Fixed detection if youtube video is music or not

README.md

@@ -96,7 +96,8 @@ and set your `User`, `Group`, `ExecStart` and `WorkingDirectory` accordingly.
 Copy `.env.EXAMPLE` to `.env` and add your `YOUTUBE_API_KEY` and `ODESLI_API_KEY`.
 To obtain one follow [YouTube's guide](https://developers.google.com/youtube/registering_an_application) to create an
 _API key_.
-If `YOUTUBE_API_KEY` is unset, no playlist will be updated.
+If `YOUTUBE_API_KEY` is unset, no playlist will be updated. Also, _all_ YouTube links will be treated as music videos,
+because the API is the only way to check if a YouTube link leads to music or something else.
 
 If `ODESLI_API_KEY` is unset, your rate limit to the song.link API will be lower.
 

src/lib/server/timeline.ts

@@ -1,5 +1,10 @@
-import { HASHTAG_FILTER, MASTODON_INSTANCE, ODESLI_API_KEY } from '$env/static/private';
-import { log } from '$lib/log';
+import {
+  HASHTAG_FILTER,
+  MASTODON_INSTANCE,
+  ODESLI_API_KEY,
+  YOUTUBE_API_KEY
+} from '$env/static/private';
+import { enableVerboseLog, log } from '$lib/log';
 import type {
   Account,
   AccountAvatar,
@@ -27,10 +32,73 @@ import sharp from 'sharp';
 import { WebSocket } from 'ws';
 
 const URL_REGEX = new RegExp(/href="(?<postUrl>[^>]+?)" target="_blank"/gm);
+const INVIDIOUS_REGEX = new RegExp(/invidious.*?watch.*?v=(?<videoId>[a-zA-Z_0-9-]+)/gm);
+const YOUTUBE_REGEX = new RegExp(
+  /https?:\/\/(www\.)?youtu((be.com\/.*?v=)|(\.be\/))(?<videoId>[a-zA-Z_0-9-]+)/gm
+);
 
 export class TimelineReader {
   private static _instance: TimelineReader;
 
+  private static async isMusicVideo(videoId: string) {
+    if (!YOUTUBE_API_KEY || YOUTUBE_API_KEY === 'CHANGE_ME') {
+      // Assume that it *is* a music link when no YT API key is provided
+      return true;
+    }
+    const searchParams = new URLSearchParams([
+      ['part', 'snippet'],
+      ['id', videoId],
+      ['key', YOUTUBE_API_KEY]
+    ]);
+    const youtubeVideoUrl = new URL(`https://www.googleapis.com/youtube/v3/videos?${searchParams}`);
+    const resp = await fetch(youtubeVideoUrl);
+    const respObj = await resp.json();
+    if (!respObj.items.length) {
+      console.warn('Could not find video with id', videoId);
+      return false;
+    }
+
+    const item = respObj.items[0];
+    if (!item.snippet) {
+      console.warn('Could not load snippet for video', videoId, item);
+      return false;
+    }
+    if (item.snippet.tags?.includes('music')) {
+      return true;
+    }
+
+    const categorySearchParams = new URLSearchParams([
+      ['part', 'snippet'],
+      ['id', item.snippet.categoryId],
+      ['key', YOUTUBE_API_KEY]
+    ]);
+    const youtubeCategoryUrl = new URL(
+      `https://www.googleapis.com/youtube/v3/videoCategories?${categorySearchParams}`
+    );
+    const categoryTitle: string = await fetch(youtubeCategoryUrl)
+      .then((r) => r.json())
+      .then((r) => r.items[0]?.snippet?.title);
+    if (enableVerboseLog) {
+      log.verbose(
+        'Video',
+        videoId,
+        'category',
+        categoryTitle,
+        'tags',
+        item.snippet.tags,
+        'category id',
+        item.snippet.categoryId,
+        'response',
+        respObj,
+        'snippet',
+        item.snippet
+      );
+    } else {
+      log.debug('Video', videoId, 'category', categoryTitle);
+    }
+    return categoryTitle === 'Music';
+  }
+
   public static async getSongInfoInPost(post: Post): Promise<SongInfo[]> {
     const urlMatches = post.content.matchAll(URL_REGEX);
     const songs: SongInfo[] = [];
@@ -71,8 +139,12 @@ export class TimelineReader {
       return null;
     }
 
+    const videoId = INVIDIOUS_REGEX.exec(url.href)?.groups?.videoId;
+    const urlString =
+      videoId !== undefined ? `https://youtube.com/watch?v=${videoId}` : url.toString();
+
     const odesliParams = new URLSearchParams();
-    odesliParams.append('url', url.toString());
+    odesliParams.append('url', urlString);
     odesliParams.append('userCountry', 'DE');
     odesliParams.append('songIfSingle', 'true');
     if (ODESLI_API_KEY && ODESLI_API_KEY !== 'CHANGE_ME') {
@@ -80,23 +152,39 @@ export class TimelineReader {
     }
     const odesliApiUrl = `https://api.song.link/v1-alpha.1/links?${odesliParams}`;
     try {
-      return fetch(odesliApiUrl).then(async (response) => {
-        if (response.status === 429) {
-          throw new Error('Rate limit reached', { cause: 429 });
-        }
-        const odesliInfo: OdesliResponse = await response.json();
-        if (!odesliInfo || !odesliInfo.entitiesByUniqueId || !odesliInfo.entityUniqueId) {
+      const response = await fetch(odesliApiUrl);
+      log.debug('received odesli response', response.status);
+      if (response.status === 429) {
+        throw new Error('Rate limit reached', { cause: 429 });
+      }
+      const odesliInfo: OdesliResponse = await response.json();
+      if (!odesliInfo || !odesliInfo.entitiesByUniqueId || !odesliInfo.entityUniqueId) {
+        return null;
+      }
+      const info = odesliInfo.entitiesByUniqueId[odesliInfo.entityUniqueId];
+      const platform: Platform = 'youtube';
+      log.debug(url, 'odesli response', info, 'YT URL', odesliInfo.linksByPlatform[platform]?.url);
+      if (info.platforms.includes(platform)) {
+        let youtubeId =
+          videoId ??
+          YOUTUBE_REGEX.exec(url.href)?.groups?.videoId ??
+          new URL(odesliInfo.pageUrl).pathname.split('/y/').pop();
+        if (youtubeId === undefined) {
+          log.warn('Looks like a youtube video, but could not extract a video id', url, odesliInfo);
           return null;
         }
-        const info = odesliInfo.entitiesByUniqueId[odesliInfo.entityUniqueId];
-        const platform: Platform = 'youtube';
-        return {
-          ...info,
-          pageUrl: odesliInfo.pageUrl,
-          youtubeUrl: odesliInfo.linksByPlatform[platform]?.url,
-          postedUrl: url.toString()
-        } as SongInfo;
-      });
+        const isMusic = await TimelineReader.isMusicVideo(youtubeId);
+        if (!isMusic) {
+          log.debug('Probably not a music video', url, odesliInfo);
+          return null;
+        }
+      }
+      return {
+        ...info,
+        pageUrl: odesliInfo.pageUrl,
+        youtubeUrl: odesliInfo.linksByPlatform[platform]?.url,
+        postedUrl: url.toString()
+      } as SongInfo;
     } catch (e) {
       if (e instanceof Error && e.cause === 429) {
         log.warn('song.link rate limit reached. Trying again in 10 seconds');