Files
traxxx/src/scrapers/boobpedia.js
2026-02-09 02:10:40 +01:00

73 lines
2.2 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
const slugify = require('../utils/slugify');
function scrapeProfile({ query }) {
const profile = {};
const bio = query.all('.infobox tr[valign="top"]')
.map((detail) => unprint.query.contents(detail, 'td'))
.reduce((acc, [key, value]) => ({ ...acc, [slugify(key, '_')]: value }), {});
/* unreliable, see: Syren De Mer
const catlinks = qa('#mw-normal-catlinks a', true);
const isTrans = catlinks.some(link => link.match(/shemale|transgender/i));
profile.gender = isTrans ? 'transsexual' : 'female';
*/
profile.dateOfBirth = query.date('.bday', 'YYYY-MM-DD');
profile.description = query.content('.mw-parser-output > p');
profile.placeOfBirth = bio.born;
profile.ethnicity = bio.ethnicity;
profile.nationality = bio.nationality.split(',')[0];
profile.measurements = bio.measurements;
if (bio.bra_cup_size) {
[profile.bust] = bio.bra_cup_size.match(/^\d+\w+/) || [];
}
if (/enhanced/i.test(bio.boobs)) profile.naturalBoobs = false;
if (/natural/i.test(bio.boobs)) profile.naturalBoobs = true;
if (bio.height) profile.height = Number(bio.height.match(/\d+\.\d+/g).slice(-1)[0]) * 100;
if (bio.weight) profile.weight = Number(bio.weight.match(/\d+/g)[1]);
if (bio.eye_color) profile.eyes = bio.eye_color;
if (bio.hair) [profile.hairColor] = bio.hair.split(/(?=[A-Z])/); // field concatted to BrunetteLong
if (bio.blood_group) profile.blood = bio.blood_group;
if (bio.also_known_as) profile.aliases = bio.also_known_as.split(', ');
const avatarThumbPath = query.img('.infobox .mw-file-description img');
if (avatarThumbPath && !/NoImageAvailable/.test(avatarThumbPath)) {
const avatarPath = avatarThumbPath.slice(0, avatarThumbPath.lastIndexOf('/')).replace('thumb/', '');
profile.avatar = unprint.prefixUrl(avatarPath, 'https://www.boobpedia.com');
}
profile.socials = query.urls('.infobox a.external');
return profile;
}
async function fetchProfile({ name: actorName }) {
const actorSlug = slugify(actorName, '_', { lower: false });
const res = await unprint.get(`https://www.boobpedia.com/boobs/${actorSlug}`);
if (res.ok) {
return scrapeProfile(res.context);
}
return null;
}
module.exports = {
fetchProfile,
};