codecs: Properly detect Matroska audio/video. Add codec support for AV1 video and DTS audio. Bump to 1.1.4.

codecs/codecs.js

@@ -53,6 +53,13 @@ const FORMAT_NAME_TO_SHORT_TYPE = {
   'wav': 'audio/wav',
 }
 
+// https://developer.mozilla.org/en-US/docs/Web/Media/Formats/Containers#webm says that only
+// the following codecs are supported for webm:
+// - video: AV1, VP8, VP9
+// - audio: Opus, Vorbis
+const WEBM_AUDIO_CODECS = [ 'opus', 'vorbis' ];
+const WEBM_VIDEO_CODECS = [ 'av1', 'vp8', 'vp9' ];
+
 /**
  * TODO: Reconcile this with file/sniffer.js findMimeType() which does signature matching.
  * @param {ProbeInfo} info
@@ -94,9 +101,16 @@ export function getShortMIMEString(info) {
     case 'ogg':
       subType = 'ogg';
       break;
-    // Should we detect .mkv files as x-matroska?
     case 'matroska,webm':
-      subType = 'webm';
+      let isWebM = true;
+      for (const stream of info.streams) {
+        if (   (stream.codec_type === 'audio' && !WEBM_AUDIO_CODECS.includes(stream.codec_name))
+            || (stream.codec_type === 'video' && !WEBM_VIDEO_CODECS.includes(stream.codec_name))) {
+              isWebM = false;
+          break;
+        }
+      }
+      subType = isWebM ? 'webm' : 'x-matroska';
       break;
     default:
       throw `Cannot handle format ${formatName} yet. ` +
@@ -132,11 +146,12 @@ export function getFullMIMEString(info) {
         default: 
           switch (stream.codec_name) {
             case 'aac': codecFrags.add(getMP4ACodecString(stream)); break;
-            case 'vorbis': codecFrags.add('vorbis'); break;
-            case 'opus': codecFrags.add('opus'); break;
             // I'm going off of what Chromium calls this one, with the dash.
             case 'ac3': codecFrags.add('ac-3'); break;
+            case 'dts': codecFrags.add('dts'); break;
             case 'flac': codecFrags.add('flac'); break;
+            case 'opus': codecFrags.add('opus'); break;
+            case 'vorbis': codecFrags.add('vorbis'); break;
             default:
               throw `Could not handle audio codec_name ${stream.codec_name}, ` +
                     `codec_tag_string ${stream.codec_tag_string} for file ${info.format.filename} yet. ` +
@@ -153,10 +168,12 @@ export function getFullMIMEString(info) {
         case 'png': continue;
         default:
           switch (stream.codec_name) {
+            case 'av1': codecFrags.add('av1'); break;
             case 'h264': codecFrags.add(getAVC1CodecString(stream)); break;
-            case 'mpeg2video': codecFrags.add('mpeg2video'); break;
             // Skip mjpeg as a video stream for the codecs string.
             case 'mjpeg': break;
+            case 'mpeg2video': codecFrags.add('mpeg2video'); break;
+            case 'vp8': codecFrags.add('vp8'); break;
             case 'vp9': codecFrags.add(getVP09CodecString(stream)); break;
             default:
               throw `Could not handle video codec_name ${stream.codec_name}, ` +

io/bitbuffer.js

@@ -32,7 +32,7 @@ export class BitBuffer {
    */
   constructor(numBytes, mtl = false) {
     if (typeof numBytes != typeof 1 || numBytes <= 0) {
-      throw "Error! ByteBuffer initialized with '" + numBytes + "'";
+      throw "Error! BitBuffer initialized with '" + numBytes + "'";
     }
 
     /**

io/bitstream.js

@@ -51,7 +51,7 @@ export class BitStream {
    */
   constructor(ab, mtl, opt_offset, opt_length) {
     if (!(ab instanceof ArrayBuffer)) {
-      throw 'Error! BitArray constructed with an invalid ArrayBuffer object';
+      throw 'Error! BitStream constructed with an invalid ArrayBuffer object';
     }
 
     const offset = opt_offset || 0;

io/bytestream.js

@@ -22,7 +22,8 @@ export class ByteStream {
    */
   constructor(ab, opt_offset, opt_length) {
     if (!(ab instanceof ArrayBuffer)) {
-      throw 'Error! BitArray constructed with an invalid ArrayBuffer object';
+      console.error(typeof ab);
+      throw 'Error! ByteStream constructed with an invalid ArrayBuffer object';
     }
 
     const offset = opt_offset || 0;

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@codedread/bitjs",
-  "version": "1.1.2",
+  "version": "1.1.4",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "@codedread/bitjs",
-      "version": "1.1.2",
+      "version": "1.1.4",
       "license": "MIT",
       "devDependencies": {
         "c8": "^7.12.0",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@codedread/bitjs",
-  "version": "1.1.3",
+  "version": "1.1.4",
   "description": "Binary Tools for JavaScript",
   "homepage": "https://github.com/codedread/bitjs",
   "author": "Jeff Schiller",

tests/codecs.spec.js

@@ -113,16 +113,30 @@ describe('codecs test suite', () => {
     it('detects WEBM video', () => {
       expect(getShortMIMEString({
         format: { format_name: 'matroska,webm' },
-        streams: [ { codec_type: 'video' } ],
+        streams: [ { codec_type: 'video', codec_name: 'vp8' } ],
       })).equals('video/webm');
     });
 
     it('detects WEBM audio', () => {
       expect(getShortMIMEString({
         format: { format_name: 'matroska,webm' },
-        streams: [ { codec_type: 'audio' } ],
+        streams: [ { codec_type: 'audio', codec_name: 'vorbis' } ],
       })).equals('audio/webm');
     });
+
+    it('detects Matroska Video', () => {
+      expect(getShortMIMEString({
+        format: { format_name: 'matroska,webm' },
+        streams: [ { codec_type: 'video', codec_name: 'h264' } ],
+      })).equals('video/x-matroska');
+    });
+
+    it('detects Matroska audio', () => {
+      expect(getShortMIMEString({
+        format: { format_name: 'matroska,webm' },
+        streams: [ { codec_type: 'audio', codec_name: 'dts' } ],
+      })).equals('audio/x-matroska');
+    });
   });
 
   describe('getFullMIMEString()', () => {
@@ -228,86 +242,193 @@ describe('codecs test suite', () => {
     });
   });
 
-  describe('VP09 / VP9', () => {
+  describe('WebM' ,() => {
-    /** @type {ProbeInfo} */
+    describe('AV1', () => {
-    let info;
+      /** @type {ProbeInfo} */
-
+      let info;
-    beforeEach(() => {
+  
-      info = {
-        format: { format_name: 'matroska,webm' },
-        streams: [{
-          codec_type: 'video',
-          codec_tag_string: 'vp09',
-        }],
-      };
-    });
-
-    describe('Profile tests', () => {
       beforeEach(() => {
-        info.streams[0].level = 20;
+        info = {
+          format: { format_name: 'matroska,webm' },
+          streams: [{ codec_type: 'video', codec_name: 'av1' }],
+        };
       });
 
-      it('detects Profile 0', () => {
+      it('outputs MIME string', () => {
-        info.streams[0].profile = 'Profile 0';
+        expect(getShortMIMEString(info)).equals('video/webm');
-        expect(getFullMIMEString(info))
+        expect(getFullMIMEString(info)).equals('video/webm; codecs="av1"')
-            .to.be.a('string')
-            .and.satisfy(s => s.startsWith('video/webm; codecs="vp09.00.'));
       });
     });
 
-    describe('Level tests', () => {
+    describe('VP8', () => {
+      /** @type {ProbeInfo} */
+      let info;
+  
       beforeEach(() => {
-        info.streams[0].profile = 'Profile 0';
+        info = {
+          format: { format_name: 'matroska,webm' },
+          streams: [{ codec_type: 'video', codec_name: 'vp8' }],
+        };
       });
 
-      it('detects 2-digit hex level', () => {
+      it('outputs MIME string', () => {
+        expect(getShortMIMEString(info)).equals('video/webm');
+        expect(getFullMIMEString(info)).equals('video/webm; codecs="vp8"')
+      });
+    });
+
+    describe('VP09 / VP9', () => {
+      /** @type {ProbeInfo} */
+      let info;
+  
+      beforeEach(() => {
+        info = {
+          format: { format_name: 'matroska,webm' },
+          streams: [{ codec_type: 'video', codec_name: 'vp9', codec_tag_string: 'vp09' }],
+        };
+      });
+  
+      describe('Profile tests', () => {
+        beforeEach(() => {
+          info.streams[0].level = 20;
+        });
+  
+        it('detects Profile 0', () => {
+          info.streams[0].profile = 'Profile 0';
+          expect(getFullMIMEString(info))
+              .to.be.a('string')
+              .and.satisfy(s => s.startsWith('video/webm; codecs="vp09.00.'));
+        });
+      });
+  
+      describe('Level tests', () => {
+        beforeEach(() => {
+          info.streams[0].profile = 'Profile 0';
+        });
+  
+        it('detects 2-digit hex level', () => {
+          info.streams[0].level = 21; // 0x15
+          expect(getFullMIMEString(info))
+              .to.be.a('string')
+              .and.satisfy(s => s.startsWith('video/webm; codecs="vp09.'))
+              .and.satisfy(s => {
+                const matches = s.match(/vp09\.[0-9]{2}\.([0-9A-F]{2})\.[0-9A-F]{2}/);
+                return matches && matches.length === 2 && matches[1] === '15';
+              });
+          });
+  
+        it('detects level = -99', () => {
+          info.streams[0].level = -99; // I'm not sure what ffprobe means by this.
+          expect(getFullMIMEString(info))
+              .to.be.a('string')
+              .and.equals('video/webm; codecs="vp9"');
+        });
+      });
+  
+      it('detects codec_name=vp9 but no codec_tag_string', () => {
+        info.streams[0].codec_name = 'vp9';
+        info.streams[0].codec_tag_string = '[0][0][0][0]';
+        info.streams[0].profile = 'Profile 0';
         info.streams[0].level = 21; // 0x15
         expect(getFullMIMEString(info))
-            .to.be.a('string')
+        .to.be.a('string')
-            .and.satisfy(s => s.startsWith('video/webm; codecs="vp09.'))
+        .and.satisfy(s => s.startsWith('video/webm; codecs="vp09.00.15'));
-            .and.satisfy(s => {
-              const matches = s.match(/vp09\.[0-9]{2}\.([0-9A-F]{2})\.[0-9A-F]{2}/);
-              return matches && matches.length === 2 && matches[1] === '15';
-            });
-        });
-
-      it('detects level = -99', () => {
-        info.streams[0].level = -99; // I'm not sure what ffprobe means by this.
-        expect(getFullMIMEString(info))
-            .to.be.a('string')
-            .and.equals('video/webm; codecs="vp9"');
       });
     });
 
-    it('detects codec_name=vp9 but no codec_tag_string', () => {
+    describe('Vorbis', () => {
-      info.streams[0].codec_name = 'vp9';
+      /** @type {ProbeInfo} */
-      info.streams[0].codec_tag_string = '[0][0][0][0]';
+      let info;
-      info.streams[0].profile = 'Profile 0';
+  
-      info.streams[0].level = 21; // 0x15
+      beforeEach(() => {
-      expect(getFullMIMEString(info))
+        info = {
-      .to.be.a('string')
+          format: { format_name: 'matroska,webm' },
-      .and.satisfy(s => s.startsWith('video/webm; codecs="vp09.00.15'));
+          streams: [{ codec_type: 'audio', codec_name: 'vorbis' }],
+        };
+      });
+  
+      it('detects vorbis', () => {
+        expect(getFullMIMEString(info))
+            .to.be.a('string')
+            .and.equals('audio/webm; codecs="vorbis"');
+      });
     });
+  
+    describe('Opus', () => {
+      /** @type {ProbeInfo} */
+      let info;
+  
+      beforeEach(() => {
+        info = {
+          format: { format_name: 'matroska,webm' },
+          streams: [{
+            codec_type: 'audio',
+            codec_name: 'opus',
+          }],
+        };
+      });
+  
+      it('detects opus', () => {
+        expect(getFullMIMEString(info))
+            .to.be.a('string')
+            .and.equals('audio/webm; codecs="opus"');
+      });
+    });  
   });
 
-  describe('MPEG2', () => {
+  describe('Matroska', () => {
-    /** @type {ProbeInfo} */
+    describe('MPEG2 codec', () => {
-    let info;
+      /** @type {ProbeInfo} */
-
+      let info;
-    beforeEach(() => {
+  
-      info = {
+      beforeEach(() => {
-        format: { format_name: 'matroska,webm' },
+        info = {
-        streams: [{
+          format: { format_name: 'matroska,webm' },
-          codec_type: 'video',
+          streams: [{ codec_type: 'video', codec_name: 'mpeg2video' }],
-          codec_name: 'mpeg2video',
+        };
-        }],
+      });
-      };
+  
+      it('outputs full MIME string', () => {
+        expect(getFullMIMEString(info))
+            .to.be.a('string')
+            .and.equals('video/x-matroska; codecs="mpeg2video"');
+      });
     });
+  
+    describe('AC-3', () => {
+      /** @type {ProbeInfo} */
+      let info;
+  
+      beforeEach(() => {
+        info = {
+          format: { format_name: 'matroska,webm' },
+          streams: [{ codec_type: 'audio', codec_name: 'ac3' }],
+        };
+      });
+  
+      it('detects AC-3', () => {
+        expect(getFullMIMEString(info))
+            .to.be.a('string')
+            .and.equals('audio/x-matroska; codecs="ac-3"');
+      });
+    });
+  
+    describe('DTS', () => {
+      /** @type {ProbeInfo} */
+      let info;
 
-    it('detects mpeg2video', () => {
+      beforeEach(() => {
-      expect(getFullMIMEString(info))
+        info = {
-          .to.be.a('string')
+          format: { format_name: 'matroska,webm' },
-          .and.equals('video/webm; codecs="mpeg2video"');
+          streams: [{ codec_type: 'audio', codec_name: 'dts' }],
+        };
+      });
+  
+      it('outputs full MIME string', () => {
+        expect(getFullMIMEString(info))
+            .to.be.a('string')
+            .and.equals('audio/x-matroska; codecs="dts"');
+      });
     });
   });
 
@@ -393,69 +514,6 @@ describe('codecs test suite', () => {
     });
   });
 
-  describe('Vorbis', () => {
-    /** @type {ProbeInfo} */
-    let info;
-
-    beforeEach(() => {
-      info = {
-        format: { format_name: 'matroska,webm' },
-        streams: [{
-          codec_type: 'audio',
-          codec_name: 'vorbis',
-        }],
-      };
-    });
-
-    it('detects vorbis', () => {
-      expect(getFullMIMEString(info))
-          .to.be.a('string')
-          .and.equals('audio/webm; codecs="vorbis"');
-    });
-  });
-
-  describe('Opus', () => {
-    /** @type {ProbeInfo} */
-    let info;
-
-    beforeEach(() => {
-      info = {
-        format: { format_name: 'matroska,webm' },
-        streams: [{
-          codec_type: 'audio',
-          codec_name: 'opus',
-        }],
-      };
-    });
-
-    it('detects opus', () => {
-      expect(getFullMIMEString(info))
-          .to.be.a('string')
-          .and.equals('audio/webm; codecs="opus"');
-    });
-  });
-
-  describe('AC-3', () => {
-    /** @type {ProbeInfo} */
-    let info;
-
-    beforeEach(() => {
-      info = {
-        format: { format_name: 'matroska,webm' },
-        streams: [{
-          codec_type: 'audio',
-          codec_name: 'ac3',
-        }],
-      };
-    });
-
-    it('detects AC-3', () => {
-      expect(getFullMIMEString(info))
-          .to.be.a('string')
-          .and.equals('audio/webm; codecs="ac-3"');
-    });
-  });
-
   describe('AVI', () => {
     it('detects AVI', () => {
       /** @type {ProbeInfo} */