141 lines
5.8 KiB
JavaScript
141 lines
5.8 KiB
JavaScript
'use strict';
|
|
|
|
/* eslint-disable newline-per-chained-call */
|
|
const bhttp = require('bhttp');
|
|
const { JSDOM } = require('jsdom');
|
|
const moment = require('moment');
|
|
|
|
async function scrapeProfileFrontpage(html, url, name) {
|
|
const { document } = new JSDOM(html).window;
|
|
const bioEl = document.querySelector('.dashboard-bio-list');
|
|
|
|
const bioUrl = `https:${document.querySelector('.seemore a').href}`;
|
|
|
|
const keys = Array.from(bioEl.querySelectorAll('dt'), el => el.textContent.trim());
|
|
const values = Array.from(bioEl.querySelectorAll('dd'), el => el.textContent.trim());
|
|
|
|
const bio = keys.reduce((acc, key, index) => ({ ...acc, [key]: values[index] }), {});
|
|
|
|
const profile = {
|
|
name,
|
|
gender: 'female',
|
|
};
|
|
|
|
const birthdateString = bio['Date of Birth:'];
|
|
const measurementsString = bio['Measurements:'];
|
|
|
|
const birthCityString = bio['Place of Birth:'];
|
|
const birthCity = birthCityString !== undefined && birthCityString !== 'Unknown' && birthCityString !== 'Unknown (add)' && birthCityString;
|
|
|
|
const birthCountryString = bio['Country of Origin:'];
|
|
const birthCountry = birthCountryString !== undefined && birthCountryString !== 'Unknown' && birthCountryString !== 'Unknown (add)' && birthCountryString;
|
|
|
|
const piercingsString = bio['Piercings:'];
|
|
const tattoosString = bio['Tattoos:'];
|
|
|
|
if (birthdateString && birthdateString !== 'Unknown (add)') profile.birthdate = moment.utc(birthdateString.slice(0, birthdateString.indexOf(' (')), 'MMMM D, YYYY').toDate();
|
|
if (measurementsString) [profile.bust, profile.waist, profile.hip] = measurementsString.split('-').map(measurement => (measurement === '??' ? null : measurement));
|
|
|
|
if (bio['Fake Boobs:']) profile.naturalBoobs = bio['Fake Boobs:'] === 'No';
|
|
profile.birthPlace = `${birthCity || ''}${birthCity ? ', ' : ''}${birthCountry || ''}`;
|
|
|
|
profile.hair = bio['Hair Color:'].toLowerCase();
|
|
profile.eyes = bio['Eye Color:'].toLowerCase();
|
|
|
|
if (piercingsString) profile.hasPiercings = !!(piercingsString !== 'Unknown (add)' && piercingsString !== 'None');
|
|
if (tattoosString) profile.hasTattoos = !!(tattoosString !== 'Unknown (add)' && tattoosString !== 'None');
|
|
|
|
if (profile.hasPiercings && piercingsString !== 'various') profile.piercings = piercingsString;
|
|
if (profile.hasTattoos && tattoosString !== 'various') profile.tattoos = tattoosString;
|
|
|
|
profile.social = Array.from(bioEl.querySelectorAll('.dashboard-socialmedia a'), el => el.href);
|
|
|
|
return {
|
|
profile,
|
|
url: bioUrl,
|
|
};
|
|
}
|
|
|
|
async function scrapeProfileBio(html, frontpageProfile, url, name) {
|
|
const { document } = new JSDOM(html).window;
|
|
const bioEl = document.querySelector('#biographyTable');
|
|
|
|
const keys = Array.from(bioEl.querySelectorAll('td:nth-child(1)'), el => el.textContent.trim());
|
|
const values = Array.from(bioEl.querySelectorAll('td:nth-child(2)'), el => el.textContent.trim());
|
|
|
|
const bio = keys.reduce((acc, key, index) => ({ ...acc, [key]: values[index] }), {});
|
|
|
|
const profile = {
|
|
...frontpageProfile,
|
|
name,
|
|
gender: 'female',
|
|
};
|
|
|
|
const birthdateString = bio['Date of Birth:'];
|
|
const measurementsString = bio['Measurements:'];
|
|
|
|
const birthCityString = bio['Place of Birth:'];
|
|
const birthCity = birthCityString !== undefined && birthCityString !== 'Unknown' && birthCityString !== 'Unknown (add)' && birthCityString;
|
|
|
|
const birthCountryString = bio['Country of Origin:'];
|
|
const birthCountry = birthCountryString !== undefined && birthCountryString !== 'Unknown' && birthCountryString !== 'Unknown (add)' && birthCountryString;
|
|
|
|
const piercingsString = bio['Piercings:'];
|
|
const tattoosString = bio['Tattoos:'];
|
|
|
|
if (birthdateString && birthdateString !== 'Unknown') profile.birthdate = moment.utc(birthdateString.slice(0, birthdateString.indexOf(' (')), 'MMMM D, YYYY').toDate();
|
|
if (measurementsString) [profile.bust, profile.waist, profile.hip] = measurementsString.split('-').map(measurement => (measurement === '??' ? null : measurement));
|
|
|
|
if (bio['Fake boobs']) profile.naturalBoobs = bio['Fake boobs:'] === 'No';
|
|
profile.ethnicity = bio['Ethnicity:'];
|
|
|
|
profile.birthPlace = `${birthCity || ''}${birthCity ? ', ' : ''}${birthCountry || ''}`;
|
|
|
|
profile.hair = bio['Hair Color:'].toLowerCase();
|
|
profile.eyes = bio['Eye Color:'].toLowerCase();
|
|
profile.height = Number(bio['Height:'].match(/\d+/)[0]);
|
|
profile.weight = Number(bio['Weight:'].match(/\d+/)[0]);
|
|
|
|
if (piercingsString) profile.hasPiercings = !!(piercingsString !== 'Unknown (add)' && piercingsString !== 'None');
|
|
if (tattoosString) profile.hasTattoos = !!(tattoosString !== 'Unknown (add)' && tattoosString !== 'None');
|
|
|
|
if (profile.hasPiercings && piercingsString !== 'various') profile.piercings = piercingsString;
|
|
if (profile.hasTattoos && tattoosString !== 'various') profile.tattoos = tattoosString;
|
|
|
|
profile.social = Array.from(bioEl.querySelectorAll('#socialmedia a'), el => el.href);
|
|
|
|
return profile;
|
|
}
|
|
|
|
async function fetchProfile(actorName) {
|
|
const slug = actorName.replace(' ', '_');
|
|
const frontpageUrl = `https://www.freeones.com/html/v_links/${slug}`;
|
|
|
|
const resFrontpage = await bhttp.get(frontpageUrl);
|
|
|
|
if (resFrontpage.statusCode === 200) {
|
|
const { url, bio } = await scrapeProfileFrontpage(resFrontpage.body.toString(), frontpageUrl, actorName);
|
|
const resBio = await bhttp.get(url);
|
|
|
|
return scrapeProfileBio(resBio.body.toString(), bio, url, actorName);
|
|
}
|
|
|
|
// apparently some actors are appended 'Babe' as their surname...
|
|
const fallbackSlug = `${slug}_Babe`;
|
|
const fallbackUrl = `https://www.freeones.com/html/s_links/${fallbackSlug}`;
|
|
const resFallback = await bhttp.get(fallbackUrl);
|
|
|
|
if (resFallback.statusCode === 200) {
|
|
const { url, profile } = await scrapeProfileFrontpage(resFallback.body.toString(), fallbackUrl, actorName);
|
|
const resBio = await bhttp.get(url);
|
|
|
|
return scrapeProfileBio(resBio.body.toString(), profile, url, actorName);
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
module.exports = {
|
|
fetchProfile,
|
|
};
|