Added Bang Bros scraper.

This commit is contained in:
ThePendulum 2019-04-12 03:12:53 +02:00
parent 17480be79a
commit b486b6abdd
8 changed files with 204 additions and 93 deletions

1
.gitignore vendored
View File

@ -2,5 +2,4 @@ node_modules/
dist/ dist/
config/* config/*
!config/default.js !config/default.js
src/utils/link-result.json
db.sqlite db.sqlite

View File

@ -26,7 +26,7 @@ Do not modify `config/default.js`, but instead create a copy at `config/local.js
* `--debug`: Show full error stack trace. * `--debug`: Show full error stack trace.
## Supported networks & sites ## Supported networks & sites
233 sites on 14 networks, continuously expanding! 285 sites on 15 networks, continuously expanding!
* **21Sextury** * **21Sextury**
* Anal Teen Angels * Anal Teen Angels
* Asshole Fever * Asshole Fever
@ -38,6 +38,59 @@ Do not modify `config/default.js`, but instead create a copy at `config/local.js
* Gapeland * Gapeland
* Lez Cuties * Lez Cuties
* Pix and Video * Pix and Video
* **Bang Bros**
* Ass Parade
* AvaSpice
* Back Room Facials
* Backroom MILF
* Ball Honeys
* Bang Bus
* Bang Casting
* Bang POV
* Bang Tryouts
* BangBros 18
* BangBros Angels
* BangBros Remastered
* Bangbros Clips
* Big Mouthfuls
* Big Tit Cream Pie
* Big Tits, Round Asses
* BlowJob Fridays
* Blowjob Ninjas
* Boob Squad
* Brown Bunnies
* Can He Score?
* Casting
* Chongas
* Colombia Fuck Fest
* Dirty World Tour
* Dorm Invasion
* Facial Fest
* Fuck Team Five
* Glory Hole Loads
* Latina Rampage
* Living With Anna
* MILF Lessons
* Magical Feet
* Milf Soup
* MomIsHorny
* Monsters of Cock
* Mr CamelToe
* Mr. Anal
* My Dirty Maid
* My Life In Brazil
* Newbie Black
* Party of 3
* Pawg
* Penny Show
* Porn Star Spa
* Power Munch
* Public Bang
* Slutty White Girls
* Stepmom Videos
* Street Ranger
* Tugjobs
* Working Latinas
* **Blowpass** * **Blowpass**
* 1000 Facials * 1000 Facials
* Immoral Live * Immoral Live

View File

@ -49,11 +49,15 @@ module.exports = {
}, },
{ {
value: 'network', value: 'network',
width: 20, width: 15,
}, },
{ {
value: 'site', value: 'site',
width: 30, width: 25,
},
{
value: 'shootId',
width: 15,
}, },
{ {
value: 'title', value: 'title',

View File

@ -91,7 +91,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Ass Parade', name: 'Ass Parade',
url: 'http://assparade.com', url: 'https://bangbros.com/websites/assparade',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -109,7 +109,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Back Room Facials', name: 'Back Room Facials',
url: 'http://backroomfacials.com', url: 'https://bangbros.com/websites/backroomfacials',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -118,7 +118,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Backroom MILF', name: 'Backroom MILF',
url: 'http://backroommilf.com', url: 'https://bangbros.com/websites/backroommilf',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -136,7 +136,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'BangBros 18', name: 'BangBros 18',
url: 'http://bangbros18.com', url: 'https://bangbros.com/websites/bangbros18',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -154,7 +154,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Bangbros Clips', name: 'Bangbros Clips',
url: 'http://bangbrosclips.com', url: 'https://bangbros.com/websites/bangbrosclips',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -163,7 +163,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'BangBros Remastered', name: 'BangBros Remastered',
url: 'https://bangbros.com/websites/remaster', url: 'https://bangbros.com/websites/bangbrosremastered',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -172,7 +172,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Bang Bus', name: 'Bang Bus',
url: 'http://bangbus.com', url: 'https://bangbros.com/websites/bangbus',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -190,7 +190,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Bang POV', name: 'Bang POV',
url: 'http://bangpov.com', url: 'https://bangbros.com/websites/bangpov',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -208,7 +208,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Big Mouthfuls', name: 'Big Mouthfuls',
url: 'http://bigmouthfuls.com', url: 'https://bangbros.com/websites/bigmouthfuls',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -217,7 +217,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Big Tit Cream Pie', name: 'Big Tit Cream Pie',
url: 'http://bigtitcreampie.com', url: 'https://bangbros.com/websites/bigtitcreampie',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -226,7 +226,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Big Tits, Round Asses', name: 'Big Tits, Round Asses',
url: 'http://bigtitsroundasses.com', url: 'https://bangbros.com/websites/bigtitsroundasses',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -235,7 +235,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'BlowJob Fridays', name: 'BlowJob Fridays',
url: 'http://blowjobfridays.com', url: 'https://bangbros.com/websites/blowjobfridays',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -244,7 +244,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Blowjob Ninjas', name: 'Blowjob Ninjas',
url: 'http://blowjobninjas.com', url: 'https://bangbros.com/websites/blowjobninjas',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -253,7 +253,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Boob Squad', name: 'Boob Squad',
url: 'http://boobsquad.com', url: 'https://bangbros.com/websites/boobsquad',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -262,7 +262,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Brown Bunnies', name: 'Brown Bunnies',
url: 'http://brownbunnies.com', url: 'https://bangbros.com/websites/brownbunnies',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -271,7 +271,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Can He Score?', name: 'Can He Score?',
url: 'http://canhescore.com', url: 'https://bangbros.com/websites/canhescore',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -289,7 +289,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Chongas', name: 'Chongas',
url: 'http://chongas.com', url: 'https://bangbros.com/websites/chongas',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -298,7 +298,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Colombia Fuck Fest', name: 'Colombia Fuck Fest',
url: 'http://colombiafuckfest.com', url: 'https://bangbros.com/websites/colombiafuckfest',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -316,7 +316,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Dorm Invasion', name: 'Dorm Invasion',
url: 'http://dorminvasion.com', url: 'https://bangbros.com/websites/dorminvasion',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -325,7 +325,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Facial Fest', name: 'Facial Fest',
url: 'http://facialfest.com', url: 'https://bangbros.com/websites/facialfest',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -334,7 +334,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Fuck Team Five', name: 'Fuck Team Five',
url: 'http://fuckteamfive.com', url: 'https://bangbros.com/websites/fuckteamfive',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -352,7 +352,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Latina Rampage', name: 'Latina Rampage',
url: 'http://latinarampage.com', url: 'https://bangbros.com/websites/latinarampage',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -361,7 +361,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Living With Anna', name: 'Living With Anna',
url: 'http://livingwithanna.com', url: 'https://bangbros.com/websites/livingwithanna',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -370,7 +370,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Magical Feet', name: 'Magical Feet',
url: 'http://magicalfeet.com', url: 'https://bangbros.com/websites/magicalfeet',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -388,7 +388,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Milf Soup', name: 'Milf Soup',
url: 'http://milfsoup.com', url: 'https://bangbros.com/websites/milfsoup',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -397,7 +397,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'MomIsHorny', name: 'MomIsHorny',
url: 'http://momishorny.com', url: 'https://bangbros.com/websites/momishorny',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -406,7 +406,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Monsters of Cock', name: 'Monsters of Cock',
url: 'http://monstersofcock.com', url: 'https://bangbros.com/websites/monstersofcock',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -415,7 +415,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Mr. Anal', name: 'Mr. Anal',
url: 'http://mranal.com', url: 'https://bangbros.com/websites/mranal',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -433,7 +433,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'My Dirty Maid', name: 'My Dirty Maid',
url: 'http://mydirtymaid.com', url: 'https://bangbros.com/websites/mydirtymaid',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -442,7 +442,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'My Life In Brazil', name: 'My Life In Brazil',
url: 'http://mylifeinbrazil.com', url: 'https://bangbros.com/websites/mylifeinbrazil',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -460,7 +460,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Party of 3', name: 'Party of 3',
url: 'http://partyof3.com', url: 'https://bangbros.com/websites/partyof3',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -469,7 +469,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Pawg', name: 'Pawg',
url: 'http://pawg.com', url: 'https://bangbros.com/websites/pawg',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -487,7 +487,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Porn Star Spa', name: 'Porn Star Spa',
url: 'http://pornstarspa.com', url: 'https://bangbros.com/websites/pornstarspa',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -505,7 +505,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Public Bang', name: 'Public Bang',
url: 'http://publicbang.com', url: 'https://bangbros.com/websites/publicbang',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -523,7 +523,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Stepmom Videos', name: 'Stepmom Videos',
url: 'http://stepmomvideos.com', url: 'https://bangbros.com/websites/stepmomvideos',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -532,7 +532,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Street Ranger', name: 'Street Ranger',
url: 'https://bangbros.com/websites/thewheeler', url: 'https://bangbros.com/websites/streetranger',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -541,7 +541,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Tugjobs', name: 'Tugjobs',
url: 'http://tugjobs.com', url: 'https://bangbros.com/websites/tugjobs',
description: null, description: null,
parameters: null, parameters: null,
}, },
@ -550,7 +550,7 @@ exports.seed = knex => Promise.resolve()
label: null, label: null,
network_id: 'bangbros', network_id: 'bangbros',
name: 'Working Latinas', name: 'Working Latinas',
url: 'http://workinglatinas.com', url: 'https://bangbros.com/websites/workinglatinas',
description: null, description: null,
parameters: null, parameters: null,
}, },

View File

@ -57,6 +57,10 @@ exports.seed = knex => Promise.resolve()
alias_for: null, alias_for: null,
group_id: 'penetration', group_id: 'penetration',
}, },
{
tag: 'amateur',
alias_for: null,
},
{ {
tag: 'American', tag: 'American',
alias_for: null, alias_for: null,

100
src/scrapers/bangbros.js Normal file
View File

@ -0,0 +1,100 @@
'use strict';
/* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
const knex = require('../knex');
const { matchTags } = require('../tags');
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.echThumb').toArray();
return sceneElements.map((element) => {
const sceneLinkElement = $(element).find('.thmb_lnk');
const title = sceneLinkElement.attr('title');
const url = `https://bangbros.com${sceneLinkElement.attr('href')}`;
const shootId = sceneLinkElement.attr('id') && sceneLinkElement.attr('id').split('-')[1];
const entryId = url.split('/')[3].slice(5);
const date = moment.utc($(element).find('.thmb_mr_2 span.faTxt').text(), 'MMM D, YYYY').toDate();
const actors = $(element).find('.cast-wrapper a.cast').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray();
const duration = moment.duration(`0:${$(element).find('.thmb_pic b.tTm').text()}`).asSeconds();
return {
url,
entryId,
shootId,
title,
actors,
date,
duration,
rating: null,
site,
};
});
}
async function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElement = $('.playerSection');
const shootId = sceneElement.find('.vdoCast:contains("Release")').text().replace('Release: ', '');
const entryId = url.split('/')[3].slice(5);
const title = sceneElement.find('.ps-vdoHdd h1').text();
const description = sceneElement.find('.vdoDesc').text().trim();
const [siteName, ...actors] = sceneElement.find('.vdoCast a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const siteId = siteName.replace(/[\s']+/g, '').toLowerCase();
const rawTags = $('.vdoTags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
const [channelSite, tags] = await Promise.all([
knex('sites')
.where({ id: siteId })
.orWhere({ name: siteName })
.first(),
matchTags(rawTags),
]);
const stars = Number(sceneElement.find('.bVdPl_it_like .bVdPl_txt').text().replace('% like', '')) / 20;
return {
url,
shootId,
entryId,
title,
description,
actors,
tags,
rating: {
stars,
},
site: channelSite || site,
};
}
async function fetchLatest(site, page = 1) {
const res = await bhttp.get(`https://bangbros.com/websites/${site.id}/${page}`);
return scrapeLatest(res.body.toString(), site);
}
async function fetchScene(url, site) {
const { origin } = new URL(url);
const res = await bhttp.get(url);
if (origin !== 'https://bangbros.com') {
throw new Error('Cannot fetch from this URL. Please find the scene on https://bangbros.com and try again.');
}
return scrapeScene(res.body.toString(), url, site);
}
module.exports = {
fetchLatest,
fetchScene,
};

View File

@ -1,6 +1,7 @@
'use strict'; 'use strict';
const twentyonesextury = require('./21sextury'); const twentyonesextury = require('./21sextury');
const bangbros = require('./bangbros');
const blowpass = require('./blowpass'); const blowpass = require('./blowpass');
const brazzers = require('./brazzers'); const brazzers = require('./brazzers');
const ddfnetwork = require('./ddfnetwork'); const ddfnetwork = require('./ddfnetwork');
@ -17,6 +18,7 @@ const xempire = require('./xempire');
module.exports = { module.exports = {
'21sextury': twentyonesextury, '21sextury': twentyonesextury,
bangbros,
blowpass, blowpass,
brazzers, brazzers,
ddfnetwork, ddfnetwork,

View File

@ -1,51 +0,0 @@
'use strict';
const Promise = require('bluebird');
const bhttp = require('bhttp');
const fs = Promise.promisifyAll(require('fs'));
const knex = require('../knex');
const argv = require('../argv');
const options = {
responseTimeout: 30000,
};
async function tryLinks() {
const sites = await knex('sites').whereIn('network_id', argv.network);
const results = await Promise.all(sites.map(async (site) => {
console.log(`Trying ${site.name} URLs`);
const [resHttp, resHttpWww, resHttps, resHttpsWww] = await Promise.all([
bhttp.get(`http://${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
bhttp.get(`http://www.${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
bhttp.get(`https://${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
bhttp.get(`https://www.${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
]);
console.log(`Got results for ${site.name}`);
return {
...site,
url: (resHttp.statusCode === 200 && `http://${site.id}.com`)
|| (resHttpWww.statusCode === 200 && `http://www.${site.id}.com`)
|| (resHttps.statusCode === 200 && `https://${site.id}.com`)
|| (resHttpsWww.statusCode === 200 && `https://www.${site.id}.com`)
|| site.url,
network_id: site.network_id,
};
}));
const sortedResults = results.sort((siteA, siteB) => {
if (siteA.id > siteB.id) return 1;
if (siteA.id < siteB.id) return -1;
return 0;
});
console.log(sortedResults);
await fs.writeFileAsync('./src/utils/link-results.json', JSON.stringify(sortedResults, null, 4));
}
tryLinks();