Merge pull request #2 from karl0ss/linksFromPage

get links from source page

.gitignore

@@ -1,4 +1,3 @@
-.env
 node_modules
 releasebb.json
 config.json

FeedFilter.js

@@ -0,0 +1,53 @@
+const fs = require('fs')
+const { linkAdder } = require('./JDLinkAdder');
+const { getLinksFromURL } = require('./LinkGrabber')
+const log = require('simple-node-logger').createSimpleLogger('jdrssdownloader.log');
+
+async function filterFeed(fileName) {
+    let myshowlist = JSON.parse(fs.readFileSync('config.json')).Shows
+    let feed = JSON.parse(fs.readFileSync(fileName));
+
+    myshowlist.forEach(async show => {
+        try {
+            // Find show on feed
+            let list_filtered_for_show = feed.filter(item => item.title.includes(show.Name))
+            if (list_filtered_for_show.length > 0) {
+                // If show is found get url then return all links on that page
+                let full_link_list_from_page = await getLinksFromURL(list_filtered_for_show[0].link)
+                // Only get urls with HEVC in name
+                let urls_with_HEVC_in_url = full_link_list_from_page.filter(item => item.includes('HEVC'))
+                if (urls_with_HEVC_in_url.length == 0) {
+                    // If no urls with HEVC check for H265
+                    urls_with_HEVC_in_url = full_link_list_from_page.filter(item => item.includes('H265'))
+                }
+                // Only keep urls that match show quality
+                let urls_with_quality_in_url = urls_with_HEVC_in_url.filter(item => item.includes(show.Quality))
+                // Remove any url trying to direct to a torrent site search
+                let urls_without_torrent_in_url = urls_with_quality_in_url.filter(item => !item.includes('torrent'))
+                // Remove any url that doesn't include MeGusta
+                let only_MeGusta_links = urls_without_torrent_in_url.filter(item => item.includes('MeGusta'))
+                // NitroFlare doesn't group with the rest of the links in JD, remove them.
+                let remove_nitroflare = only_MeGusta_links.filter(item => !item.includes('nitro'))
+                // Send Links to JDdownloader
+                if (remove_nitroflare.length !== 0) {
+                    log.info(remove_nitroflare.length + ' links for ' + show.Name + ' have been sent to JDdownloader')
+                    linkAdder(remove_nitroflare)
+                } else {
+                    // No HEVC links found
+                    log.info(remove_nitroflare.length + ' HEVC links for ' + show.Name + ' have been found')
+                }
+
+            } else {
+                // Show not found on the current feed cache
+                log.info(show.Name + ' not on feed')
+            }
+        } catch (error) {
+            log.error('Something went wrong ' + error)
+        }
+    })
+}
+
+module.exports = {
+    filterFeed
+}
+

LinkGrabber.js

@@ -0,0 +1,26 @@
+const axios = require('axios');
+var cheerio = require('cheerio');
+
+async function getLinksFromURL(url) {
+
+    try {
+        let links = [];
+        let httpResponse = await axios.get(url);
+
+        let $ = cheerio.load(httpResponse.data);
+        let linkObjects = $('a'); // get all hyperlinks
+
+        linkObjects.each((index, element) => {
+            links.push(
+                $(element).attr('href'), // get the href attribute
+            );
+        });
+
+        return links;
+    } catch (e) { console.log(e) }
+
+}
+
+module.exports = {
+    getLinksFromURL
+}

feed.js

@@ -1,29 +0,0 @@
-const extractUrls = require("extract-urls");
-const fs = require('fs')
-const { linkAdder } = require('./jd-link-adder');
-const log = require('simple-node-logger').createSimpleLogger('jdrssdownloader.log');
-
-function filterFeed(fileName) {
-    let myshowlist = JSON.parse(fs.readFileSync('config.json')).Shows
-    let feed = JSON.parse(fs.readFileSync(fileName));
-
-    myshowlist.forEach(show => {
-        try {
-            let list_filtered_for_show = feed.filter(item => item.title.includes(show.Name))
-            let extracted_urls_for_show = extractUrls(list_filtered_for_show[0]["content:encoded"]);
-            let urls_with_HEVC_in_url = extracted_urls_for_show.filter(item => item.includes('HEVC') || item.includes('H265'))
-            let urls_with_quality_in_url = urls_with_HEVC_in_url.filter(item => item.includes(show.Quality))
-            let urls_without_torrent_in_url = urls_with_quality_in_url.filter(item => !item.includes('torrent'))
-            log.info(show.Name + ' - ' + urls_without_torrent_in_url)
-            linkAdder(urls_without_torrent_in_url)
-            // console.log(urls_without_torrent_in_url)
-        } catch (error) {
-            log.info(show.Name + ' not on feed')
-        }
-    })
-}
-
-module.exports = {
-    filterFeed
-}
-

main.js

@@ -1,11 +1,11 @@
 // Import dependencies
 const fs = require("fs");
 const Parser = require("rss-parser");
-const { filterFeed } = require("./feed");
+const { filterFeed } = require("./FeedFilter");
 const lodash = require('lodash');
 const log = require('simple-node-logger').createSimpleLogger('jdrssdownloader.log');
 
-    (async function main() {
+(async function main() {
 
 
     // Make a new RSS Parser
@@ -22,6 +22,7 @@ const log = require('simple-node-logger').createSimpleLogger('jdrssdownloader.lo
     if (fs.existsSync(fileName)) {
         items = require(`./${fileName}`);
     }
+    // Compare existing cache and new items and merge differences
     let updatedArray = lodash.unionBy(feed.items, items, 'title');
 
     // Save the file
@@ -30,4 +31,4 @@ const log = require('simple-node-logger').createSimpleLogger('jdrssdownloader.lo
 
     // run next part
     filterFeed(fileName)
-    })();
+})();

package.json

@@ -9,6 +9,8 @@
   "author": "Karl0ss",
   "license": "ISC",
   "dependencies": {
+    "axios": "^0.27.2",
+    "cheerio": "^1.0.0-rc.11",
     "extract-urls": "^1.3.2",
     "jdownloader-client": "^1.0.0",
     "lodash": "^4.17.21",