feat(editor): audio block (#10947)

AudioMedia entity for loading & controlling a single audio media
AudioMediaManagerService: Global audio state synchronization across tabs
AudioAttachmentService + AudioAttachmentBlock for manipulating AttachmentBlock in affine - e.g., filling transcription (using mock endpoint for now)
Added AudioBlock + AudioPlayer for rendering audio block in affine (new transcription block whose renderer is provided in affine)

fix AF-2292
fix AF-2337

packages/common/infra/src/index.ts

@@ -2,6 +2,7 @@ export * from './app-config-storage';
 export * from './atom';
 export * from './framework';
 export * from './livedata';
+export * from './media';
 export * from './orm';
 export * from './storage';
 export * from './sync';

packages/common/infra/src/media/index.ts

@@ -0,0 +1 @@
+export * from './types';

packages/common/infra/src/media/types.ts

@@ -0,0 +1,72 @@
+/**
+ * Attachment block audio media.
+ * blockId/docId/workspaceId are used to identify the source of the media
+ * to control the exclusivity playback state of audio across the whole application.
+ */
+export interface AttachmentBlockAudioMedia {
+  blobId: string; // aka sourceId
+  blockId: string;
+  docId: string;
+  workspaceId: string;
+}
+
+export interface AudioMediaDescriptor {
+  key: AudioMediaKey;
+  tabId: string | null;
+  name: string;
+  size: number;
+  blobId: string; // aka sourceId
+}
+
+// workspaceId/docId/blockId/blobId
+export type AudioMediaKey = `${string}/${string}/${string}/${string}`;
+
+export const attachmentBlockAudioMediaKey = (
+  media: AttachmentBlockAudioMedia
+): AudioMediaKey => {
+  return `${media.workspaceId}/${media.docId}/${media.blockId}/${media.blobId}`;
+};
+
+export const parseAudioMediaKey = (
+  key: AudioMediaKey
+): AttachmentBlockAudioMedia => {
+  const [workspaceId, docId, blockId, blobId] = key.split('/');
+  return {
+    workspaceId,
+    docId,
+    blockId,
+    blobId,
+  };
+};
+
+type State = 'idle' | 'playing' | 'paused' | 'stopped';
+
+export interface MediaStats {
+  key: AudioMediaKey;
+  tabId: string | null;
+  duration: number;
+  name: string;
+  size: number;
+  waveform: number[]; // for drawing waveform, maxmium of 1000 samples
+}
+
+export interface PlaybackState {
+  key: AudioMediaKey;
+  tabId: string | null;
+  state: State;
+  /**
+   * Whenever the user seek the media, the startSeekOffset will be updated
+   */
+  seekOffset: number;
+  /**
+   * Whenever the media state is updated.
+   * the updateTime will be updated. It is in milliseconds (unix timestamp).
+   * The current playback position (0-based, in seconds) is calculated by `seekOffset + (Date.now() - updateTime) / 1000 * rate`
+   */
+  updateTime: number;
+  /**
+   * the playback rate
+   * Not implemented yet. Always 1.0.
+   */
+  // rate: number;
+}

packages/common/infra/src/utils/object-pool.ts

@@ -1,6 +1,6 @@
 import { Unreachable } from '@affine/env/constant';
 
-export interface RcRef<T> {
+export interface RcRef<T> extends Disposable {
   obj: T;
   release: () => void;
 }
@@ -21,17 +21,19 @@ export class ObjectPool<Key, T> {
     if (exist) {
       exist.rc++;
       let released = false;
+      const release = () => {
+        // avoid double release
+        if (released) {
+          return;
+        }
+        released = true;
+        exist.rc--;
+        this.requestGc();
+      };
       return {
         obj: exist.obj,
-        release: () => {
-          // avoid double release
-          if (released) {
-            return;
-          }
-          released = true;
-          exist.rc--;
-          this.requestGc();
-        },
+        release,
+        [Symbol.dispose]: release,
       };
     }
     return null;