boiii/traxxx
1
0
Fork 0
forked from DebaucheryLibrarian/traxxx
Code Issues Pull Requests Projects Releases Wiki Activity

Moved Arch Angel to Full Porn Network and adapted scraper.

Browse Source
This commit is contained in:
DebaucheryLibrarian
2026-01-20 04:28:49 +01:00
parent b2116f728f
commit 2a4dce106e
8 changed files with 113 additions and 219 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
src/scrapers/pornhub.js
View File

@@ -4,20 +4,13 @@ const { JSDOM } = require('jsdom');
const moment = require('moment');
const http = require('../utils/http');
const ethnicityMap = {
White: 'Caucasian',
};
const hairMap = {
Brunette: 'brown',
};
const slugify = require('../utils/slugify');
async function scrapeProfile(html, _url, actorName) {
const { document } = new JSDOM(html).window;
const entries = Array.from(document.querySelectorAll('.infoPiece'), (el) => el.textContent.replace(/\n|\t/g, '').split(':'));
const bio = entries.reduce((acc, [key, value]) => (key ? { ...acc, [key.trim()]: value.trim() } : acc), {});
const bio = entries.reduce((acc, [key, value]) => (key ? { ...acc, [slugify(key, '_')]: value.trim() } : acc), {});
const profile = {
name: actorName,
@@ -26,25 +19,26 @@ async function scrapeProfile(html, _url, actorName) {
const descriptionString = document.querySelector('div[itemprop="description"]') || document.querySelector('.longBio');
const avatarEl = document.querySelector('#getAvatar') || document.querySelector('.thumbImage img');
if (bio.Gender) profile.gender = bio.Gender.toLowerCase();
if (bio.ethnicity) profile.ethnicity = ethnicityMap[bio.Ethnicity] || bio.Ethnicity;
if (bio.gender) profile.gender = bio.gender;
if (bio.ethnicity) profile.ethnicity = bio.ethnicity;
if (descriptionString) profile.description = descriptionString.textContent;
if (bio.Birthday && !/-0001/.test(bio.Birthday)) profile.birthdate = moment.utc(bio.Birthday, 'MMM D, YYYY').toDate(); // birthyear sometimes -0001, see Spencer Bradley as of january 2020
if (bio.Born) profile.birthdate = moment.utc(bio.Born, 'YYYY-MM-DD').toDate();
if (bio.birthday && !/-0001/.test(bio.birthday)) profile.birthdate = moment.utc(bio.Birthday, 'MMM D, YYYY').toDate(); // birthyear sometimes -0001, see Spencer Bradley as of january 2020
if (bio.born) profile.birthdate = moment.utc(bio.born, 'YYYY-MM-DD').toDate();
profile.birthPlace = bio['Birth Place'] || bio.Birthplace;
profile.residencePlace = bio['City and Country'];
profile.birthPlace = bio.birth_place || bio.birthplace;
profile.residencePlace = bio.city_and_country;
if (bio.Measurements && bio.Measurements !== '--') [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-');
if (bio['Fake Boobs']) profile.naturalBoobs = bio['Fake Boobs'] === 'No';
if (bio.measurements && bio.measurements !== '--') profile.measurements = bio.measurements;
if (bio.fake_boobs) profile.naturalBoobs = bio.fake_boobs.toLowerCase() === 'no';
if (bio.Height) profile.height = Number(bio.Height.match(/\(\d+/)[0].slice(1));
if (bio.Weight) profile.weight = Number(bio.Weight.match(/\(\d+/)[0].slice(1));
if (bio['Hair Color']) profile.hair = hairMap[bio['Hair Color']] || bio['Hair Color'].toLowerCase();
if (bio.Piercings) profile.hasPiercings = bio.Piercings === 'Yes';
if (bio.Tattoos) profile.hasTattoos = bio.Tattoos === 'Yes';
if (bio.height) profile.height = Number(bio.height.match(/\(\d+/)[0].slice(1));
if (bio.weight) profile.weight = Number(bio.weight.match(/\(\d+/)[0].slice(1));
if (bio.hair_color) profile.hairColor = bio.hair_color;
if (bio.eyes) profile.eyeColor = bio.eye_color;
if (bio.piercings) profile.hasPiercings = bio.piercings.toLowerCase() === 'yes';
if (bio.tattoos) profile.hasTattoos = bio.tattoos.toLowerCase() === 'yes';
if (avatarEl && !/default\//.test(avatarEl.src)) profile.avatar = avatarEl.src;
profile.social = Array.from(document.querySelectorAll('.socialList a'), (el) => el.href).filter((link) => link !== 'https://www.twitter.com/'); // PH links to Twitter itself for some reason