fix(CEA): Fix multi byte language support in CEA-708 (#7929)

Fixes #7926
shaka-project · Jan 23, 2025 · 0f84e02 · 0f84e02
1 parent 1c3b66a
commit 0f84e02
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 19 deletions.
diff --git a/lib/cea/cea708_service.js b/lib/cea/cea708_service.js
@@ -169,24 +169,16 @@ shaka.cea.Cea708Service = class {
       const firstByte = dtvccPacket.readByte().value;
       const secondByte = dtvccPacket.readByte().value;
 
-      const isTextBlock = (b) => {
-        return (b >= 0x20 && b <= 0x7f) || (b >= 0xa0 && b <= 0xff);
+      const toHexString = (byteArray) => {
+        return byteArray.map((byte) => {
+          return ('0' + (byte & 0xFF).toString(16)).slice(-2);
+        }).join('');
       };
-
-      if (isTextBlock(firstByte) && isTextBlock(secondByte)) {
-        const toHexString = (byteArray) => {
-          return byteArray.map((byte) => {
-            return ('0' + (byte & 0xFF).toString(16)).slice(-2);
-          }).join('');
-        };
-        const unicode = toHexString([firstByte, secondByte]);
-        // Takes a unicode hex string and creates a single character.
-        const char = String.fromCharCode(parseInt(unicode, 16));
-        this.currentWindow_.setCharacter(char);
-        return null;
-      } else {
-        dtvccPacket.rewind(2);
-      }
+      const unicode = toHexString([firstByte, secondByte]);
+      // Takes a unicode hex string and creates a single character.
+      const char = String.fromCharCode(parseInt(unicode, 16));
+      this.currentWindow_.setCharacter(char);
+      return null;
     }
 
     const window = this.currentWindow_;

diff --git a/test/cea/cea708_service_unit.js b/test/cea/cea708_service_unit.js
@@ -121,8 +121,7 @@ describe('Cea708Service', () => {
     expect(captions).toEqual(expectedCaptions);
   });
 
-
-  it('decodes multibyte unstyled caption text', () => {
+  it('decodes multibyte unstyled caption text (Korean)', () => {
     const controlCodes = [
       ...defineWindow,
       // Series of C0 control codes that add multi-byte text.
@@ -150,6 +149,34 @@ describe('Cea708Service', () => {
     expect(captions).toEqual(expectedCaptions);
   });
 
+  it('decodes multibyte unstyled caption text (Polish)', () => {
+    const controlCodes = [
+      ...defineWindow,
+      // Series of C0 control codes that add multi-byte text.
+      0x18, 0x01, 0x7c, 0xF3, 0x18, 0x01, 0x42, 0x18, 0x01, 0x07, // ż, ó, ł, ć
+    ];
+
+    const packet1 = createCea708PacketFromBytes(controlCodes, startTime);
+    const packet2 = createCea708PacketFromBytes(hideWindow, endTime);
+
+    const text = 'żółć'; // cspell:ignore żółć
+    const topLevelCue = CeaUtils.createWindowedCue(startTime, endTime, '',
+        serviceNumber, windowId, rowCount, colCount, anchorId);
+    topLevelCue.nestedCues = [
+      CeaUtils.createDefaultCue(startTime, endTime, /* payload= */ text),
+    ];
+
+    const expectedCaptions = [
+      {
+        stream,
+        cue: topLevelCue,
+      },
+    ];
+
+    const captions = getCaptionsFromPackets(service, packet1, packet2);
+    expect(captions).toEqual(expectedCaptions);
+  });
+
   it('setPenLocation sets the pen location correctly', () => {
     const controlCodes = [
       ...defineWindow,