Added generic Gamma photo and actor scraper for XEmpire, 21Sextury, Blowpass and Evil Angel.

This commit is contained in:
2020-01-22 22:25:58 +01:00
parent 4e4323704a
commit f8175f6054
17 changed files with 347 additions and 290 deletions

View File

@@ -1,51 +1,10 @@
'use strict';
const Promise = require('bluebird');
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
async function fetchPhotos(photoPath) {
const res = await bhttp.get(`https://21sextury.com${photoPath}`);
return res.body.toString();
}
function scrapePhotos(html) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const unlockedPhotos = $('.preview .imgLink.pgUnlocked')
.map((photoIndex, photoElement) => $(photoElement).attr('href')).toArray();
const lockedThumbnails = $('.preview .imgLink.lockedPicture img')
.map((photoIndex, photoElement) => $(photoElement)
.attr('src'))
// .replace('_tb.jpg', '.jpg')) does not always work
.toArray();
return unlockedPhotos.concat(lockedThumbnails);
}
async function getPhotos(photoPath) {
if (!photoPath || photoPath.match('join')) {
return [];
}
const html = await fetchPhotos(photoPath);
const $ = cheerio.load(html, { normalizeWhitespace: true });
const photos = scrapePhotos(html);
const pages = $('.paginatorPages a').map((pageIndex, pageElement) => $(pageElement).attr('href')).toArray();
const otherPhotos = await Promise.map(pages, async (pagePath) => {
const pageHtml = await fetchPhotos(pagePath);
return scrapePhotos(pageHtml);
}, {
concurrency: 2,
});
return photos.concat(otherPhotos.flat());
}
const { getPhotos, fetchProfile } = require('./gamma');
function scrape(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
@@ -131,8 +90,7 @@ async function scrapeScene(html, url, site) {
const poster = videoData.picPreview;
const trailer = `${videoData.playerOptions.host}${videoData.url}`;
const photoPath = $('.picturesItem a').attr('href');
const photos = await getPhotos(photoPath, site);
const photos = await getPhotos($('.picturesItem a').attr('href'), '21sextury.com', site);
const tags = data.keywords.split(', ');
const siteName = data.productionCompany ? data.productionCompany.name : $('#logoLink a').attr('title');
@@ -181,8 +139,13 @@ async function fetchScene(url, site) {
return scrapeScene(res.body.toString(), url, site);
}
async function networkFetchProfile(actorName) {
return fetchProfile(actorName, '21sextury', true);
}
module.exports = {
fetchLatest,
fetchProfile: networkFetchProfile,
fetchUpcoming,
fetchScene,
};