Added PornPros satellite sites.

This commit is contained in:
2020-01-15 01:01:16 +01:00
parent 6d5dde774f
commit 5dda81535d
49 changed files with 143 additions and 12 deletions

View File

@@ -43,7 +43,7 @@ async function scrapeUniqueReleases(scraper, site, afterDate = getAfterDate(), a
.filter(release => !duplicateReleaseIds.has(String(release.entryId)) // release is already in database
&& moment(release.date).isAfter(afterDate)); // release is older than specified date limit
logger.info(`\x1b[90m${site.name}: Scraped page ${page}, ${uniqueReleases.length} unique recent releases\x1b[0m`);
logger.info(`${site.name}: Scraped page ${page}, ${uniqueReleases.length} unique recent releases`);
if (
uniqueReleases.length > 0

View File

@@ -3,14 +3,18 @@
const bhttp = require('bhttp');
const { JSDOM } = require('jsdom');
const moment = require('moment');
const ex = require('../utils/ex');
function scrapeLatest(html, site) {
const { document } = new JSDOM(html).window;
const s = ex(html);
const { origin } = new URL(site.url);
return Array.from(document.querySelectorAll('.site-list .card'), (scene) => {
const videos = s.qa('.video-releases-list').slice(-1)[0];
return Array.from(videos.querySelectorAll('.card'), (scene) => {
const release = { site };
release.url = `https://pornpros.com${scene.querySelector(':scope > a').href}`;
release.url = `${origin}${scene.querySelector(':scope > a').href}`;
release.entryId = scene.dataset.videoId;
release.title = scene.querySelector('.card-title').textContent;
release.date = moment.utc(scene.dataset.date, 'MMMM DD, YYYY').toDate();
@@ -26,18 +30,25 @@ function scrapeLatest(html, site) {
});
}
function scrapeScene(html, site) {
function scrapeScene(html, site, url) {
const { document } = new JSDOM(html).window;
const release = { site };
const scene = document.querySelector('#t2019-2col');
release.url = url;
release.title = scene.querySelector('.t2019-stitle').textContent.trim();
release.description = scene.querySelector('#t2019-description').textContent.trim();
release.actors = Array.from(scene.querySelectorAll('#t2019-models a'), el => el.textContent);
const durationEl = scene.querySelector('#t2019-stime');
if (durationEl) release.duration = Number(durationEl.textContent.match(/\d+/)[0]) * 60;
const durationEls = Array.from(scene.querySelectorAll('#t2019-stime span'));
if (durationEls.length > 1) {
release.date = moment.utc(durationEls[0].textContent, 'MMMM DD, YYYY').toDate();
release.duration = Number(durationEls[1].textContent.match(/\d+/)[0]) * 60;
} else {
release.duration = Number(durationEls[0].textContent.match(/\d+/)[0]) * 60;
}
release.photos = Array.from(scene.querySelectorAll('#t2019-main .t2019-thumbs img'), el => `https:${el.src}`);
@@ -54,21 +65,21 @@ function scrapeScene(html, site) {
}
async function fetchLatest(site, page = 1) {
const url = `https://pornpros.com/site/${site.slug}?page=${page}`;
const url = `${site.url}?page=${page}`;
const res = await bhttp.get(url);
if (res.statusCode === 200) {
return scrapeLatest(res.body.toString(), site);
}
return null;
return [];
}
async function fetchScene(url, site) {
const res = await bhttp.get(url);
if (res.statusCode === 200) {
return scrapeScene(res.body.toString(), site);
return scrapeScene(res.body.toString(), site, url);
}
return null;

View File

@@ -60,7 +60,7 @@ function destructConfigNetworks(networks) {
}
async function findSiteByUrl(url) {
const { hostname } = new URL(url);
const { hostname, origin } = new URL(url);
const domain = hostname.replace(/www.|tour./, '');
const sites = await knex('sites')
@@ -70,6 +70,7 @@ async function findSiteByUrl(url) {
'networks.name as network_name', 'networks.slug as network_slug', 'networks.url as network_url', 'networks.description as network_description', 'networks.parameters as network_parameters',
)
.where('sites.url', 'like', `${domain}`)
.orWhere('sites.url', 'like', `${origin}`)
.orWhere('sites.url', url);
if (sites.length > 0) {

23
src/utils/ex.js Normal file
View File

@@ -0,0 +1,23 @@
'use strict';
const { JSDOM } = require('jsdom');
function q(context, selector) {
return context.querySelector(selector);
}
function qa(context, selector) {
return Array.from(context.querySelectorAll(selector));
}
function ex(html) {
const { document } = new JSDOM(html).window;
return {
document,
q: selector => q(document, selector),
qa: selector => qa(document, selector),
};
}
module.exports = ex;