Added Naughty America profile and actor releases scraper.

This commit is contained in:
2020-01-31 19:25:42 +01:00
parent c882862af6
commit 504bcd02e3
6 changed files with 37 additions and 4 deletions

View File

@@ -5,6 +5,9 @@ const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
const slugify = require('../utils/slugify');
const { ex } = require('../utils/q');
function titleExtractor(pathname) {
const components = pathname.split('/')[2].split('-');
const entryId = components.slice(-1)[0];
@@ -48,7 +51,7 @@ function scrapeLatest(html, site) {
});
}
async function scrapeScene(html, url, site) {
function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElement = $('.scene-info');
@@ -97,6 +100,20 @@ async function scrapeScene(html, url, site) {
};
}
function scrapeProfile(html) {
const { q, qu } = ex(html);
const profile = {};
profile.description = q('.bio_about_text', true);
const avatar = q('img.performer-pic', 'src');
if (avatar) profile.avatar = `https:${avatar}`;
profile.releases = qu('.scene-item > a:first-child');
return profile;
}
async function fetchLatest(site, page = 1) {
const res = await bhttp.get(`${site.url}?page=${page}`);
@@ -109,7 +126,20 @@ async function fetchScene(url, site) {
return scrapeScene(res.body.toString(), url, site);
}
async function fetchProfile(actorName) {
const actorSlug = slugify(actorName);
const res = await bhttp.get(`https://www.naughtyamerica.com/pornstar/${actorSlug}`);
if (res.statusCode === 200) {
return scrapeProfile(res.body.toString());
}
return null;
}
module.exports = {
fetchLatest,
fetchScene,
fetchProfile,
};