Added Kelly Madison profile scraper.

src/actors.js

@@ -2,6 +2,7 @@
 
 const Promise = require('bluebird');
 const UrlPattern = require('url-pattern');
+const moment = require('moment');
 
 const knex = require('./knex');
 const argv = require('./argv');
@@ -49,6 +50,10 @@ async function curateActor(actor) {
         scrapedAt: actor.scraped_at,
     };
 
+    if (curatedActor.birthdate) {
+        curatedActor.age = moment().diff(curatedActor.birthdate, 'years');
+    }
+
     if (actor.birth_city) curatedActor.origin.city = actor.birth_city;
     if (actor.birth_state) curatedActor.origin.state = actor.birth_state;
 

src/releases.js

@@ -1,6 +1,8 @@
 'use strict';
 
 const Promise = require('bluebird');
+const moment = require('moment');
+
 const knex = require('./knex');
 const argv = require('./argv');
 const whereOr = require('./utils/where-or');
@@ -17,9 +19,14 @@ const { fetchSites, findSiteByUrl } = require('./sites');
 async function curateRelease(release) {
     const [actors, tags, media] = await Promise.all([
         knex('actors_associated')
-            .select('actors.id', 'actors.name', 'actors.gender', 'actors.slug', 'media.thumbnail as avatar')
+            .select(
+                'actors.id', 'actors.name', 'actors.gender', 'actors.slug', 'actors.birthdate',
+                'birth_countries.alpha2 as birth_country_alpha2', 'birth_countries.name as birth_country_name', 'birth_countries.alias as birth_country_alias',
+                'media.thumbnail as avatar',
+            )
             .where({ release_id: release.id })
             .leftJoin('actors', 'actors.id', 'actors_associated.actor_id')
+            .leftJoin('countries as birth_countries', 'actors.birth_country_alpha2', 'birth_countries.alpha2')
             .leftJoin('media', (builder) => {
                 builder
                     .on('media.target_id', 'actors.id')
@@ -52,7 +59,21 @@ async function curateRelease(release) {
         url: release.url,
         shootId: release.shoot_id,
         entryId: release.entry_id,
-        actors,
+        actors: actors.map(actor => ({
+            id: actor.id,
+            slug: actor.slug,
+            name: actor.name,
+            gender: actor.gender,
+            birthdate: actor.birthdate,
+            age: moment().diff(actor.birthdate, 'years'),
+            avatar: actor.avatar,
+            origin: {
+                country: {
+                    name: actor.birth_country_alias,
+                    alpha2: actor.birth_country_alpha2,
+                },
+            },
+        })),
         director: release.director,
         tags,
         duration: release.duration,

src/scrapers/kellymadison.js

@@ -4,6 +4,8 @@ const bhttp = require('bhttp');
 const { JSDOM } = require('jsdom');
 const moment = require('moment');
 
+const { feetInchesToCm } = require('../utils/convert');
+
 const siteMapByKey = {
     PF: 'pornfidelity',
     TF: 'teenfidelity',
@@ -112,6 +114,31 @@ function scrapeScene(html, url, site, shallowRelease) {
     return release;
 }
 
+function scrapeProfile(html, actorName) {
+    const { document } = new JSDOM(html).window;
+    const profile = { name: actorName };
+
+    const bioKeys = Array.from(document.querySelectorAll('table.table td:nth-child(1)'), el => el.textContent.slice(0, -1));
+    const bioValues = Array.from(document.querySelectorAll('table.table td:nth-child(2)'), el => el.textContent);
+    const bio = bioKeys.reduce((acc, key, index) => ({ ...acc, [key]: bioValues[index] }), {});
+
+    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);
+    }
+
+    if (bio.Ethnicity) profile.ethnicity = bio.Ethnicity;
+
+    const avatarEl = Array.from(document.querySelectorAll('img')).find(photo => photo.src.match('model'));
+
+    if (avatarEl) profile.avatar = avatarEl.src;
+
+    return profile;
+}
+
 async function fetchLatest(site, page = 1) {
     const url = `https://kellymadison.com/episodes/search?page=${page}`; // TLS issues with teenfidelity.com, same overview on all sites
     const res = await bhttp.get(url, {
@@ -139,7 +166,23 @@ async function fetchScene(url, site, shallowRelease) {
     return scrapeScene(res.body.toString(), url, site, shallowRelease);
 }
 
+async function fetchProfile(actorName) {
+    const actorSlug = actorName.toLowerCase().replace(/\s+/g, '-');
+    const res = await bhttp.get(`https://www.kellymadison.com/models/${actorSlug}`, {
+        headers: {
+            'X-Requested-With': 'XMLHttpRequest',
+        },
+    });
+
+    if (res.statusCode === 200) {
+        return scrapeProfile(res.body.toString(), actorName);
+    }
+
+    return null;
+}
+
 module.exports = {
     fetchLatest,
+    fetchProfile,
     fetchScene,
 };

src/scrapers/scrapers.js

@@ -6,7 +6,6 @@ const bangbros = require('./bangbros');
 const blowpass = require('./blowpass');
 const dogfart = require('./dogfart');
 const evilangel = require('./evilangel');
-const kellymadison = require('./kellymadison');
 const kink = require('./kink');
 const mikeadriano = require('./mikeadriano');
 const mofos = require('./mofos');
@@ -20,6 +19,7 @@ const vixen = require('./vixen');
 const ddfnetwork = require('./ddfnetwork');
 const brazzers = require('./brazzers');
 const julesjordan = require('./julesjordan');
+const kellymadison = require('./kellymadison');
 const legalporno = require('./legalporno');
 const xempire = require('./xempire');
 
@@ -55,6 +55,7 @@ module.exports = {
         xempire,
         brazzers,
         freeones,
+        kellymadison,
         julesjordan,
         legalporno,
         pornhub,