traxxx/src/scrapers/template.js

99 lines
2.2 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
const slugify = require('../utils/slugify');
function scrapeAll(scenes) {
return scenes.map(({ query }) => {
const release = {};
release.url = query.url('.title a');
release.entryId = new URL(release.url).pathname.match(/\/scene\/(\d+)/)[1];
release.title = query.content('.title a');
release.description = query.content('.description');
release.date = query.date('.date', 'MMM DD, YYYY');
release.actors = query.contents('.models a.model');
release.poster = query.img('img.poster');
release.teaser = { src: query.video('.teaser video') };
release.stars = query.number('.rating');
release.likes = query.number('.likes');
console.log(release);
return release;
});
}
function scrapeScene({ query }, url) {
const release = {};
release.entryId = new URL(url).pathname.match(/\/scene\/(\d+)/)[1];
release.title = query.content('h3.title');
release.description = query.content('p.description');
[release.poster, ...release.photos] = query.imgs('.preview-thumb');
release.trailer = query.video('.trailer video');
console.log(release);
return release;
}
function scrapeProfile({ query }, actorName, entity, include) {
const profile = {};
profile.description = query.content('.bio-text');
profile.birthPlace = query.content('.birth-place span');
profile.avatar = query.img('.actor-photo img');
if (include.releases) {
return scrapeAll(unprint.initAll(query.all('.scene')));
}
console.log(profile);
return profile;
}
async function fetchLatest(channel, page = 1) {
const url = `${channel.url}/${page}`;
const res = await unprint.get(url, { selectAll: '.scene' });
if (res.ok) {
return scrapeAll(res.context, channel);
}
return res.status;
}
async function fetchScene(url, channel) {
const res = await unprint.get(url);
if (res.ok) {
return scrapeScene(res.context, url, channel);
}
return res.status;
}
async function fetchProfile({ name: actorName }, entity, include) {
const url = `${entity.url}/actors/${slugify(actorName, '_')}`;
const res = await unprint.get(url);
if (res.ok) {
return scrapeProfile(res.context, actorName, entity, include);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchScene,
fetchProfile,
};