traxxx/src/scrapers/hussiepass.js

166 lines
5.2 KiB
JavaScript
Raw Normal View History

2020-03-05 01:47:52 +00:00
'use strict';
// const util = require('util');
const { get, geta, fd } = require('../utils/q');
const slugify = require('../utils/slugify');
function scrapeLatest(scenes, site) {
return scenes.map(({ q, qu, qd, ql }) => {
const release = {};
release.entryId = q('.modelimg img', 'id').match(/set-target-(\d+)/)[1];
release.title = q('h3 a', 'title') || q('h3 a', true);
release.url = qu('h3 a');
release.date = qd('.modeldata p', 'YYYY-MM-DD', /\d{4}-\d{2}-\d{2}/);
release.duration = ql('.modeldata p');
if (/bts|behind the scenes/i.test(release.title)) release.tags = ['behind-the-scenes'];
release.poster = [
q('.modelimg img', 'src0_3x'),
q('.modelimg img', 'src0_2x'),
q('.modelimg img', 'src0_1x'),
].filter(Boolean).map(src => `${site.url}${src}`);
return release;
});
}
function scrapeLatestT1(scenes, site) {
return scenes.map(({ q, qi, qd, ql, qu }) => {
const release = {};
release.title = q('h4 a', 'title') || q('h4 a', true);
release.url = qu('h4 a');
release.date = qd('.more-info-div', 'MMM D, YYYY');
release.duration = ql('.more-info-div');
release.entryId = q('.img-div img', 'id')?.match(/set-target-(\d+)/)[1]
|| `${slugify(fd(release.date, 'YYYY-MM-DD'))}-${slugify(release.title)}`;
const posterPath = q('.img-div img', 'src0_1x') || qi('img.video_placeholder');
if (posterPath) {
const poster = /^http/.test(posterPath) ? posterPath : `${site.url}${posterPath}`;
release.poster = [
poster.replace('-1x', '-3x'),
poster.replace('-1x', '-2x'),
poster,
];
}
return release;
});
}
function scrapeScene({ html, q, qa, qd, ql }, site, url, baseRelease) {
const release = {};
release.entryId = html.match(/set-target-(\d+)/)[1];
release.title = q('.centerwrap h2', true);
release.description = q('.videocontent p', true);
release.date = qd('.videodetails .date', 'MM/DD/YYYY');
release.duration = ql('.videodetails .date');
release.actors = qa('.modelname a', true);
const posterPath = html.match(/poster="([\w-/.]+)"/)?.[1];
if (posterPath && !/400.jpg/.test(posterPath)) {
const poster = `${site.url}${posterPath}`;
release.poster = [
poster,
// upscaled
poster.replace('-1x', '-2x'),
poster.replace('-1x', '-3x'),
];
if (baseRelease?.poster) release.photos = [baseRelease.poster];
}
const trailerPath = html.match(/\/trailers\/.*.mp4/);
if (trailerPath) release.trailer = { src: `${site.url}${trailerPath}` };
const stars = q('.modelrates + p', true).match(/\d.\d/)?.[0];
if (stars) release.stars = Number(stars);
return release;
}
function scrapeSceneT1({ html, q, qa, qd, ql, qtx }, site, url, baseRelease) {
const release = {};
release.title = q('.trailer-section-head .section-title', true);
release.description = qtx('.row .update-info-block');
release.date = qd('.update-info-row', 'MMM D, YYYY', /\w+ \d{1,2}, \d{4}/);
release.duration = ql('.update-info-row:nth-child(2)');
release.entryId = q('.player-thumb img', 'id').match(/set-target-(\d+)/)[1]
|| `${slugify(fd(release.date, 'YYYY-MM-DD'))}-${slugify(release.title)}`;
release.actors = qa('.models-list-thumbs a').map(el => ({
name: q(el, 'span', true),
avatar: [
q(el, 'img', 'src0_3x'),
q(el, 'img', 'src0_2x'),
q(el, 'img', 'src0_1x'),
].filter(Boolean).map(src => `${site.url}${src}`),
}));
release.tags = qa('.tags a', true);
const posterPath = html.match(/poster="(.*\.jpg)/)?.[1];
if (posterPath && !/400.jpg/.test(posterPath)) {
const poster = `${site.url}${posterPath}`;
release.poster = [
poster,
// upscaled
poster.replace('-1x', '-2x'),
poster.replace('-1x', '-3x'),
];
if (baseRelease?.poster) release.photos = [baseRelease.poster];
}
const trailer = html.match(/<video.*src="(.*\.mp4)/)?.[1];
if (trailer && /^http/.test(trailer)) release.trailer = { src: trailer, referer: url };
else if (trailer) release.trailer = { src: `${site.url}${trailer}`, referer: url };
const stars = q('.update-rating', true).match(/\d.\d/)?.[0];
if (stars) release.stars = Number(stars);
return release;
}
async function fetchLatest(site, page = 1) {
const url = site.parameters?.t1
? `${site.url}/t1/categories/movies_${page}_d.html`
: `${site.url}/categories/movies_${page}_d.html`;
const qLatest = await geta(url, '.modelfeature, .item-video');
if (!qLatest) return null;
return site.parameters?.t1 ? scrapeLatestT1(qLatest, site) : scrapeLatest(qLatest, site);
}
async function fetchScene(url, site, baseRelease) {
const qScene = await get(url);
if (!qScene) return null;
return site.parameters?.t1 ? scrapeSceneT1(qScene, site, url, baseRelease) : scrapeScene(qScene, site, url, baseRelease);
}
module.exports = {
fetchLatest,
fetchScene,
};