Added avatars. Added PornHub and LegalPorno actor profile scrapers.

src/scrapers/freeones.js

@@ -7,7 +7,7 @@ const moment = require('moment');
 
 const knex = require('../knex');
 
-async function scrapeActorFrontpage(html, url, name) {
+async function scrapeProfileFrontpage(html, url, name) {
     const { document } = new JSDOM(html).window;
     const bioEl = document.querySelector('.dashboard-bio-list');
 
@@ -26,7 +26,6 @@ async function scrapeActorFrontpage(html, url, name) {
     const boobsSizeString = bio['Measurements:'];
     const boobsSize = boobsSizeString === '??-??-??' ? null : boobsSizeString;
     const boobsNatural = bio['Fake Boobs:'] === 'No';
-    const active = bio['Career Status:'].trim() === 'Active';
 
     const residenceCountryName = bio['Country of Origin:'];
     const countryEntry = await knex('countries').where({ name: residenceCountryName }).first();
@@ -59,14 +58,13 @@ async function scrapeActorFrontpage(html, url, name) {
             eyes,
             piercings,
             tattoos,
-            active,
             social,
         },
         url: bioUrl,
     };
 }
 
-async function scrapeActorBio(html, frontpageBio, url, name) {
+async function scrapeProfileBio(html, frontpageBio, url, name) {
     const { document } = new JSDOM(html).window;
     const bioEl = document.querySelector('#biographyTable');
 
@@ -92,6 +90,8 @@ async function scrapeActorBio(html, frontpageBio, url, name) {
 
     const hair = bio['Hair Color:'].toLowerCase();
     const eyes = bio['Eye Color:'].toLowerCase();
+    const height = Number(bio['Height:'].match(/\d+/)[0]);
+    const weight = Number(bio['Weight:'].match(/\d+/)[0]);
 
     const piercingsString = bio['Piercings:'];
     const piercings = piercingsString === 'None' ? null : piercingsString;
@@ -113,6 +113,8 @@ async function scrapeActorBio(html, frontpageBio, url, name) {
             size: boobsSize,
             natural: boobsNatural,
         },
+        height,
+        weight,
         hair,
         eyes,
         piercings,
@@ -121,23 +123,34 @@ async function scrapeActorBio(html, frontpageBio, url, name) {
     };
 }
 
-async function fetchActor(actorName) {
+async function fetchProfile(actorName) {
     const slug = actorName.replace(' ', '_');
-    const frontpageUrl = `https://freeones.com/html/v_links/${slug}`;
+    const frontpageUrl = `https://www.freeones.com/html/v_links/${slug}`;
 
     const resFrontpage = await bhttp.get(frontpageUrl);
 
     if (resFrontpage.statusCode === 200) {
-        const { url, bio } = await scrapeActorFrontpage(resFrontpage.body.toString(), frontpageUrl, actorName);
-
+        const { url, bio } = await scrapeProfileFrontpage(resFrontpage.body.toString(), frontpageUrl, actorName);
         const resBio = await bhttp.get(url);
 
-        return scrapeActorBio(resBio.body.toString(), bio, url, actorName);
+        return scrapeProfileBio(resBio.body.toString(), bio, url, actorName);
+    }
+
+    // apparently some actors are appended 'Babe' as their surname...
+    const fallbackSlug = `${slug}_Babe`;
+    const fallbackUrl = `https://www.freeones.com/html/s_links/${fallbackSlug}`;
+    const resFallback = await bhttp.get(fallbackUrl);
+
+    if (resFallback.statusCode === 200) {
+        const { url, bio } = await scrapeProfileFrontpage(resFallback.body.toString(), fallbackUrl, actorName);
+        const resBio = await bhttp.get(url);
+
+        return scrapeProfileBio(resBio.body.toString(), bio, url, actorName);
     }
 
     return null;
 }
 
 module.exports = {
-    fetchActor,
+    fetchProfile,
 };