Fix WAV parser, sync issue with Broadcast Wave Format (BWF) Audio Ext…

…ension chunk Strips null characters from BWF extension strings. Resolves: #1163
Borewit · Jul 4, 2022 · b486eb6 · b486eb6
1 parent 55344b5
commit b486eb6
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 5 deletions.
diff --git a/lib/wav/BwfChunk.ts b/lib/wav/BwfChunk.ts
@@ -1,5 +1,6 @@
 import { IGetToken } from 'strtok3/lib/core';
 import * as Token from 'token-types';
+import { stripNulls } from '../common/Util';
 
 export interface IBroadcastAudioExtensionChunk {
   description: string;
@@ -22,11 +23,11 @@ export const BroadcastAudioExtensionChunk: IGetToken<IBroadcastAudioExtensionChu
 
   get: (uint8array, off) => {
     return {
-      description: new Token.StringType(256, 'ascii').get(uint8array, off).trim(),
-      originator: new Token.StringType(32, 'ascii').get(uint8array, off + 256).trim(),
-      originatorReference: new Token.StringType(32, 'ascii').get(uint8array, off + 288).trim(),
-      originationDate: new Token.StringType(10, 'ascii').get(uint8array, off + 320).trim(),
-      originationTime: new Token.StringType(8, 'ascii').get(uint8array, off + 330).trim(),
+      description: stripNulls(new Token.StringType(256, 'ascii').get(uint8array, off)).trim(),
+      originator: stripNulls(new Token.StringType(32, 'ascii').get(uint8array, off + 256)).trim(),
+      originatorReference: stripNulls(new Token.StringType(32, 'ascii').get(uint8array, off + 288)).trim(),
+      originationDate: stripNulls(new Token.StringType(10, 'ascii').get(uint8array, off + 320)).trim(),
+      originationTime: stripNulls(new Token.StringType(8, 'ascii').get(uint8array, off + 330)).trim(),
       timeReferenceLow: Token.UINT32_LE.get(uint8array, off + 338),
       timeReferenceHigh: Token.UINT32_LE.get(uint8array, off + 342),
       version: Token.UINT16_LE.get(uint8array, off + 346),

diff --git a/lib/wav/WaveParser.ts b/lib/wav/WaveParser.ts
@@ -128,6 +128,8 @@ export class WaveParser extends BasicParser {
           Object.keys(bext).forEach(key => {
             this.metadata.addTag('exif', 'bext.' + key, bext[key]);
           });
+          const bextRemaining = header.chunkSize - BroadcastAudioExtensionChunk.len;
+          await this.tokenizer.ignore(bextRemaining);
           break;
 
         case '\x00\x00\x00\x00': // padding ??

diff --git a/test/samples/wav/issue-1163.bwf b/test/samples/wav/issue-1163.bwf
diff --git a/test/test-file-wav.ts b/test/test-file-wav.ts
@@ -178,5 +178,22 @@ describe('Parse RIFF/WAVE audio format', () => {
     assert.approximately(format.duration, 2478 / 16000, 5 / 1000, 'format.duration');
   });
 
+  // https://github.com/Borewit/music-metadata/issues/1163
+  it('Support chunk size larger then BWF extension', async () => {
+    // const filePath = path.join(wavSamples, 'unreadable-tags.wav');
+    const filePath = path.join(wavSamples, 'issue-1163.bwf');
+    const {format, common, native} = await mm.parseFile(filePath);
+
+    assert.strictEqual(format.container, 'WAVE', 'format.container');
+    assert.strictEqual(format.codec, 'PCM', 'format.codec');
+
+    assert.strictEqual(common.artist, 'Some Composer', 'common.artists');
+    assert.strictEqual(common.title, 'Title Redacted', 'common.title');
+    assert.deepStrictEqual(common.track, {no: 1, of: 12}, 'common.track');
+
+    const exif = mm.orderTags(native.exif);
+    assert.deepStrictEqual(exif['bext.originator'], ['Pro Tools'], 'BWF: exif.bext.originator');
+  });
+
 });