'use strict'; const bhttp = require('bhttp'); const cheerio = require('cheerio'); const moment = require('moment'); const { matchTags } = require('../tags'); function scrapeLatest(html, site) { const $ = cheerio.load(html, { normalizeWhitespace: true }); const scenesElements = $('.update_details').toArray(); return scenesElements.map((element) => { const photoElement = $(element).find('a img.thumbs'); const photoCount = Number(photoElement.attr('cnt')); const photos = Array.from({ length: photoCount }, (value, index) => photoElement.attr(`src${index}_1x`)).filter(photoUrl => photoUrl !== undefined); const sceneLinkElement = $(element).children('a').eq(1); const url = sceneLinkElement.attr('href'); const title = sceneLinkElement.text(); const shootId = $(element).attr('data-setid'); const date = moment .utc($(element).find('.update_date').text(), 'MM/DD/YYYY') .toDate(); const actors = $(element).find('.update_models a') .map((actorIndex, actorElement) => $(actorElement).text()) .toArray(); return { url, shootId, title, actors, date, site, photos, }; }); } function scrapeUpcoming(html, site) { const $ = cheerio.load(html, { normalizeWhitespace: true }); const scenesElements = $('#coming_soon_carousel').find('.table').toArray(); return scenesElements.map((element) => { const photoElement = $(element).find('a img.thumbs'); const photoCount = Number(photoElement.attr('cnt')); const photos = Array.from({ length: photoCount }, (value, index) => photoElement.attr(`src${index}_1x`)).filter(photoUrl => photoUrl !== undefined); const shootId = $(element).find('.upcoming_updates_thumb').attr('id').match(/\d+/)[0]; const details = $(element).find('.update_details_comingsoon') .eq(1) .children() .remove(); const title = details .end() .text() .trim(); const actors = details .text() .trim() .split(', '); const date = moment .utc($(element).find('.update_date_comingsoon').text().slice(7), 'MM/DD/YYYY') .toDate(); return { url: null, shootId, title, date, actors, photos, rating: null, site, }; }); } async function scrapeScene(html, url, site) { const $ = cheerio.load(html, { normalizeWhitespace: true }); const title = $('.title_bar_hilite').text().trim(); const date = moment .utc($('.update_date').text(), 'MM/DD/YYYY') .toDate(); const actors = $('.backgroundcolor_info > .update_models a') .map((_actorIndex, actorElement) => $(actorElement).text()) .toArray(); const description = $('.update_description').text().trim(); const stars = Number($('.avg_rating').text().trim().replace(/[\s|Avg Rating:]/g, '')); const rawTags = $('.update_tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray(); const tags = await matchTags(rawTags); return { url, title, date, actors, description, rating: { stars, }, tags, site, }; } async function fetchLatest(site, page = 1) { const res = await bhttp.get(`${site.url}/trial/categories/movies_${page}_d.html`); return scrapeLatest(res.body.toString(), site); } async function fetchUpcoming(site) { const res = await bhttp.get(`${site.url}/trial/index.php`); return scrapeUpcoming(res.body.toString(), site); } async function fetchScene(url, site) { const res = await bhttp.get(url); return scrapeScene(res.body.toString(), url, site); } module.exports = { fetchLatest, fetchUpcoming, fetchScene, };