Added basic actor and network overview. Added DDF Network actor scraper. Various bug fixes and layout improvements.
This commit is contained in:
@@ -2,9 +2,10 @@
|
||||
|
||||
const bhttp = require('bhttp');
|
||||
const cheerio = require('cheerio');
|
||||
const { JSDOM } = require('jsdom');
|
||||
const moment = require('moment');
|
||||
|
||||
// const knex = require('../knex');
|
||||
const knex = require('../knex');
|
||||
const { matchTags } = require('../tags');
|
||||
|
||||
/* eslint-disable newline-per-chained-call */
|
||||
@@ -105,6 +106,67 @@ async function scrapeScene(html, url, site) {
|
||||
};
|
||||
}
|
||||
|
||||
async function scrapeProfile(html, _url, actorName) {
|
||||
const { document } = new JSDOM(html).window;
|
||||
|
||||
const keys = Array.from(document.querySelectorAll('.about-title'), el => el.textContent.trim().replace(':', ''));
|
||||
const values = Array.from(document.querySelectorAll('.about-info'), (el) => {
|
||||
if (el.children.length > 0) {
|
||||
return Array.from(el.children, child => child.textContent.trim()).join(', ');
|
||||
}
|
||||
|
||||
return el.textContent.trim();
|
||||
});
|
||||
|
||||
const bio = keys.reduce((acc, key, index) => {
|
||||
if (values[index] === '-') {
|
||||
return acc;
|
||||
}
|
||||
|
||||
return {
|
||||
...acc,
|
||||
[key]: values[index],
|
||||
};
|
||||
}, {});
|
||||
|
||||
const descriptionEl = document.querySelector('.description-box');
|
||||
const avatarEl = document.querySelector('.pornstar-details .card-img-top');
|
||||
|
||||
const country = await knex('countries')
|
||||
.where('nationality', 'ilike', `%${bio.Nationality}%`)
|
||||
.orderBy('priority', 'desc')
|
||||
.first();
|
||||
|
||||
const profile = {
|
||||
name: actorName,
|
||||
};
|
||||
|
||||
profile.birthdate = moment.utc(bio.Birthday, 'MMMM DD, YYYY').toDate();
|
||||
if (country) profile.birthPlace = country.name;
|
||||
|
||||
if (bio['Bra size']) [profile.bust] = bio['Bra size'].match(/\d+\w+/);
|
||||
if (bio.Waist) profile.waist = Number(bio.Waist.match(/\d+/)[0]);
|
||||
if (bio.Hips) profile.hip = Number(bio.Hips.match(/\d+/)[0]);
|
||||
|
||||
if (bio.Height) profile.height = Number(bio.Height.match(/\d{2,}/)[0]);
|
||||
|
||||
if (bio['Tit Style'] && bio['Tit Style'].match('Enhanced')) profile.naturalBoobs = false;
|
||||
if (bio['Tit Style'] && bio['Tit Style'].match('Natural')) profile.naturalBoobs = true;
|
||||
|
||||
if (bio['Body Art'] && bio['Body Art'].match('Tattoo')) profile.hasTattoos = true;
|
||||
if (bio['Body Art'] && bio['Body Art'].match('Piercing')) profile.hasPiercings = true;
|
||||
|
||||
if (bio['Hair Style']) profile.hair = bio['Hair Style'].split(',')[0].trim().toLowerCase();
|
||||
if (bio['Eye Color']) profile.eyes = bio['Eye Color'].match(/\w+/)[0].toLowerCase();
|
||||
|
||||
if (bio['Shoe size']) profile.shoes = Number(bio['Shoe size'].split('|')[1]);
|
||||
|
||||
if (descriptionEl) profile.description = descriptionEl.textContent.trim();
|
||||
if (avatarEl) profile.avatar = `https:${avatarEl.dataset.src}`;
|
||||
|
||||
return profile;
|
||||
}
|
||||
|
||||
async function fetchLatest(site, page = 1) {
|
||||
const res = await bhttp.get(`https://ddfnetwork.com/videos/search/latest/ever/${new URL(site.url).hostname}/-/${page}`);
|
||||
|
||||
@@ -117,7 +179,41 @@ async function fetchScene(url, site) {
|
||||
return scrapeScene(res.body.toString(), url, site);
|
||||
}
|
||||
|
||||
async function fetchProfile(actorName) {
|
||||
const resSearch = await bhttp.post('https://ddfnetwork.com/search/ajax',
|
||||
{
|
||||
type: 'hints',
|
||||
word: actorName,
|
||||
},
|
||||
{
|
||||
decodeJSON: true,
|
||||
headers: {
|
||||
'x-requested-with': 'XMLHttpRequest',
|
||||
},
|
||||
});
|
||||
|
||||
if (resSearch.statusCode !== 200 || Array.isArray(resSearch.body.list)) {
|
||||
return null;
|
||||
}
|
||||
|
||||
if (!resSearch.body.list.pornstarsName || resSearch.body.list.pornstarsName.length === 0) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const [actor] = resSearch.body.list.pornstarsName;
|
||||
const url = `https://ddfnetwork.com${actor.href}`;
|
||||
|
||||
const resActor = await bhttp.get(url);
|
||||
|
||||
if (resActor.statusCode !== 200) {
|
||||
return null;
|
||||
}
|
||||
|
||||
return scrapeProfile(resActor.body.toString(), url, actorName);
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
fetchLatest,
|
||||
fetchProfile,
|
||||
fetchScene,
|
||||
};
|
||||
|
||||
Reference in New Issue
Block a user