2019-04-04 02:00:28 +00:00
|
|
|
'use strict';
|
|
|
|
|
2019-04-07 03:19:44 +00:00
|
|
|
/* eslint-disable newline-per-chained-call */
|
2019-04-04 02:00:28 +00:00
|
|
|
const bhttp = require('bhttp');
|
|
|
|
const cheerio = require('cheerio');
|
|
|
|
const moment = require('moment');
|
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
function getPosterFallbacks(poster) {
|
|
|
|
return poster
|
|
|
|
.filter(image => /landscape/i.test(image.name))
|
|
|
|
.sort((imageA, imageB) => imageB.height - imageA.height)
|
|
|
|
.map((image) => {
|
|
|
|
const sources = [image.src, image.highdpi?.['2x'], image.highdpi?.['3x']];
|
|
|
|
// high DPI images for full HD source are huge, only prefer for smaller fallback sources
|
|
|
|
return image.height === 1080 ? sources : sources.reverse();
|
|
|
|
})
|
|
|
|
.flat();
|
|
|
|
}
|
|
|
|
|
2019-04-04 02:00:28 +00:00
|
|
|
function scrapeLatest(html, site) {
|
|
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
|
2019-09-26 01:27:01 +00:00
|
|
|
const stateScript = $('script:contains("INITIAL_STATE")').html();
|
|
|
|
const { videos: scenes } = JSON.parse(stateScript.slice(stateScript.indexOf('{'), stateScript.indexOf('};') + 1));
|
2019-04-04 02:00:28 +00:00
|
|
|
|
|
|
|
return scenes.map((scene) => {
|
2019-11-16 02:33:36 +00:00
|
|
|
const entryId = String(scene.newId);
|
2019-09-25 02:52:58 +00:00
|
|
|
|
|
|
|
const {
|
|
|
|
title,
|
|
|
|
models: actors,
|
|
|
|
} = scene;
|
|
|
|
|
2019-04-04 02:00:28 +00:00
|
|
|
const url = `${site.url}${scene.targetUrl}`;
|
|
|
|
const date = moment.utc(scene.releaseDateFormatted, 'MMMM DD, YYYY').toDate();
|
|
|
|
const stars = Number(scene.textRating) / 2;
|
|
|
|
|
2019-09-25 02:52:58 +00:00
|
|
|
// largest thumbnail. poster is the same image but bigger, too large for storage space efficiency
|
|
|
|
const poster = scene.images.listing.slice(-1)[0].src;
|
2020-02-22 03:37:48 +00:00
|
|
|
const teaser = scene.previews.listing.slice(-1)[0];
|
2019-09-25 02:52:58 +00:00
|
|
|
|
2019-04-04 02:00:28 +00:00
|
|
|
return {
|
|
|
|
url,
|
2019-11-16 02:33:36 +00:00
|
|
|
entryId,
|
2019-04-04 02:00:28 +00:00
|
|
|
title,
|
|
|
|
actors,
|
|
|
|
date,
|
2019-09-25 02:52:58 +00:00
|
|
|
poster,
|
2020-02-22 03:37:48 +00:00
|
|
|
teaser: {
|
|
|
|
src: teaser.src,
|
|
|
|
type: teaser.type,
|
|
|
|
quality: teaser.height,
|
2019-09-25 02:52:58 +00:00
|
|
|
},
|
2019-04-04 02:00:28 +00:00
|
|
|
rating: {
|
|
|
|
stars,
|
|
|
|
},
|
|
|
|
site,
|
|
|
|
};
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
2020-02-22 03:37:48 +00:00
|
|
|
function scrapeUpcoming(html, site) {
|
|
|
|
const statePrefix = html.indexOf('__INITIAL_STATE__');
|
|
|
|
const stateString = html.slice(html.indexOf('{', statePrefix), html.indexOf('};', statePrefix) + 1);
|
|
|
|
const data = JSON.parse(stateString);
|
|
|
|
|
|
|
|
const scene = data.page.data['/'].data?.nextScene;
|
2020-02-22 04:29:02 +00:00
|
|
|
if (!scene || scene.isPreReleasePeriod) return null;
|
2020-02-22 03:37:48 +00:00
|
|
|
|
|
|
|
const release = {};
|
|
|
|
|
|
|
|
release.title = scene.targetUrl
|
|
|
|
.slice(1)
|
|
|
|
.split('-')
|
|
|
|
.map(component => `${component.charAt(0).toUpperCase()}${component.slice(1)}`)
|
|
|
|
.join(' ');
|
|
|
|
|
|
|
|
release.date = moment.utc(scene.releaseDate).toDate();
|
|
|
|
release.url = `${site.url}${scene.targetUrl}`;
|
|
|
|
|
|
|
|
release.actors = scene.models;
|
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
release.poster = getPosterFallbacks(scene.images.poster);
|
2020-02-22 03:37:48 +00:00
|
|
|
release.teaser = scene.previews.poster
|
|
|
|
.filter(teaser => /landscape/i.test(teaser.name))
|
|
|
|
.map(teaser => ({
|
|
|
|
src: teaser.src,
|
|
|
|
type: teaser.type,
|
|
|
|
quality: Number(String(teaser.height).replace('353', '360')),
|
|
|
|
}));
|
|
|
|
|
|
|
|
release.entryId = (release.poster[0] || release.teaser[0])?.match(/\/(\d+)/)?.[1];
|
|
|
|
|
|
|
|
return [release];
|
|
|
|
}
|
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
async function scrapeScene(html, url) {
|
2019-04-04 02:00:28 +00:00
|
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
|
|
|
|
const stateObject = $('script:contains("INITIAL_STATE")');
|
|
|
|
const data = JSON.parse(stateObject.html().trim().slice(27, -1));
|
|
|
|
|
2020-01-18 22:10:31 +00:00
|
|
|
const pageData = data.page.data[data.location.pathname].data;
|
2020-02-22 04:29:02 +00:00
|
|
|
const scene = data.videos.find(video => video.newId === pageData.video);
|
2019-09-25 00:44:25 +00:00
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
const release = {
|
|
|
|
url,
|
|
|
|
title: scene.title,
|
|
|
|
description: scene.description,
|
|
|
|
actors: scene.models,
|
|
|
|
director: scene.directorNames,
|
|
|
|
duration: scene.runLength,
|
|
|
|
stars: scene.totalRateVal,
|
|
|
|
tags: scene.tags,
|
|
|
|
};
|
2020-02-22 03:37:48 +00:00
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
release.entryId = pageData.video;
|
|
|
|
release.actors = scene.models;
|
2019-09-25 02:52:58 +00:00
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
// release.poster = scene.rotatingThumbsUrlSizes[0]['1040w'];
|
|
|
|
release.poster = getPosterFallbacks(scene.images.poster);
|
|
|
|
release.photos = pageData.pictureset.map(photo => photo.main[0].src);
|
2019-04-04 02:00:28 +00:00
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
const trailer = scene.previews.listing.find(preview => preview.height === 353);
|
|
|
|
if (trailer) release.trailer = { src: trailer };
|
2019-04-04 02:00:28 +00:00
|
|
|
|
2020-02-22 04:29:02 +00:00
|
|
|
// trailer must exist!
|
|
|
|
|
|
|
|
release.teaser = scene.previews.poster
|
|
|
|
.filter(teaser => /landscape/i.test(teaser.name))
|
|
|
|
.map(teaser => ({
|
|
|
|
src: teaser.src,
|
|
|
|
type: teaser.type,
|
|
|
|
quality: Number(String(teaser.height).replace('353', '360')),
|
|
|
|
}));
|
|
|
|
|
|
|
|
release.date = new Date(scene.releaseDate);
|
|
|
|
|
|
|
|
return release;
|
2019-04-04 02:00:28 +00:00
|
|
|
}
|
|
|
|
|
2019-04-05 01:45:40 +00:00
|
|
|
async function fetchLatest(site, page = 1) {
|
2020-01-27 02:07:06 +00:00
|
|
|
const url = `${site.url}/videos?page=${page}&size=7`;
|
|
|
|
const res = await bhttp.get(url);
|
|
|
|
|
|
|
|
if (res.statusCode === 200) {
|
|
|
|
return scrapeLatest(res.body.toString(), site);
|
|
|
|
}
|
2019-04-04 02:00:28 +00:00
|
|
|
|
2020-02-22 03:37:48 +00:00
|
|
|
return res.statusCode;
|
|
|
|
}
|
|
|
|
|
|
|
|
async function fetchUpcoming(site) {
|
|
|
|
const res = await bhttp.get(site.url);
|
|
|
|
|
|
|
|
if (res.statusCode === 200) {
|
|
|
|
return scrapeUpcoming(res.body.toString(), site);
|
|
|
|
}
|
|
|
|
|
|
|
|
return res.statusCode;
|
2019-04-04 02:00:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
async function fetchScene(url, site) {
|
|
|
|
const res = await bhttp.get(url);
|
|
|
|
|
2020-01-27 02:07:06 +00:00
|
|
|
if (res.statusCode === 200) {
|
|
|
|
return scrapeScene(res.body.toString(), url, site);
|
|
|
|
}
|
|
|
|
|
|
|
|
throw new Error(`Vixen response not OK for scene (${url}): ${res.statusCode}`);
|
2019-04-04 02:00:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
module.exports = {
|
|
|
|
fetchLatest,
|
2020-02-22 03:37:48 +00:00
|
|
|
fetchUpcoming,
|
2019-04-04 02:00:28 +00:00
|
|
|
fetchScene,
|
|
|
|
};
|