Refactored Vixen scene scraper. Using better poster source for Vixen. Returning video as teaser instead of trailer.

This commit is contained in:
ThePendulum 2020-02-22 05:29:02 +01:00
parent c8c638d201
commit 5e1a1005f1
1 changed files with 48 additions and 52 deletions

View File

@ -5,6 +5,18 @@ const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
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();
}
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
@ -53,7 +65,7 @@ function scrapeUpcoming(html, site) {
const data = JSON.parse(stateString);
const scene = data.page.data['/'].data?.nextScene;
if (!scene) return null;
if (!scene || scene.isPreReleasePeriod) return null;
const release = {};
@ -68,16 +80,7 @@ function scrapeUpcoming(html, site) {
release.actors = scene.models;
release.poster = scene.images.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();
release.poster = getPosterFallbacks(scene.images.poster);
release.teaser = scene.previews.poster
.filter(teaser => /landscape/i.test(teaser.name))
.map(teaser => ({
@ -91,56 +94,49 @@ function scrapeUpcoming(html, site) {
return [release];
}
async function scrapeScene(html, url, site) {
async function scrapeScene(html, url) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const stateObject = $('script:contains("INITIAL_STATE")');
const data = JSON.parse(stateObject.html().trim().slice(27, -1));
const pageData = data.page.data[data.location.pathname].data;
const entryId = pageData.video;
const scene = data.videos.find(video => video.newId === entryId);
const scene = data.videos.find(video => video.newId === pageData.video);
console.log(scene, data, pageData);
const poster = scene.rotatingThumbsUrlSizes[0]['1040w'];
const photos = pageData.pictureset.map(photo => photo.main[0].src);
const trailer = scene.previews.listing.find(preview => preview.height === 353) || null;
const {
title,
description,
models: actors,
totalRateVal: stars,
runLength: duration,
directorNames: director,
tags,
} = scene;
const date = new Date(scene.releaseDate);
return {
const release = {
url,
entryId,
title,
description,
actors,
director,
date,
duration,
tags,
photos,
poster,
trailer: trailer && {
src: trailer.src,
type: trailer.type,
quality: 353,
},
rating: {
stars,
},
site,
title: scene.title,
description: scene.description,
actors: scene.models,
director: scene.directorNames,
duration: scene.runLength,
stars: scene.totalRateVal,
tags: scene.tags,
};
release.entryId = pageData.video;
release.actors = scene.models;
// release.poster = scene.rotatingThumbsUrlSizes[0]['1040w'];
release.poster = getPosterFallbacks(scene.images.poster);
release.photos = pageData.pictureset.map(photo => photo.main[0].src);
const trailer = scene.previews.listing.find(preview => preview.height === 353);
if (trailer) release.trailer = { src: trailer };
// 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;
}
async function fetchLatest(site, page = 1) {