fix: Fix data URI parsing when charset present

Data URIs can have additional parameters other than just the MIME type
and encoding.  While working on TTML fixes, I was debugging by
inserting data URIs as TTML text tracks.  But the data URI encoder I
used was inserting a "charset" parameter into the URI, which was
rejected by Shaka Player.

This fixes the data URI parser to understand these extra parameters.

While researching this fix, I also found that only the base64 encoding
is specified for data URIs.  So now only that specific parameter is
supported as an encoding, and the UNKNOWN_DATA_URI_ENCODING error code
has been retired.

Change-Id: I7022529b0931e80a77a1120fc86c25b42e1fb3d9

lib/net/data_uri_plugin.js

@@ -79,30 +79,30 @@ shaka.net.DataUriPlugin = class {
     const info = infoAndData[0];
     const dataStr = window.decodeURIComponent(infoAndData.slice(1).join(','));
 
-    // Extract the encoding (optional).
-    const typeAndEncoding = info.split(';');
-    let encoding = null;
-    if (typeAndEncoding.length > 1) {
-      encoding = typeAndEncoding[1];
+    // The MIME type is always the first thing in the semicolon-separated list
+    // of type parameters.  It may be blank.
+    const typeInfoList = info.split(';');
+    const contentType = typeInfoList[0];
+
+    // Check for base64 encoding, which is always the last in the
+    // semicolon-separated list if present.
+    let base64Encoded = false;
+    if (typeInfoList.length > 1 &&
+        typeInfoList[typeInfoList.length - 1] == 'base64') {
+      base64Encoded = true;
+      typeInfoList.pop();
     }
 
     // Convert the data.
     /** @type {BufferSource} */
     let data;
-    if (encoding == 'base64') {
+    if (base64Encoded) {
       data = shaka.util.Uint8ArrayUtils.fromBase64(dataStr);
-    } else if (encoding) {
-      shaka.log.error('Bad data URI, unknown encoding');
-      throw new shaka.util.Error(
-          shaka.util.Error.Severity.CRITICAL,
-          shaka.util.Error.Category.NETWORK,
-          shaka.util.Error.Code.UNKNOWN_DATA_URI_ENCODING,
-          uri);
     } else {
       data = shaka.util.StringUtils.toUTF8(dataStr);
     }
 
-    return {data: data, contentType: typeAndEncoding[0]};
+    return {data: data, contentType};
   }
 };
 

lib/util/error.js

@@ -238,11 +238,7 @@ shaka.util.Error.Code = {
    */
   'MALFORMED_DATA_URI': 1004,
 
-  /**
-   * A network request was made with a data URI using an unknown encoding.
-   * <br> error.data[0] is the URI.
-   */
-  'UNKNOWN_DATA_URI_ENCODING': 1005,
+  // RETIRED: 'UNKNOWN_DATA_URI_ENCODING': 1005,
 
   /**
    * A request filter threw an error.

test/net/data_uri_plugin_unit.js

@@ -23,18 +23,35 @@ describe('DataUriPlugin', () => {
         'data:;base64,SGVsbG8sIFdvcmxkIQ%3D%3D', '', 'Hello, World!');
   });
 
-  it('supports extra colin', async () => {
+  it('supports extra colon', async () => {
     await testSucceeds('data:,Hello:', '', 'Hello:');
   });
 
-  it('supports extra semi-colin', async () => {
+  it('supports extra semi-colon', async () => {
     await testSucceeds('data:,Hello;', '', 'Hello;');
   });
 
   it('supports extra comma', async () => {
     await testSucceeds('data:,Hello,', '', 'Hello,');
   });
 
+  it('supports character set metadata', async () => {
+    await testSucceeds(
+        'data:text/plain;charset=UTF-8,Hello,', 'text/plain', 'Hello,');
+  });
+
+  it('supports arbitrary metadata', async () => {
+    await testSucceeds(
+        'data:text/plain;foo=bar,Hello,', 'text/plain', 'Hello,');
+  });
+
+  it('supports arbitrary metadata with base64 encoding', async () => {
+    await testSucceeds(
+        'data:text/plain;foo=bar;base64,SGVsbG8sIFdvcmxkIQ%3D%3D',
+        'text/plain',
+        'Hello, World!');
+  });
+
   it('fails for empty URI', async () => {
     await testFails('', shaka.util.Error.Code.MALFORMED_DATA_URI);
   });