add support for shorts urls, channel handles, channel /c/ and /user/

src/router/websocket.ts

@@ -128,6 +128,7 @@ app.ws('/savechannel', {
 
     const channelId = await validateChannel(ws.data.query.url);
     if (!channelId) return sendError(ws, 'Invalid channel URL.');
+    if (typeof channelId !== 'string') return sendError(ws, `Failed to fetch channel ID - ${channelId.error}`)
 
     ws.send('DATA - This process is automatic. Your video will start archiving shortly.')
     ws.send('CAPTCHA - Solving a cryptographic challenge before downloading.')

src/utils/regex.ts

@@ -1,35 +1,29 @@
 function validateVideo(input: string): string | false {
   try {
     const url = new URL(input);
+    let videoId: string = ''
     const hostnames = [
       'youtube.com',
       'www.youtube.com',
       'm.youtube.com'
     ];
 
-    // basic hostname check
     if (hostnames.includes(url.hostname)) {
-      // basic url
       if (url.pathname === '/watch') {
-        const videoId = url.searchParams.get('v');
+        if (!url.searchParams.get('v')) return false
-        return videoId || false;
+        videoId = url.searchParams.get('v')!
-      }
+      } else if (url.pathname.startsWith('/shorts/')) {
-
+        videoId = url.pathname.replace('/shorts/', '')
-      // embed url
+      } else return false
-      const embedMatch = url.pathname.match(/^\/embed\/([a-zA-Z0-9_-]+)/);
+      // removed - embed url
-      if (embedMatch) {
-        return embedMatch[1];
-      }
-
-      return false;
     }
 
     // short urls
     if (url.hostname === 'youtu.be') {
-      const videoId = url.pathname.replace(/^\//, '');
+      videoId = url.pathname.replace(/^\//, '');
-      return videoId || false;
     }
 
+    if (videoId && videoId.match(/[\w\-_]{11}/)) return videoId
     return false;
   } catch {
     return false;
@@ -59,7 +53,7 @@ function validatePlaylist(input: string): string | false {
   }
 }
 
-async function validateChannel(input: string): Promise<string | false> {
+async function validateChannel(input: string): Promise<string | { error: string; } | false> {
   try {
     const url = new URL(input);
     const hostnames = [
@@ -67,19 +61,43 @@ async function validateChannel(input: string): Promise<string | false> {
       'www.youtube.com', 
       'm.youtube.com'
     ];
+    const whereIsIt: Record<string, (string|number)[]> = {
+      channel: ['metadata', 'channelMetadataRenderer', 'externalId'],
+      handle: ['responseContext', 'serviceTrackingParams', 0, 'params'],
+      user: ['metadata', 'channelMetadataRenderer', 'externalId']
+    }
 
     if (hostnames.includes(url.hostname)) {
-      // @ urls
+      let whatIsIt = ''
-      const atMatch = url.pathname.match(/^\/@([a-zA-Z0-9.-]+)/);
+
-      if (atMatch) {
+      // many thanks to Benjamin Loison (@Benjamin-Loison) for his PHP implementation of this
-        const channelId = await (await fetch(`https://yt.jaybee.digital/api/channels?part=channels&handle=${atMatch[1]}`)).json()
+      // https://github.com/Benjamin-Loison/YouTube-operational-API/blob/main/channels.php
-        return channelId['items'][0]['id']
+      // and the stackoverflow answer with all the possible options, thank you.
+      // https://stackoverflow.com/a/75843807
+
+      if (url.pathname.startsWith('/channel/')) { // /channel/[id]
+        return url.pathname.match(/UC[\w\-_]{22}/gm)?.[0] || false
+      } else if (url.pathname.startsWith('/c/')) { // /c/[custom]
+        whatIsIt = 'channel'
+      } else if (url.pathname.startsWith('/user/')) {
+        whatIsIt = 'user'
+      } else if (url.pathname.match(/@[\w\-_.]{3,}/gm)) { // /@[handle]
+        whatIsIt = 'handle'
+      } else return false 
+
+      const channelReq = await fetch(`${process.env.METADATA}/getWebpageJson?url=${url}`)
+      if (!channelReq.ok) return {
+        error: `Failed to fetch Youtube with status ${channelReq.status}. Please retry.`
       }
 
-      // /channel/ and /c/
+      const channelJson = await channelReq.json()
-      const channelMatch = url.pathname.match(/^\/(channel|c)\/([a-zA-Z0-9_-]+)/);
+      let channelId: string | Record<any, any> = getByPath(channelJson, whereIsIt[whatIsIt]!);
-      if (channelMatch) {
+      if (whatIsIt == 'handle') {
-        return channelMatch[2];
+        channelId = channelId.find((c:any) => c.key == 'browse_id').value
+      }
+
+      return typeof channelId == 'string' ? channelId : {
+        error: 'Failed to extract channel ID from the Youtube provided JSON.'
       }
     }
 
@@ -89,4 +107,11 @@ async function validateChannel(input: string): Promise<string | false> {
   }
 }
 
+function getByPath(obj: Record<any, any>, path: (string|number)[]) {
+  return path.reduce((acc, key) => {
+    if (acc === undefined || acc === null) return undefined
+    return acc[key]
+  }, obj)
+}
+
 export { validateVideo, validatePlaylist, validateChannel }