Added experimental youtube url scraping support

Author: Azarattum

apps/core/event/pool.ts

@@ -43,10 +43,15 @@ const relate = pool<
   | ((type: "artist", to: ArtistInfo, page: number) => Aggregated<ArtistInfo>)
 >("relate", { transform: aggregate, timeout, rate });
 
+const scrape = pool<(url: string, page: number) => Aggregated<TrackInfo>>(
+  "scrape",
+  { transform: aggregate, timeout },
+);
+
 const recognize = pool<
   (
     stream: () => ReadableStream<Uint8Array>,
-    page: number
+    page: number,
   ) => Aggregated<TrackInfo>
 >("recognize", { transform: aggregate, timeout });
 
@@ -68,6 +73,7 @@ export {
   search,
   relate,
   expand,
+  scrape,
   desource,
   recognize,
   transcribe,

plugins/telegram/handlers/message.ts

@@ -4,6 +4,7 @@ import {
   message,
   search,
   lookup,
+  scrape,
   voice,
   fetch,
   info,
@@ -29,6 +30,20 @@ message(function* (text) {
   const page = pages.get(id);
   if (!page) return;
 
+  if (text.match(/^https?:\/\//)) {
+    const reply = this.reply.bind(this);
+    yield* map(scrape(text, 8), function* (state) {
+      if (state.completed && state.items.length === 1) {
+        page.close();
+        state.close();
+        yield* reply(state.items);
+      } else {
+        yield* page.update(state);
+      }
+    });
+    return;
+  }
+
   this.signal.addEventListener("abort", page.close, { once: true });
   yield* map(search("track", text, 8), function* (state) {
     yield* page.update(state);

plugins/telegram/plugin.ts

@@ -23,6 +23,7 @@ export const {
   fetch,
   users,
   search,
+  scrape,
   relate,
   expand,
   lookup,
@@ -40,7 +41,7 @@ export const {
         token: defaulted(string(), ""),
         webhook: defaulted(string(), ""),
       }),
-      {}
+      {},
     ),
   },
   settings: {
@@ -70,11 +71,11 @@ const command = pool<(command: string, replied?: number) => void>("command");
 const changed = pool<(entries: number[]) => void>("changed");
 const callback =
   pool<(request: Infer<typeof query>, message: number, chat: number) => void>(
-    "callback"
+    "callback",
   );
 
 update.catch((error: any) =>
-  wrn(error.cause?.message || error.cause || error.message)
+  wrn(error.cause?.message || error.cause || error.message),
 );
 
 export {

plugins/youtube/index.ts

@@ -5,6 +5,7 @@ import {
   init,
   lookup,
   relate,
+  scrape,
   search,
   transcribe,
 } from "./plugin";
@@ -15,7 +16,7 @@ import { convert } from "./types";
 init(function* () {
   const loadMessage = setTimeout(
     () => info("Retrieving player script..."),
-    1000
+    1000,
   );
 
   const youtube = yield* async(Innertube.create());
@@ -46,7 +47,7 @@ desource(function* (track) {
     this.youtube.music
       .getInfo(id)
       .then((x) => x.chooseFormat({ type: "audio", quality: "best" }))
-      .then((x) => x.decipher(this.youtube.player))
+      .then((x) => x.decipher(this.youtube.player)),
   );
 });
 
@@ -65,12 +66,12 @@ expand(function* (type, what, _) {
       : this.youtube.music
           .getAlbum(id)
           .then((x) => x.url?.match(/list=([^&]+)/)?.[1])
-          .catch(() => undefined)
+          .catch(() => undefined),
   );
   if (!playlist) return;
 
   let result = yield* async(
-    this.youtube.music.getPlaylist(playlist).catch(() => undefined)
+    this.youtube.music.getPlaylist(playlist).catch(() => undefined),
   );
   while (result) {
     yield* convert(result.items as any, "track");
@@ -90,8 +91,8 @@ relate(function* (type, to, _) {
           x.sections
             .find((x: any) => x.header?.title?.text === "Fans might also like")
             ?.contents.filter((x) => x.type === "MusicTwoRowItem"),
-        () => undefined
-      )
+        () => undefined,
+      ),
     );
     return yield* convert(artists as any, "artist");
   }
@@ -101,11 +102,19 @@ relate(function* (type, to, _) {
       .getRelated(id)
       .then((x) => x.find((x) => x.type === "MusicCarouselShelf"))
       .then((x: any) => x?.contents)
-      .catch(() => undefined)
+      .catch(() => undefined),
   );
   yield* convert(items, type);
 });
 
+scrape(function* (url) {
+  const id = url.match(/youtu(be.com|.be)\/(watch\?v=)?([a-zA-Z0-9_-]+)/)?.[3];
+  if (!id) return;
+
+  const info = yield* async(this.youtube.music.getInfo(id));
+  yield* convert([info["basic_info"]], "track");
+});
+
 transcribe(function* (track) {
   const id = yield* identify(track, "track");
   if (!id) return;
@@ -114,14 +123,14 @@ transcribe(function* (track) {
     this.youtube.music
       .getLyrics(id)
       .then((x) => x?.description.toString())
-      .catch(() => undefined)
+      .catch(() => undefined),
   );
   if (lyrics) yield lyrics;
 });
 
 function* identify(
   data: { sources?: string[]; title?: string },
-  type: "track" | "artist" | "album"
+  type: "track" | "artist" | "album",
 ) {
   return (
     data.sources?.find((x) => x.startsWith("youtube/"))?.slice(8) ||

plugins/youtube/plugin.ts

@@ -8,6 +8,7 @@ export const {
   search,
   relate,
   expand,
+  scrape,
   lookup,
   desource,
   transcribe,

plugins/youtube/types.ts

@@ -1,13 +1,10 @@
-import type {
-  MusicResponsiveListItem,
-  MusicTwoRowItem,
-} from "youtubei.js/dist/src/parser/nodes";
+import type { YTMusic } from "youtubei.js";
 
-function toAssets(url?: string) {
+function toAssets(url?: string, secondary = url) {
   if (!url) return { arts: [], thumbnails: [] };
   return {
     arts: [url.replace(/w[0-9]+-h[0-9]+/, "w800-h800")],
-    thumbnails: [url.replace(/w[0-9]+-h[0-9]+/, "w120-h120")],
+    thumbnails: [secondary!.replace(/w[0-9]+-h[0-9]+/, "w120-h120")],
   };
 }
 
@@ -35,7 +32,7 @@ function toAlbum(data: MusicResponsiveListItem) {
         (x: any) =>
           x.endpoint?.payload?.browseEndpointContextSupportedConfigs
             ?.browseEndpointContextMusicConfig?.pageType ===
-            "MUSIC_PAGE_TYPE_ARTIST" && x.endpoint?.payload?.browseId
+            "MUSIC_PAGE_TYPE_ARTIST" && x.endpoint?.payload?.browseId,
       )
       .map((x: any) => ({
         title: x.text as string,
@@ -44,37 +41,55 @@ function toAlbum(data: MusicResponsiveListItem) {
   };
 }
 
-function toTrack(data: MusicResponsiveListItem) {
-  if (!data.artists?.length) return;
+function toTrack(data: MusicResponsiveListItem & TrackInfo) {
+  if (!data.artists?.length && !data.view_count) return;
 
   return {
     title: data.title + (data.subtitle ? ` (${data.subtitle})` : ""),
-    duration: data.duration?.seconds || 0,
+    duration: data.duration?.seconds || +data.duration! || 0,
     sources: [`youtube/${data.id}`],
     album: {
       title: data.album?.name || data.title || "Unknown",
       sources: data.album?.id ? [`youtube/${data.album?.id}`] : [],
       year: 0,
-      ...toAssets(data.thumbnail?.contents[0].url),
+      ...toAssets(
+        data.thumbnail?.contents?.[0].url || data.thumbnail?.[0].url,
+        data.thumbnail?.[data.thumbnail.length - 1]?.url,
+      ),
     },
     artists:
       data.artists?.map((x) => ({
         title: x.name,
         sources: x.channel_id ? [`youtube/${x.channel_id}`] : [],
-      })) || [],
+      })) ||
+      (typeof data.author === "string"
+        ? [
+            {
+              title: data.author.toString(),
+              sources: data.channel_id ? [`youtube/${data.channel_id}`] : [],
+            },
+          ]
+        : []),
   };
 }
 
 function convert<T extends "track" | "artist" | "album">(
-  data: MusicResponsiveListItem[] | undefined,
-  type: T
+  data: (MusicResponsiveListItem | TrackInfo)[] | undefined,
+  type: T,
 ) {
   const map = { track: toTrack, artist: toArtist, album: toAlbum }[type];
   const truthy = <T>(x: T): x is NonNullable<T> => !!x;
   const convert = map as (
-    x: Parameters<typeof map>[0]
+    x: Parameters<typeof map>[0],
   ) => ReturnType<typeof map>;
-  return data?.map((x) => convert(x)).filter(truthy) || [];
+  return data?.map((x: any) => convert(x)).filter(truthy) || [];
 }
 
+type TrackInfo = YTMusic.TrackInfo["basic_info"];
+type MusicResponsiveListItem = YTMusic.Album["contents"][number];
+type MusicTwoRowItem = Extract<
+  YTMusic.Album["sections"][number]["contents"][number],
+  { id: string | undefined }
+>;
+
 export { convert };