Added Brazzers scraper. Removed scene-only sites in database in favor of automatic network fallback.

This commit is contained in:
2019-04-04 20:22:47 +02:00
parent 784542253b
commit 6190247ac5
9 changed files with 494 additions and 29 deletions

View File

@@ -9,14 +9,16 @@ const scrapers = require('./scrapers');
async function findSite(url) {
const { origin } = new URL(url);
const site = await knex('sites').where({ url: origin }).first();
const site = await knex('sites').where({ url: origin }).first()
// scenes might be listed on network site, let network scraper find channel site
|| await knex('networks').where({ url: origin }).first();
return {
id: site.id,
name: site.name,
description: site.description,
url: site.url,
networkId: site.network_id,
networkId: site.network_id || site.id,
};
}

118
src/scrapers/brazzers.js Normal file
View File

@@ -0,0 +1,118 @@
'use strict';
/* eslint-disable */
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
const knex = require('../knex');
const { matchTags } = require('../tags');
function scrape(html, site, upcoming) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.release-card.scene').toArray();
return sceneElements.reduce((acc, element) => {
const isUpcoming = $(element).find('.icon-upcoming.active').length === 1;
if ((upcoming && !isUpcoming) || (!upcoming && isUpcoming)) {
return acc;
}
const sceneLinkElement = $(element).find('a');
const url = `https://www.brazzers.com${sceneLinkElement.attr('href')}`;
const title = sceneLinkElement.attr('title');
const shootId = url.split('/').slice(-3, -2)[0];
const date = moment.utc($(element).find('time').text(), 'MMMM DD, YYYY').toDate();
const actors = $(element).find('.model-names a').map((actorIndex, actorElement) => $(actorElement).attr('title')).toArray();
const likes = Number($(element).find('.label-rating .like-amount').text());
const dislikes = Number($(element).find('.label-rating .dislike-amount').text());
return acc.concat({
url,
shootId,
title,
actors,
date,
rating: {
likes,
dislikes,
},
site,
});
}, []);
}
async function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const shootId = url.split('/').slice(-3, -2)[0];
const title = $('.scene-title[itemprop="name"]').text();
const description = $('#scene-description p[itemprop="description"]')
.contents()
.first()
.text()
.trim();
const date = moment.utc($('.more-scene-info .scene-date').text(), 'MMMM DD, YYYY').toDate();
const actors = $('.related-model a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const duration = Number($('.scene-length[itemprop="duration"]').attr('content').slice(1, -1)) * 60;
const likes = Number($('.label-rating .like').text());
const dislikes = Number($('.label-rating .dislike').text());
const siteElement = $('.niche-site-logo');
const siteUrl = `https://www.brazzers.com${siteElement.attr('href').slice(0, -1)}`;
const siteName = siteElement.attr('title');
const rawTags = $('.tag-card-container a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
const [tags, channelSite] = await Promise.all([
matchTags(rawTags),
knex('sites')
.where({ url: siteUrl })
.orWhere({ name: siteName })
.first(),
]);
return {
url,
shootId,
title,
actors,
date,
rating: {
likes,
dislikes,
},
site: channelSite || site,
};
}
async function fetchLatest(site) {
const res = await bhttp.get(`${site.url}/`);
return scrape(res.body.toString(), site, false);
}
async function fetchUpcoming(site) {
const res = await bhttp.get(`${site.url}/`);
return scrape(res.body.toString(), site, true);
}
async function fetchScene(url, site) {
const res = await bhttp.get(url);
return scrapeScene(res.body.toString(), url, site);
}
module.exports = {
fetchLatest,
fetchUpcoming,
fetchScene,
};

View File

@@ -1,5 +1,6 @@
'use strict';
const brazzers = require('./brazzers');
const julesjordan = require('./julesjordan');
const kink = require('./kink');
const legalporno = require('./legalporno');
@@ -9,6 +10,7 @@ const vixen = require('./vixen');
const xempire = require('./xempire');
module.exports = {
brazzers,
julesjordan,
kink,
legalporno,

View File

@@ -57,13 +57,16 @@ async function scrapeScene(html, url, site) {
const siteElement = $('.content-wrapper .logos-sites a');
const siteUrl = siteElement.attr('href').slice(0, -1);
const siteName = siteElement.text();
const channelSite = await knex('sites')
.where({ url: siteUrl })
.orWhere({ name: siteName })
.first();
const rawTags = $('.content-desc .scene-tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
const tags = await matchTags(rawTags);
const [tags, channelSite] = await Promise.all([
matchTags(rawTags),
knex('sites')
.where({ url: siteUrl })
.orWhere({ name: siteName })
.first(),
]);
return {
url,

View File

@@ -9,7 +9,7 @@ const { matchTags } = require('../tags');
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.scenes-latest');
const sceneElements = $('.scenes-latest').toArray();
return sceneElements.map((element) => {
return {
@@ -30,7 +30,7 @@ function scrapeLatest(html, site) {
function scrapeUpcoming(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.scenes-upcoming');
const sceneElements = $('.scenes-upcoming').toArray();
return sceneElements.map((element) => {
return {