-white-ter span.tag', 'YYYY-MM-DD');
release.actors = query.cnts('.has-text-white-ter a.tag[href*="home.php"]');
release.tags = query.cnts('.has-background-black-ter > div:nth-child(6) > span');
release.poster = query.img('#videoPlayer, #iodvideo', 'poster');
release.photos = query.imgs('body > div:nth-child(6) img');
release.entryId = `${qu.formatDate(release.date, 'YYYY-MM-DD')}-${slugify(release.title)}`;
release.trailer = query.video();
if (!release.trailer) {
const trailerRes = await http.get(`${channel.url}/api/play-api.php`, { session });
if (trailerRes.ok) {
release.trailer = trailerRes.body;
}
}
return release;
}
async function fetchLatest(site, page = 1) {
const url = (site.parameters?.scraper === 'alt' && `${site.parameters.latest}/home.php?o=latest&p=${page}`)
// || (site.slug === 'paintoy' && `${site.url}/corporal/punishment/gallery.php?type=brief&page=${page}`) // paintoy's site is (was?) partially broken, use front page
|| `${site.url}/scripts/switch_tour.php?type=brief&page=${page}`;
const res = await ((site.parameters?.scraper === 'alt' && qu.getAll(url, 'body > .columns .column'))
// || (site.slug === 'paintoy' && qu.getAll(url, '#articleTable table[cellspacing="2"]'))
|| qu.get(url)); // JSON containing html as a property
if (res.ok) {
if (site.parameters?.scraper === 'alt') {
return scrapeLatestAlt(res.items, site);
}
/*
if (site.slug === 'paintoy') {
return scrapeLatest(res.items, site);
}
*/
return scrapeLatest(qu.extractAll(res.body.html, '#articleTable > tbody > tr:nth-child(2) > td > table'), site);
}
return res.status;
}
async function fetchScene(url, site) {
const session = http.session();
const res = await qu.get(url, null, null, { session });
if (res.ok) {
if (site.parameters?.scraper === 'alt') {
return scrapeSceneAlt(res.item, url, site, session);
}
return scrapeScene(res.item, site);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchScene,
};