Separated actor expand buttons. Refactored Brazzers scraper. Fixed actor releases not included in shallow scrape. Added number query and data-src default to qu img. Updated README. Removed post-install migrate and seed.

This commit is contained in:
ThePendulum 2020-07-09 02:00:54 +02:00
parent 17d46e804e
commit 44a8ced30c
10 changed files with 289 additions and 326 deletions

View File

@ -7,50 +7,45 @@ Use [nvm](https://github.com/creationix/nvm) to install a recent version of Node
`npm install` `npm install`
### Set up database ### Set up database
Install PostgreSQL, make sure password authentication is enabled (scram-sha-256) and create a database with a privileged user. Install PostgreSQL, make sure password authentication is enabled (scram-sha-256) and create a database with a privileged user. For optimal search engine performance, copy `traxxx.stop` to your PostgresQL text search directory, usually `/usr/share/postgresql/tsearch_data/ or `/usr/local/share/postgresql/tsearch_data/`.
For optimal search engine performance, copy `traxxx.stop` to your PostgresQL text search directory, usually `/usr/share/postgresql/tsearch_data/ or `/usr/local/share/postgresql/tsearch_data/`. ### Configuration
Do not modify `config/default.js`, but instead create a copy at `config/local.js` containing the properties you wish to change. If you have set `NODE_ENV`, copy `assets/js/config/default.js` to `assets/js/config/[environment].js`. After setting up PostgreSQL and configuring the details, run the following commands to create and populate the tables, and build the project:
`npm run migrate` `npm run migrate`
`npm run seed` `npm run seed`
`npm start` You can also use `npm run flush` to run both steps at once, and wipe the database completely later.
### Configuration ### Building
Do not modify `config/default.js`, but instead create a copy at `config/local.js` containing the properties you wish to change. If you have set `NODE_ENV`, copy `assets/js/config/default.js` to `assets/js/config/[environment].js`. To build traxxx, run the following command:
### Options `npm run build`
`npm start -- --option value`
To generate the thumbnails for logos and tag photos, run:
`npm run logos-thumbs`
`npm run tags-thumbs`
### Run
`./traxxx --option value` or `npm start -- --option value`
* `--server`: Run the web server * `--server`: Run the web server
* `--fetch`: Fetch updates instead of running the webserver. Without further arguments, it will use the networks and sites defined in the configuration file. * `--all`: Fetch updates from the channels and networks in the configuration file.
* `--site [site ID]`: Fetch updates from a specific site. The site ID is typically the site name in lowercase and without cases or special characters. For example, Teens Like It Big is teenslikeitbig. * `--channel [slug] [slug]`: Fetch updates from specific channels. The slug is the channel's name in lowercase and without cases or special characters. For example, Teens Like It Big is teenslikeitbig.
* `--network [network ID]`: Fetch updates from all sites of a specific network. The network ID is composed similarly to the site ID. * `--network [slug] [slug]`: Fetch updates from all sites of a specific network. The network slug is composed similarly to the channel slug.
* `--after "[time]"`: Do not fetch scenes older than this. Example values are: `"1 month"`, `"2 weeks"`, `"3 years"`. * `--after "[time]"`: Do not fetch scenes older than this period or date. Example values are: `"1 month"`, `"3 years"`, `"2019-01-01"`.
* `--scene [URL]`: Try to retrieve scene details from its official site or network URL. * `--scene [URL]`: Try to retrieve scene details from its official channel or network URL.
* `--deep`: Follow each release link found running `--site` or `--network` and scrape it for more details. Enabled by default at the moment of writing; use `--no-deep` to only save information found on the overview pages. * `--deep`: Follow each release link found running `--channel` or `--network` and scrape it for more details. Enabled by default ; use `--no-deep` to only save information found on the overview pages.
* `--copy`: Try to copy relevant results to the clipboard. When used with `--scene`, it will copy the filename as defined in the config with all the details filled in.
## Developers #### Developers
### Options
* `--no-save`: Do not store retrieved information in local database, forcing re-fetch. * `--no-save`: Do not store retrieved information in local database, forcing re-fetch.
* `--debug`: Show full error stack trace. * `--level`: Change log level to `silly`, `verbose`, `info`, `warn` or `error`.
### Generating thumbnails
Ensure each tag or sfw category directory has a `thumbs` and `lazy` directory: `for dir in \*; do mkdir "$dir/thumbs $dir/lazy"; done`
Using ImageMagick's bulk tool `mogrify` to generate 240px thumbnails and 90px lazy pre-loading images:
* Generate thumbnails within tag or sfw directory: `mogrify -path lazy -resize x240 -quality 90% \*.jpeg`
* Generate lazy loading images within tag or sfw directory: `mogrify -path lazy -resize x90 -quality 90% \*.jpeg`
* Generate thumbnails for all tags or categories in `tags` or `sfw` directory: `for dir in \*; do mogrify -path "$dir/thumbs" -resize x240 -quality 90% "$dir/\*.jpeg"; done`
* Generate lazy loading images for all tags categories in `tags` or `sfw` directory: `for dir in \*; do mogrify -path "$dir/lazy" -resize x90 -quality 90% "$dir/\*.jpeg"; done`
## Supported networks & sites ## Supported networks & sites
768 sites on 62 networks, continuously expanding! 896 channels on 64 networks, continuously expanding!
* 21Naturals * 21Naturals
* 21Sextreme * 21Sextreme
@ -60,7 +55,6 @@ Using ImageMagick's bulk tool `mogrify` to generate 240px thumbnails and 90px la
* Assylum * Assylum
* Aziani * Aziani
* Babes * Babes
* BAM Visions
* Bang! * Bang!
* Bang Bros * Bang Bros
* Blowpass * Blowpass
@ -75,24 +69,20 @@ Using ImageMagick's bulk tool `mogrify` to generate 240px thumbnails and 90px la
* Fame Digital * Fame Digital
* Fantasy Massage * Fantasy Massage
* Full Porn Network * Full Porn Network
* Gamma Entertainment
* Girlsway * Girlsway
* Hush
* Hush Pass * Hush Pass
* Hussie Pass * Hussie Pass
* Insex * Insex
* Interracial Pass * Interracial Pass
* JayRock Productions * JayRock Productions
* Jesse Loads Monster Facials
* Jules Jordan * Jules Jordan
* Kelly Madison Media * Kelly Madison Media (Teen Fidelity)
* Kink * Kink
* LegalPorno * LegalPorno
* Men * Men
* Metro HD * Metro HD
* Mike Adriano * Mike Adriano
* Mile High Media * Mile High Media
* Mind Geek
* MOFOS * MOFOS
* Naughty America * Naughty America
* New Sensations * New Sensations
@ -107,10 +97,11 @@ Using ImageMagick's bulk tool `mogrify` to generate 240px thumbnails and 90px la
* SCORE * SCORE
* Sexy Hub * Sexy Hub
* Team Skeet * Team Skeet
* Teen Core Club
* Twistys * Twistys
* Vivid * Vivid
* Vixen * Vixen
* VogoV * VogoV
* Whale Member * Whale Member (Holed, POVD)
* Wicked * Wicked
* XEmpire * XEmpire

View File

@ -32,7 +32,7 @@
<div class="actor-inner"> <div class="actor-inner">
<div <div
class="profile" class="profile"
:class="{ expanded, 'with-avatar': !!actor.avatar }" :class="{ expanded: bioExpanded, 'with-avatar': !!actor.avatar }"
> >
<a <a
v-if="actor.avatar" v-if="actor.avatar"
@ -48,11 +48,12 @@
> >
</a> </a>
<span <Expand
v-show="expanded" v-if="bioExpanded"
class="expand collapse-header noselect" :expanded="bioExpanded"
@click="expanded = false" class="expand expand-dark"
><Icon icon="arrow-up3" /></span> @expand="(state) => bioExpanded = state"
/>
<ul class="bio nolist"> <ul class="bio nolist">
<li <li
@ -96,10 +97,10 @@
<span> <span>
<span <span
v-if="actor.origin.city" v-if="actor.origin.city"
class="city hideable" class="city"
>{{ actor.origin.city }}</span><span >{{ actor.origin.city }}</span><span
v-if="actor.origin.state && (!actor.origin.city || (actor.origin.country && actor.origin.country.alpha2 === 'US'))" v-if="actor.origin.state && (!actor.origin.city || (actor.origin.country && actor.origin.country.alpha2 === 'US'))"
class="state hideable" class="state"
>{{ actor.origin.city ? `, ${actor.origin.state}` : actor.origin.state }}</span> >{{ actor.origin.city ? `, ${actor.origin.state}` : actor.origin.state }}</span>
<span <span
@ -116,17 +117,17 @@
<li <li
v-if="actor.residence" v-if="actor.residence"
class="bio-item residence" class="bio-item residence hideable"
> >
<dfn class="bio-label"><Icon icon="location" />Lives in</dfn> <dfn class="bio-label"><Icon icon="location" />Lives in</dfn>
<span> <span>
<span <span
v-if="actor.residence.city" v-if="actor.residence.city"
class="city hideable" class="city"
>{{ actor.residence.city }}</span><span >{{ actor.residence.city }}</span><span
v-if="actor.residence.state && actor.residence.country && actor.residence.country.alpha2 === 'US'" v-if="actor.residence.state && actor.residence.country && actor.residence.country.alpha2 === 'US'"
class="state hideable" class="state"
>{{ actor.residence.city ? `, ${actor.residence.state}` : actor.residence.state }}</span> >{{ actor.residence.city ? `, ${actor.residence.state}` : actor.residence.state }}</span>
<span <span
@ -235,12 +236,6 @@
<li class="bio-item scraped hideable">Updated {{ formatDate(actor.updatedAt, 'YYYY-MM-DD HH:mm') }}, ID: {{ actor.id }}</li> <li class="bio-item scraped hideable">Updated {{ formatDate(actor.updatedAt, 'YYYY-MM-DD HH:mm') }}, ID: {{ actor.id }}</li>
</ul> </ul>
<span
v-show="!expanded"
class="expand expand-header collapse-header noselect"
@click="expanded = true"
><Icon icon="arrow-down3" /></span>
<div class="descriptions-container"> <div class="descriptions-container">
<div <div
v-if="actor.descriptions && actor.descriptions.length > 0" v-if="actor.descriptions && actor.descriptions.length > 0"
@ -275,23 +270,23 @@
class="profile-social" class="profile-social"
/> />
<span <Expand
v-show="expanded" :expanded="bioExpanded"
class="expand expand-header collapse-header noselect" class="expand expand-dark"
@click="expanded = false" @expand="(state) => bioExpanded = state"
><Icon icon="arrow-up3" /></span> />
</div> </div>
<div class="actor-content"> <div class="actor-content">
<Scroll <Scroll
v-if="actor.avatar || (actor.photos && actor.photos.length > 0)" v-if="actor.avatar || (actor.photos && actor.photos.length > 0)"
:expanded="expanded" :expanded="photosExpanded"
class="scroll-light" class="scroll-light"
@expand="(state) => expanded = state" @expand="(state) => photosExpanded = state"
> >
<Photos <Photos
:actor="actor" :actor="actor"
:class="{ expanded }" :class="{ expanded: photosExpanded }"
/> />
</Scroll> </Scroll>
@ -319,6 +314,7 @@ import Pagination from '../pagination/pagination.vue';
import FilterBar from '../header/filter-bar.vue'; import FilterBar from '../header/filter-bar.vue';
import Releases from '../releases/releases.vue'; import Releases from '../releases/releases.vue';
import Photos from './photos.vue'; import Photos from './photos.vue';
import Expand from '../expand/expand.vue';
import Scroll from '../scroll/scroll.vue'; import Scroll from '../scroll/scroll.vue';
import Gender from './gender.vue'; import Gender from './gender.vue';
import Social from './social.vue'; import Social from './social.vue';
@ -358,6 +354,7 @@ export default {
Pagination, Pagination,
Photos, Photos,
Scroll, Scroll,
Expand,
Releases, Releases,
Gender, Gender,
Social, Social,
@ -369,7 +366,8 @@ export default {
totalCount: 0, totalCount: 0,
limit: 10, limit: 10,
pageTitle: null, pageTitle: null,
expanded: false, bioExpanded: false,
photosExpanded: false,
}; };
}, },
computed: { computed: {
@ -641,56 +639,6 @@ export default {
.expand { .expand {
display: none; display: none;
justify-content: center;
align-items: center;
padding: .5rem .25rem;
font-weight: bold;
font-size: .9rem;
cursor: pointer;
.icon {
fill: var(--shadow);
}
&:hover {
background: var(--shadow-hint);
.icon {
fill: var(--shadow-strong);
}
}
}
.expand-sidebar:hover {
background: var(--shadow-hint);
}
.expand-header {
display: none;
&:hover {
background: var(--shadow-hint);
}
}
.collapse-header {
display: none;
width: 100%;
justify-content: center;
align-items: center;
padding: 0;
background: var(--profile);
.icon {
width: 100%;
fill: var(--highlight);
padding: .5rem 0;
}
&:hover .icon {
background: var(--highlight-hint);
fill: var(--text-contrast);
}
} }
.scroll { .scroll {
@ -718,7 +666,6 @@ export default {
height: auto; height: auto;
max-height: none; max-height: none;
flex-direction: column; flex-direction: column;
padding: 0 0 .5rem 0;
&.with-avatar { &.with-avatar {
height: auto; height: auto;
@ -742,29 +689,12 @@ export default {
margin: 0; margin: 0;
} }
.expand, .expanded .bio-value {
.expand-header {
display: flex;
}
/*
.expanded .descriptions-container {
display: block;
max-width: 100%;
max-height: 30rem;
margin: 0;
padding: 0 1rem;
}
*/
.expanded {
.collapse-header {
display: block;
}
.bio-value {
white-space: normal; white-space: normal;
} }
.expand {
display: block;
} }
} }

View File

@ -12,8 +12,13 @@ module.exports = {
sfwPort: 5001, sfwPort: 5001,
}, },
// include: [], // include: [],
// exclude: [], exclude: {
exclude: [ networks: [
'gamma',
'mindgeek',
'julesjordan',
],
channels: [
// 21sextreme, no longer updated // 21sextreme, no longer updated
'mightymistress', 'mightymistress',
'dominatedgirls', 'dominatedgirls',
@ -64,6 +69,7 @@ module.exports = {
// mindgeek // mindgeek
'pornhub', 'pornhub',
], ],
},
profiles: [ profiles: [
[ [
'evilangel', 'evilangel',

View File

@ -4,7 +4,6 @@
"description": "All the latest porn releases in one place", "description": "All the latest porn releases in one place",
"main": "src/app.js", "main": "src/app.js",
"scripts": { "scripts": {
"postinstall": "npm run migrate && npm run seed",
"start": "node -r source-map-support/register dist/init.js", "start": "node -r source-map-support/register dist/init.js",
"webpack": "webpack --env=production --mode=production", "webpack": "webpack --env=production --mode=production",
"webpack-dev": "webpack --env=development --mode=development", "webpack-dev": "webpack --env=development --mode=development",

View File

@ -4825,7 +4825,7 @@ const sites = [
name: 'Lil Humpers', name: 'Lil Humpers',
url: 'https://lilhumpers.com', url: 'https://lilhumpers.com',
description: '', description: '',
parameters: { siteId: 310 }, parameters: { siteId: 336 },
slug: 'lilhumpers', slug: 'lilhumpers',
parent: 'realitykings', parent: 'realitykings',
}, },

View File

@ -281,7 +281,7 @@ async function curateProfile(profile) {
curatedProfile.nationality = profile.nationality?.trim() || null; // used to derive country when country not available curatedProfile.nationality = profile.nationality?.trim() || null; // used to derive country when country not available
curatedProfile.ethnicity = ethnicities[profile.ethnicity?.trim().toLowerCase()] || null; curatedProfile.ethnicity = ethnicities[profile.ethnicity?.trim().toLowerCase()] || null;
curatedProfile.hairColor = hairColors[(profile.hairColor || profile.hair)?.trim().toLowerCase()] || null; curatedProfile.hairColor = hairColors[(profile.hairColor || profile.hair)?.toLowerCase().replace('hair', '').trim()] || null;
curatedProfile.eyes = eyeColors[profile.eyes?.trim().toLowerCase()] || null; curatedProfile.eyes = eyeColors[profile.eyes?.trim().toLowerCase()] || null;
curatedProfile.tattoos = profile.tattoos?.trim() || null; curatedProfile.tattoos = profile.tattoos?.trim() || null;

View File

@ -24,11 +24,11 @@ async function init() {
const actors = argv.actors && await scrapeActors(argv.actors); const actors = argv.actors && await scrapeActors(argv.actors);
const actorBaseScenes = argv.actors && argv.actorScenes && actors.map(actor => actor.releases).flat().filter(Boolean); const actorBaseScenes = argv.actors && argv.actorScenes && actors.map(actor => actor.releases).flat().filter(Boolean);
const updateBaseScenes = (argv.scrape || argv.channels || argv.networks) && await fetchUpdates(); const updateBaseScenes = (argv.all || argv.channels || argv.networks) && await fetchUpdates();
const deepScenes = argv.deep const deepScenes = argv.deep
? await fetchScenes([...(argv.scenes || []), ...(updateBaseScenes || []), ...(actorBaseScenes || [])]) ? await fetchScenes([...(argv.scenes || []), ...(updateBaseScenes || []), ...(actorBaseScenes || [])])
: updateBaseScenes; : [...(updateBaseScenes || []), ...(actorBaseScenes || [])];
const sceneMovies = deepScenes && argv.sceneMovies && deepScenes.map(scene => scene.movie).filter(Boolean); const sceneMovies = deepScenes && argv.sceneMovies && deepScenes.map(scene => scene.movie).filter(Boolean);
const deepMovies = await fetchMovies([...(argv.movies || []), ...(sceneMovies || [])]); const deepMovies = await fetchMovies([...(argv.movies || []), ...(sceneMovies || [])]);

View File

@ -85,7 +85,54 @@ async function fetchChannelsFromArgv() {
} }
async function fetchChannelsFromConfig() { async function fetchChannelsFromConfig() {
const rawSites = await knex('entities') const rawNetworks = await knex.raw(`
WITH RECURSIVE children AS (
SELECT
id, parent_id, name, slug, type, url, description, parameters
FROM
entities
WHERE
CASE WHEN array_length(?, 1) IS NOT NULL
THEN slug = ANY(?)
ELSE true
END
AND NOT
slug = ANY(?)
AND
entities.type = 'network'
UNION ALL
SELECT
entities.id, entities.parent_id, entities.name, entities.slug, entities.type, entities.url, entities.description, entities.parameters
FROM
entities
INNER JOIN
children ON children.id = entities.parent_id
)
SELECT
entities.*, row_to_json(parents) as parent, json_agg(children) as children
FROM
children
LEFT JOIN
entities ON entities.id = children.parent_id
LEFT JOIN
entities AS parents ON parents.id = entities.parent_id
WHERE
children.type = 'channel'
GROUP BY
children.parent_id, entities.id, entities.name, parents.id
`, [
config.include.networks,
config.include.networks,
config.exclude.networks,
]);
console.log(rawNetworks.rows);
/*
const curatedSites = await curateEntities(rawChannels, true);
logger.info(`Found ${curatedSites.length} entities in database`);
const rawChannels = await knex('entities')
.select(knex.raw('entities.*, row_to_json(parents) as parent')) .select(knex.raw('entities.*, row_to_json(parents) as parent'))
.leftJoin('entities as parents', 'parents.id', 'entities.parent_id') .leftJoin('entities as parents', 'parents.id', 'entities.parent_id')
.where((builder) => { .where((builder) => {
@ -97,10 +144,10 @@ async function fetchChannelsFromConfig() {
builder.whereIn('entities.slug', config.exclude || []); builder.whereIn('entities.slug', config.exclude || []);
}); });
const curatedSites = await curateEntities(rawSites, true); console.log(rawChannels);
logger.info(`Found ${curatedSites.length} entities in database`); */
return curatedSites; // return curatedSites;
} }
async function fetchIncludedEntities() { async function fetchIncludedEntities() {

View File

@ -1,142 +1,111 @@
'use strict'; 'use strict';
/* eslint-disable newline-per-chained-call */ /* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp'); const qu = require('../utils/qu');
const cheerio = require('cheerio');
const { JSDOM } = require('jsdom');
const moment = require('moment');
const { get, ex } = require('../utils/q');
const slugify = require('../utils/slugify'); const slugify = require('../utils/slugify');
const { heightToCm, lbsToKg } = require('../utils/convert'); const { heightToCm, lbsToKg } = require('../utils/convert');
const hairMap = { function scrapeAll(items, channel, upcoming) {
Blonde: 'blonde', return items.reduce((acc, { query }) => {
Brunette: 'brown', const isUpcoming = query.exists('.icon-upcoming.active');
'Black Hair': 'black',
Redhead: 'red',
};
function scrapeAll(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)) { if ((upcoming && !isUpcoming) || (!upcoming && isUpcoming)) {
return acc; return acc;
} }
const sceneLinkElement = $(element).find('a'); const release = {};
const pathname = query.url('a');
const url = `https://www.brazzers.com${sceneLinkElement.attr('href')}`; release.url = `https://www.brazzers.com${pathname}`;
const title = sceneLinkElement.attr('title'); release.entryId = pathname.match(/(\/view\/id\/|\/episode\/)(\d+)/)[2];
const entryId = url.split('/').slice(-3, -2)[0];
const date = moment.utc($(element).find('time').text(), 'MMMM DD, YYYY').toDate(); release.title = query.q('a', 'title');
const actors = $(element).find('.model-names a').map((actorIndex, actorElement) => $(actorElement).attr('title')).toArray(); release.date = query.date('time', 'MMMM DD, YYYY');
const likes = Number($(element).find('.label-rating .like-amount').text()); release.actors = query.all('.model-names a', 'title');
const dislikes = Number($(element).find('.label-rating .dislike-amount').text());
const poster = `https:${$(element).find('.card-main-img').attr('data-src')}`; release.likes = query.number('.label-rating .like-amount');
const photos = $(element).find('.card-overlay .image-under').map((photoIndex, photoElement) => `https:${$(photoElement).attr('data-src')}`).toArray(); release.dislikes = query.number('.label-rating .dislike-amount');
const channel = slugify($(element).find('.collection').attr('title'), ''); release.poster = query.img('.card-main-img');
release.photos = query.imgs('.card-overlay .image-under');
return acc.concat({ release.channel = slugify(query.q('.collection', 'title'), '');
url,
entryId, return acc.concat(release);
title,
actors,
date,
poster,
photos,
rating: {
likes,
dislikes,
},
channel,
site,
});
}, []); }, []);
} }
async function scrapeScene(html, url, _site) { function getVideoData(html) {
const $ = cheerio.load(html, { normalizeWhitespace: true }); try {
const videoScriptStart = html.indexOf('window.videoUiOptions');
const videoScript = html.slice(videoScriptStart, html.indexOf('};', videoScriptStart));
const videoString = videoScript.slice(videoScript.indexOf('{"stream_info"'), videoScript.lastIndexOf('},') + 1);
return JSON.parse(videoString);
} catch (error) {
return null;
}
}
async function scrapeScene({ query, html }, url, _site) {
const release = {}; const release = {};
const videoJson = $('script:contains("window.videoUiOptions")').html(); release.entryId = new URL(url).pathname.match(/(\/view\/id\/|\/episode\/)(\d+)/)[2];
const videoString = videoJson.slice(videoJson.indexOf('{"stream_info":'), videoJson.lastIndexOf('},') + 1);
const videoData = JSON.parse(videoString);
[release.entryId] = url.split('/').slice(-3, -2); release.title = query.q('.scene-title[itemprop="name"]', true);
release.title = $('.scene-title[itemprop="name"]').text(); release.description = query.text('#scene-description p[itemprop="description"]');
release.description = $('#scene-description p[itemprop="description"]') release.date = query.date('.more-scene-info .scene-date', 'MMMM DD, YYYY');
.contents() release.duration = Number(query.q('#trailer-player-container', 'data-duration')) // more accurate
.first() || Number(query.q('.scene-length[itemprop="duration"]', 'content').slice(1, -1) * 60);
.text()
.trim();
release.date = moment.utc($('.more-scene-info .scene-date').text(), 'MMMM DD, YYYY').toDate(); // actor cards have avatar, but truncated name
release.duration = Number($('.scene-length[itemprop="duration"]').attr('content').slice(1, -1)) * 60; const actorImagesByActorId = query.imgs('.featured-model .card-image img').reduce((acc, img) => ({
...acc,
[img.match(/\/models\/(\d+)/)[1]]: [
img.replace('medium', 'large'),
img,
],
}), {});
const actorsFromCards = $('.featured-model .card-image a').map((actorIndex, actorElement) => { release.actors = query.all('.related-model a').map(actorEl => ({
const avatar = `https:${$(actorElement).find('img').attr('data-src')}`; name: query.q(actorEl, null, 'title'),
avatar: actorImagesByActorId[query.url(actorEl, null).match(/\/view\/id\/(\d+)/)?.[1]],
}));
return { release.likes = query.number('.label-rating .like');
name: $(actorElement).attr('title'), release.dislikes = query.number('.label-rating .dislike');
avatar: [avatar.replace('medium.jpg', 'large.jpg'), avatar],
};
}).toArray();
release.actors = actorsFromCards || $('.related-model a').map((actorIndex, actorElement) => $(actorElement).text()).toArray(); const tags = query.all('.tag-card-container a', true);
const categories = query.all('.timeline a[href*="/categories"]', 'title');
release.likes = Number($('.label-rating .like').text());
release.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');
release.channel = slugify(siteName, '');
const tags = $('.tag-card-container a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
const categories = $('.timeline a[href*="/categories"]').map((tagIndex, categoryElement) => $(categoryElement).attr('title')).toArray();
release.tags = tags.concat(categories); release.tags = tags.concat(categories);
release.photos = $('.carousel-thumb a').map((photoIndex, photoElement) => `https:${$(photoElement).attr('href')}`).toArray(); release.channel = slugify(query.q('.scene-site .label-text', true) || query.q('.niche-site-logo', 'title'), '');
const posterPath = videoData?.poster || $('meta[itemprop="thumbnailUrl"]').attr('content') || $('#trailer-player-container').attr('data-player-img'); const videoData = getVideoData(html);
if (posterPath) release.poster = `https:${posterPath}`; const poster = videoData?.poster || query.meta('itemprop="thumbnailUrl"') || query.q('#trailer-player-container', 'data-player-img');
release.poster = qu.prefixUrl(poster);
release.photos = query.urls('.carousel-thumb a');
if (videoData) { if (videoData) {
release.trailer = Object.entries(videoData.stream_info.http.paths).map(([quality, path]) => ({ release.trailer = Object.entries(videoData.stream_info.http.paths).map(([quality, path]) => ({
src: `https:${path}`, src: qu.prefixUrl(path),
quality: Number(quality.match(/\d{3,}/)[0]), quality: Number(quality.match(/\d{3,}/)[0]),
})); }));
} }
console.log(release);
return release; return release;
} }
function scrapeActorSearch(html, url, actorName) { async function fetchActorReleases({ query }, accReleases = []) {
const { document } = new JSDOM(html).window; const releases = scrapeAll(qu.initAll(query.all('.release-card.scene')));
const actorLink = document.querySelector(`a[title="${actorName}" i]`); const next = query.url('.pagination .next a');
return actorLink ? actorLink.href : null;
}
async function fetchActorReleases({ qu, html }, accReleases = []) {
const releases = scrapeAll(html);
const next = qu.url('.pagination .next a');
if (next) { if (next) {
const url = `https://www.brazzers.com${next}`; const url = `https://www.brazzers.com${next}`;
const res = await get(url); const res = await qu.get(url);
if (res.ok) { if (res.ok) {
return fetchActorReleases(res.item, accReleases.concat(releases)); return fetchActorReleases(res.item, accReleases.concat(releases));
@ -146,12 +115,9 @@ async function fetchActorReleases({ qu, html }, accReleases = []) {
return accReleases.concat(releases); return accReleases.concat(releases);
} }
async function scrapeProfile(html, url, actorName, include) { async function scrapeProfile({ query }, url, actorName, include) {
const qProfile = ex(html); const bioKeys = query.all('.profile-spec-list label', true).map(key => key.replace(/\n+|\s{2,}/g, '').trim());
const { q, qa } = qProfile; const bioValues = query.all('.profile-spec-list var', true).map(value => value.replace(/\n+|\s{2,}/g, '').trim());
const bioKeys = qa('.profile-spec-list label', true).map(key => key.replace(/\n+|\s{2,}/g, '').trim());
const bioValues = qa('.profile-spec-list var', true).map(value => value.replace(/\n+|\s{2,}/g, '').trim());
const bio = bioKeys.reduce((acc, key, index) => ({ ...acc, [key]: bioValues[index] }), {}); const bio = bioKeys.reduce((acc, key, index) => ({ ...acc, [key]: bioValues[index] }), {});
@ -159,17 +125,17 @@ async function scrapeProfile(html, url, actorName, include) {
name: actorName, name: actorName,
}; };
profile.description = q('.model-profile-specs p', true); profile.description = query.q('.model-profile-specs p', true);
if (bio.Ethnicity) profile.ethnicity = bio.Ethnicity; if (bio.Ethnicity) profile.ethnicity = bio.Ethnicity;
if (bio.Measurements && bio.Measurements.match(/\d+[A-Z]+-\d+-\d+/)) [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-'); if (bio.Measurements && bio.Measurements.match(/\d+[A-Z]+-\d+-\d+/)) [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-');
if (bio['Date of Birth'] && bio['Date of Birth'] !== 'Unknown') profile.birthdate = moment.utc(bio['Date of Birth'], 'MMMM DD, YYYY').toDate(); if (bio['Date of Birth'] && bio['Date of Birth'] !== 'Unknown') profile.birthdate = qu.extractDate(bio['Date of Birth'], 'MMMM DD, YYYY');
if (bio['Birth Location']) profile.birthPlace = bio['Birth Location']; if (bio['Birth Location']) profile.birthPlace = bio['Birth Location'];
if (bio['Pussy Type']) profile.pussy = bio['Pussy Type'].split(',').slice(-1)[0].toLowerCase(); if (bio['Pussy Type']) profile.pussy = bio['Pussy Type'].split(',').slice(-1)[0].toLowerCase();
if (bio.Height) profile.height = heightToCm(bio.Height); if (bio.Height) profile.height = heightToCm(bio.Height);
if (bio.Weight) profile.weight = lbsToKg(bio.Weight.match(/\d+/)[0]); if (bio.Weight) profile.weight = lbsToKg(bio.Weight.match(/\d+/)[0]);
if (bio['Hair Color']) profile.hair = hairMap[bio['Hair Color']] || bio['Hair Color'].toLowerCase(); if (bio['Hair Color']) profile.hair = bio['Hair Color'].toLowerCase();
if (bio['Tits Type'] && bio['Tits Type'].match('Natural')) profile.naturalBoobs = true; if (bio['Tits Type'] && bio['Tits Type'].match('Natural')) profile.naturalBoobs = true;
if (bio['Tits Type'] && bio['Tits Type'].match('Enhanced')) profile.naturalBoobs = false; if (bio['Tits Type'] && bio['Tits Type'].match('Enhanced')) profile.naturalBoobs = false;
@ -177,49 +143,60 @@ async function scrapeProfile(html, url, actorName, include) {
if (bio['Body Art'] && bio['Body Art'].match('Tattoo')) profile.hasTattoos = true; if (bio['Body Art'] && bio['Body Art'].match('Tattoo')) profile.hasTattoos = true;
if (bio['Body Art'] && bio['Body Art'].match('Piercing')) profile.hasPiercings = true; if (bio['Body Art'] && bio['Body Art'].match('Piercing')) profile.hasPiercings = true;
const avatarEl = q('.big-pic-model-container img'); const avatarEl = query.q('.big-pic-model-container img');
if (avatarEl) profile.avatar = `https:${avatarEl.src}`; if (avatarEl) profile.avatar = `https:${avatarEl.src}`;
if (include.releases) { if (include.releases) {
profile.releases = await fetchActorReleases(qProfile); profile.releases = await fetchActorReleases({ query });
} }
return profile; return profile;
} }
async function fetchLatest(site, page = 1) { async function fetchLatest(channel, page = 1) {
const res = await bhttp.get(`${site.url}/page/${page}/`); const res = await qu.getAll(`${channel.url}/page/${page}/`, '.release-card.scene');
return scrapeAll(res.body.toString(), site, false); if (res.ok) {
return scrapeAll(res.items, channel, false);
} }
async function fetchUpcoming(site) { return res.status;
const res = await bhttp.get(`${site.url}/`); }
return scrapeAll(res.body.toString(), site, true); async function fetchUpcoming(channel) {
const res = await qu.getAll(`${channel.url}/page/1`, '.release-card.scene');
if (res.ok) {
return scrapeAll(res.items, channel, true);
}
return res.status;
} }
async function fetchScene(url, site) { async function fetchScene(url, site) {
const res = await bhttp.get(url); const res = await qu.get(url);
return scrapeScene(res.body.toString(), url, site); if (res.ok) {
return scrapeScene(res.item, url, site);
}
return res.status;
} }
async function fetchProfile(actorName, context, include) { async function fetchProfile(actorName, context, include) {
const searchUrl = 'https://brazzers.com/pornstars-search/'; const searchRes = await qu.get('https://brazzers.com/pornstars-search/', `a[title="${actorName}" i]`, {
const searchRes = await bhttp.get(searchUrl, {
headers: {
Cookie: `textSearch=${encodeURIComponent(actorName)};`, Cookie: `textSearch=${encodeURIComponent(actorName)};`,
},
}); });
const actorLink = scrapeActorSearch(searchRes.body.toString(), searchUrl, actorName); const actorLink = searchRes.ok && searchRes.item.qu.url(null);
if (actorLink) { if (actorLink) {
const url = `https://brazzers.com${actorLink}`; const url = `https://brazzers.com${actorLink}`;
const res = await bhttp.get(url); const res = await qu.get(url);
return scrapeProfile(res.body.toString(), url, actorName, include); if (res.ok) {
return scrapeProfile(res.item, url, actorName, include);
}
} }
return null; return null;

View File

@ -103,6 +103,12 @@ function text(context, selector, applyTrim = true) {
return applyTrim ? trim(textValue) : textValue; return applyTrim ? trim(textValue) : textValue;
} }
function number(context, selector, attr = true) {
const value = q(context, selector, attr);
return value ? Number(value) : null;
}
function meta(context, selector, attrArg = 'content', applyTrim = true) { function meta(context, selector, attrArg = 'content', applyTrim = true) {
if (/meta\[.*\]/.test(selector)) { if (/meta\[.*\]/.test(selector)) {
return q(context, selector, attrArg, applyTrim); return q(context, selector, attrArg, applyTrim);
@ -119,17 +125,22 @@ function date(context, selector, format, match, attr = 'textContent') {
return extractDate(dateString, format, match); return extractDate(dateString, format, match);
} }
function image(context, selector = 'img', attr = 'src', origin, protocol = 'https') { function image(context, selector = 'img', attr, origin, protocol = 'https') {
const imageEl = q(context, selector, attr); const imageEl = (attr && q(context, selector, attr))
|| q(context, selector, 'src')
|| q(context, selector, 'data-src');
// no attribute means q output will be HTML element return prefixUrl(imageEl, origin, protocol);
return attr ? prefixUrl(imageEl, origin, protocol) : imageEl;
} }
function images(context, selector = 'img', attr = 'src', origin, protocol = 'https') { function images(context, selector = 'img', attr, origin, protocol = 'https') {
const imageEls = all(context, selector, attr); const attribute = attr
|| (q(context, selector, 'src') && 'src')
|| (q(context, selector, 'data-src') && 'data-src');
return attr ? imageEls.map(imageEl => prefixUrl(imageEl, origin, protocol)) : imageEls; const imageEls = all(context, selector, attribute);
return imageEls.map(imageEl => prefixUrl(imageEl, origin, protocol));
} }
function url(context, selector = 'a', attr = 'href', origin, protocol = 'https') { function url(context, selector = 'a', attr = 'href', origin, protocol = 'https') {
@ -225,6 +236,8 @@ const quFuncs = {
imgs: images, imgs: images,
length: duration, length: duration,
meta, meta,
number,
num: number,
poster, poster,
q, q,
text, text,