Added media to LegalPorno scraper.

assets/css/_theme.scss

@@ -1,4 +1,5 @@
-$primary: #ff886c;
+/* $primary: #ff886c; */
+$primary: #ff6c88;
 
 $text: #222;
 $text-contrast: #fff;

public/css/style.css

@@ -1,3 +1,4 @@
+/* $primary: #ff886c; */
 .filters-bar[data-v-5533e378] {
   display: block;
   background: rgba(0, 0, 0, 0.1);
@@ -27,7 +28,7 @@
 }
 .filters .toggle.active[data-v-5533e378] {
       color: #fff;
-      background: #ff886c;
+      background: #ff6c88;
 }
 .filter[data-v-5533e378] {
   display: inline-block;
@@ -142,6 +143,7 @@
   width: 300px;
 }
 
+/* $primary: #ff886c; */
 .banner[data-v-2bc41e74] {
   background: #222;
   white-space: nowrap;
@@ -182,6 +184,7 @@
     width: .6rem;
 }
 
+/* $primary: #ff886c; */
 .noselect {
   user-select: none;
   -webkit-user-select: none;
@@ -206,12 +209,13 @@ body {
     display: inline-block; }
 
 .heading {
-  color: #ff886c;
+  color: #ff6c88;
   margin: 0 0 1rem 0; }
 
+/* $primary: #ff886c; */
 .header[data-v-10b7ec04] {
   color: #fff;
-  background: #ff886c;
+  background: #ff6c88;
   padding: 1rem;
 }
 .logo-link[data-v-10b7ec04] {
@@ -241,6 +245,7 @@ body {
   overflow-y: auto;
 }
 
+/* $primary: #ff886c; */
 .icon {
   fill: #222;
   display: inline-block;

src/fetch-releases.js

@@ -142,10 +142,11 @@ async function storePhotos(release, releaseEntry) {
 async function storePoster(release, releaseEntry) {
     console.log(`Storing poster for (${release.site.name}, ${releaseEntry.id}) "${release.title}"`);
 
-    const { pathname } = new URL(release.poster);
-    const mimetype = mime.getType(pathname);
-
     const res = await bhttp.get(release.poster);
+
+    const { pathname } = new URL(release.poster);
+    const mimetype = res.headers['content-type'] || mime.getType(pathname) || 'image/jpeg';
+
     const filepath = path.join(release.site.slug, releaseEntry.id.toString(), `poster.${mime.getExtension(mimetype)}`);
     await fs.writeFile(path.join(config.photoPath, filepath), res.body);
 

src/scrapers/legalporno.js

@@ -15,6 +15,28 @@ function extractTitle(originalTitle) {
     return { shootId, title };
 }
 
+function getPoster(posterElement, sceneId) {
+    const posterStyle = posterElement.attr('style');
+
+    if (posterStyle) {
+        return posterStyle.slice(posterStyle.indexOf('(') + 1, -1);
+    }
+
+    const posterRange = posterElement.attr('data-casting');
+    const posterRangeData = posterRange ? JSON.parse(posterRange) : null;
+    const posterTimeRange = posterRangeData[Math.floor(Math.random() * posterRangeData.length)];
+
+    if (typeof posterTimeRange === 'number') {
+        // poster time is already a single time value
+        return `https://legalporno.com/casting/${sceneId}/${posterTimeRange}`;
+    }
+
+    const [max, min] = posterTimeRange.split('-');
+    const posterTime = Math.floor(Math.random() * (Number(max) - Number(min) + 1) + Number(min));
+
+    return `https://legalporno.com/casting/${sceneId}/${posterTime}`;
+}
+
 function scrapeLatest(html, site) {
     const $ = cheerio.load(html, { normalizeWhitespace: true });
     const scenesElements = $('.thumbnails > div').toArray();
@@ -29,19 +51,27 @@ function scrapeLatest(html, site) {
 
         const date = moment.utc($(element).attr('release'), 'YYYY/MM/DD').toDate();
 
+        const sceneId = $(element).attr('data-content');
+        const posterElement = $(element).find('.thumbnail-avatar');
+
+        const poster = getPoster(posterElement, sceneId);
+
         return {
             url,
             shootId,
             entryId,
             title,
             date,
+            poster,
             site,
         };
     });
 }
 
-async function scrapeScene(html, url, site) {
+async function scrapeScene(html, url, site, useGallery) {
     const $ = cheerio.load(html, { normalizeWhitespace: true });
+    const playerObject = $('script:contains("new VideoPlayer")').html();
+    const data = JSON.parse(playerObject.slice(playerObject.indexOf('{"swf":'), playerObject.indexOf('} );') + 1));
 
     const originalTitle = $('h1.watchpage-title').text().trim();
     const { shootId, title } = extractTitle(originalTitle);
@@ -56,6 +86,15 @@ async function scrapeScene(html, url, site) {
 
     const duration = moment.duration($('span[title="Runtime"]').text().trim()).asSeconds();
 
+    const posterStyle = $('#player').attr('style');
+    const poster = posterStyle.slice(posterStyle.indexOf('(') + 1, -1);
+
+    const photos = useGallery
+        ? $('.gallery a img').map((photoIndex, photoElement) => $(photoElement).attr('src')).toArray()
+        : $('.screenshots img').map((photoIndex, photoElement) => $(photoElement).attr('src')).toArray();
+
+    const trailer = data.clip.qualities.find(clip => clip.quality === 'vga' || clip.quality === 'hd');
+
     const rawTags = $(tagsElement).find('a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
     const tags = await matchTags(rawTags);
 
@@ -67,6 +106,13 @@ async function scrapeScene(html, url, site) {
         date,
         actors,
         duration,
+        poster,
+        photos,
+        trailer: {
+            src: trailer.src,
+            type: trailer.type,
+            quality: trailer.quality === 'vga' ? 480 : 720,
+        },
         tags,
         site,
     };
@@ -79,9 +125,13 @@ async function fetchLatest(site, page = 1) {
 }
 
 async function fetchScene(url, site) {
-    const res = await bhttp.get(url);
+    const useGallery = true;
 
-    return scrapeScene(res.body.toString(), url, site);
+    const res = useGallery
+        ? await bhttp.get(`${url}/gallery#gallery`)
+        : await bhttp.get(`${url}/screenshots#screenshots`);
+
+    return scrapeScene(res.body.toString(), url, site, useGallery);
 }
 
 module.exports = {