traxxx/src/scrapers/kellymadison.js

188 lines
5.8 KiB
JavaScript
Executable File

'use strict';
const slugify = require('../utils/slugify');
const qu = require('../utils/qu');
const http = require('../utils/http');
const { feetInchesToCm } = require('../utils/convert');
const siteMapByKey = {
PF: 'pornfidelity',
TF: 'teenfidelity',
KM: 'kellymadison',
'5KP': '5kporn',
'5KT': '5kteens',
};
const siteMapBySlug = Object.entries(siteMapByKey).reduce((acc, [key, value]) => ({ ...acc, [value]: key }), {});
function scrapeLatest(scenes, site) {
return scenes.reduce((acc, { query }) => {
const release = {};
release.shootId = query.q('.card-meta .text-right, .row .text-right, .card-footer-item:last-child', true);
const siteId = release.shootId.match(/\d?\w{2}/)[0];
const siteSlug = siteMapByKey[siteId];
const { pathname } = new URL(query.url('h5 a, .ep-title a, .title a'));
[release.entryId] = pathname.match(/\d+$/);
release.url = `${site.url}${pathname}`;
release.title = query.cnt('h5 a, .ep-title a, .title a');
release.date = query.date('.card-meta .text-left, .row .col-4:first-child, .card-footer-item:first-child', ['MMM D', 'MMM D, YYYY'], /\w+ \d+(, \w+)?/);
release.actors = query.cnts('.models a, .ep-models a, a[href*="models/"]');
release.duration = query.dur('.content a');
const duration = query.cnt('.content a, .ep-runtime strong, .subtitle:last-child a')?.match(/(\d+) min/)?.[1];
if (duration) release.duration = Number(duration) * 60;
if (query.exists('.episodes-preview')) {
[release.poster, ...release.photos] = query.imgs('.episodes-preview img');
} else {
release.poster = query.img('.card-img-top, .image img');
release.teaser = {
src: query.video('video'),
};
}
if (site.slug !== siteSlug) {
// using generic network overview, scene is not from the site we want
return { ...acc, unextracted: [...acc.unextracted, release] };
}
return { ...acc, scenes: [...acc.scenes, release] };
}, {
scenes: [],
unextracted: [],
});
}
async function scrapeScene({ query, html }, url, baseRelease, channel, session) {
const { pathname } = new URL(url);
const release = {};
[release.entryId] = pathname.match(/\d+$/);
const titleString = query.cnt('.card-header.row h4, .trailer-starring span, .level-left .level-item');
const episode = titleString?.match(/#\d+$/)?.[0];
release.title = query.cnt('.trailer-title') || titleString?.match(/(?:Trailer: )?(.+) -/)?.[1];
release.channel = slugify(titleString?.match(/([\w\s]+) #\d+$/)?.[1], '');
const siteKey = siteMapBySlug[release.channel];
release.shootId = `${siteKey} ${episode}`;
release.description = query.text('p.card-text, h5.heavy + p, .card-content .is-three-fifths');
// order not reliable, get keys
const detailElsByKey = query.all('.card-body h4.card-title, .video-summary h5, .columns li').reduce((acc, rowEl) => ({
...acc,
[slugify(rowEl?.textContent.match(/(\w+):/)?.[1])]: rowEl,
}), {});
release.date = query.date(detailElsByKey.published, null, 'YYYY-MM-DD', /\d{4}-\d{2}-\d{2}/);
release.duration = query.dur(detailElsByKey.episode);
release.actors = query.cnts(detailElsByKey.starring, 'a');
const posterPrefix = html.indexOf('poster:');
const poster = query.img('.trailer-poster') || html.slice(html.indexOf('http', posterPrefix), html.indexOf('.jpg', posterPrefix) + 4);
if (poster) {
if (baseRelease?.poster) {
release.photos = [poster, ...(baseRelease.photos || [])];
} else {
release.poster = poster;
}
}
// const token = query.meta('name=_token');
// const trailerInfoUrl = `${channel.url}/episodes/trailer/sources/${release.entryId}?type=trailer`;
const trailerInfoUrl = html.match(/'(http.*\/trailer\/sources.*)'/)?.[1];
const trailerInfoRes = await http.post(trailerInfoUrl, null, { session });
if (trailerInfoRes.ok && trailerInfoRes.body.sources?.length > 0) {
release.trailer = trailerInfoRes.body.sources.map((trailer) => ({
src: trailer.src,
type: trailer.type,
/* unreliable, sometimes actual video is 720p
quality: trailer.res
.replace(4000, 2160)
.replace(5000, 2880),
*/
}));
}
console.log(release);
return release;
}
function scrapeProfile({ query }) {
const profile = {};
const bioKeys = query.all('table.table td:nth-child(1)', true);
const bioValues = query.all('table.table td:nth-child(2)', true);
const bio = bioKeys.reduce((acc, key, index) => ({ ...acc, [key.slice(0, -1)]: bioValues[index] }), {});
if (bio.Ethnicity) profile.ethnicity = bio.Ethnicity;
if (bio.Measurements) [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-');
if (bio.Birthplace) profile.birthPlace = bio.Birthplace;
if (bio.Height) {
const [feet, inches] = bio.Height.match(/\d+/g);
profile.height = feetInchesToCm(feet, inches);
}
profile.avatar = query.img('img[src*="model"]');
return profile;
}
async function fetchLatest(channel, page = 1) {
const url = `${channel.url}/episodes/search?page=${page}`; // TLS issues with teenfidelity.com, same overview on all sites
const res = await http.get(url, {
headers: {
'X-Requested-With': 'XMLHttpRequest',
},
});
if (res.ok && res.body.status === 'success') {
return scrapeLatest(qu.extractAll(res.body.html, '.episode, .ep'), channel);
}
return res.status;
}
async function fetchScene(url, channel, baseRelease) {
const session = http.session();
const res = await qu.get(url, null, {
'X-Requested-With': 'XMLHttpRequest',
}, { session });
return res.ok ? scrapeScene(res.item, url, baseRelease, channel, session) : res.status;
}
async function fetchProfile({ name: actorName }) {
const actorSlug = slugify(actorName);
const res = await qu.get(`https://www.kellymadison.com/models/${actorSlug}`, null, {
headers: {
'X-Requested-With': 'XMLHttpRequest',
},
});
if (res.ok) {
return scrapeProfile(res.item);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
fetchScene,
};