traxxx/src/scrapers/freeones.js

'use strict';

/* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp');
const { JSDOM } = require('jsdom');
const moment = require('moment');

const knex = require('../knex');

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 birthdateString = bio['Date of Birth:'];
    const birthdate = birthdateString && birthdateString !== 'Unknown (Add)'
        ? moment.utc(birthdateString.slice(0, birthdateString.indexOf(' (')), 'MMMM D, YYYY').toDate()
        : null;

    const measurementsString = bio['Measurements:'];
    const [bust, waist, hip] = measurementsString.split('-').map(measurement => (measurement === '??' ? null : measurement));
    const naturalBoobs = bio['Fake Boobs:'] === 'No';

    const residenceCountryName = bio['Country of Origin:'];
    const countryEntry = await knex('countries').where({ name: residenceCountryName }).first();
    const residenceCountry = countryEntry ? countryEntry.alpha2 : null;
    const birthPlace = bio['Place of Birth:'];

    const hair = bio['Hair Color:'].toLowerCase();
    const eyes = bio['Eye Color:'].toLowerCase();

    const piercingsString = bio['Piercings:'];
    const hasPiercings = !!(piercingsString !== undefined && piercingsString !== 'Unknown (add)' && piercingsString !== 'None');
    const piercings = hasPiercings && piercingsString;

    const tattoosString = bio['Tattoos:'];
    const hasTattoos = !!(tattoosString !== undefined && tattoosString !== 'Unknown (add)' && tattoosString !== 'None');
    const tattoos = hasTattoos && tattoosString;

    const social = Array.from(bioEl.querySelectorAll('.dashboard-socialmedia a'), el => el.href);

    return {
        bio: {
            name,
            gender: 'female',
            birthdate,
            residenceCountry,
            birthPlace,
            naturalBoobs,
            bust,
            waist,
            hip,
            hair,
            eyes,
            piercings,
            tattoos,
            social,
        },
        url: bioUrl,
    };
}

async function scrapeProfileBio(html, frontpageBio, 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 birthdateString = bio['Date of Birth:'];
    const birthdate = birthdateString && birthdateString !== 'Unknown'
        ? moment.utc(birthdateString.slice(0, birthdateString.indexOf(' (')), 'MMMM D, YYYY').toDate()
        : null;

    const measurementsString = bio['Measurements:'];
    const [bust, waist, hip] = measurementsString.split('-').map(measurement => (measurement === '??' ? null : measurement));
    const boobsNatural = bio['Fake boobs:'] === 'No';
    const ethnicity = bio['Ethnicity:'];

    const residenceCountryName = bio['Country of Origin:'];
    const countryEntry = await knex('countries').where({ name: residenceCountryName }).first();
    const residenceCountry = countryEntry ? countryEntry.alpha2 : null;
    const birthPlace = bio['Place of Birth:'];

    const hair = bio['Hair Color:'].toLowerCase();
    const eyes = bio['Eye Color:'].toLowerCase();
    const height = Number(bio['Height:'].match(/\d+/)[0]);
    const weight = Number(bio['Weight:'].match(/\d+/)[0]);

    const piercingsString = bio['Piercings:'];
    const hasPiercings = !!(piercingsString !== undefined && piercingsString !== 'Unknown (add)' && piercingsString !== 'None');
    const piercings = hasPiercings && piercingsString;

    const tattoosString = bio['Tattoos:'];
    const hasTattoos = !!(tattoosString !== undefined && tattoosString !== 'Unknown (add)' && tattoosString !== 'None');
    const tattoos = hasTattoos && tattoosString;

    const social = Array.from(bioEl.querySelectorAll('#socialmedia a'), el => el.href);

    return {
        ...frontpageBio,
        name,
        gender: 'female',
        birthdate,
        residenceCountry,
        birthPlace,
        ethnicity,
        naturalBoobs: boobsNatural,
        bust,
        waist,
        hip,
        height,
        weight,
        hair,
        eyes,
        hasPiercings,
        hasTattoos,
        piercings,
        tattoos,
        social,
    };
}

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, bio } = await scrapeProfileFrontpage(resFallback.body.toString(), fallbackUrl, actorName);
        const resBio = await bhttp.get(url);

        return scrapeProfileBio(resBio.body.toString(), bio, url, actorName);
    }

    return null;
}

module.exports = {
    fetchProfile,
};