TextDecoder: fix decode() ignoring TypedArray's byteLength/byteOffset

HtmlUnit · Aug 27, 2024 · 335f99f · 335f99f
1 parent af041bc
commit 335f99f
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 17 deletions.
diff --git a/src/main/java/org/htmlunit/javascript/host/TextDecoder.java b/src/main/java/org/htmlunit/javascript/host/TextDecoder.java
@@ -15,6 +15,7 @@
 package org.htmlunit.javascript.host;
 
 import java.nio.charset.Charset;
+import java.util.Arrays;
 import java.util.Locale;
 
 import org.htmlunit.corejs.javascript.typedarrays.NativeArrayBuffer;
@@ -79,29 +80,37 @@ public String decode(final Object buffer) {
             return "";
         }
 
-        NativeArrayBuffer arrayBuffer = null;
         if (buffer instanceof NativeArrayBuffer) {
-            arrayBuffer = (NativeArrayBuffer) buffer;
-        }
-        else if (buffer instanceof NativeArrayBufferView) {
-            arrayBuffer = ((NativeArrayBufferView) buffer).getBuffer();
+            return new String(((NativeArrayBuffer) buffer).getBuffer(), getEncoding(whatwgEncoding_));
         }
 
-        if (arrayBuffer != null) {
-            if (XUserDefinedCharset.NAME.equalsIgnoreCase(whatwgEncoding_)) {
-                return new String(arrayBuffer.getBuffer(), XUserDefinedCharset.INSTANCE);
+        if (buffer instanceof NativeArrayBufferView) {
+            final NativeArrayBufferView arrayBufferView = (NativeArrayBufferView) buffer;
+            final NativeArrayBuffer arrayBuffer = arrayBufferView.getBuffer();
+            if (arrayBuffer != null) {
+                final int byteLength = arrayBufferView.getByteLength();
+                final int byteOffset = arrayBufferView.getByteOffset();
+                final byte[] backedBytes = arrayBuffer.getBuffer();
+                final byte[] bytes = Arrays.copyOfRange(backedBytes, byteOffset, byteOffset + byteLength);
+                return new String(bytes, getEncoding(whatwgEncoding_));
             }
-
-            final String ianaEncoding = StandardEncodingTranslator
-                    .ENCODING_TO_IANA_ENCODING.getOrDefault(whatwgEncoding_, whatwgEncoding_);
-            // Convert our IANA encoding names to Java charset names
-            final String javaEncoding = StandardEncodingTranslator
-                    .IANA_TO_JAVA_ENCODINGS.getOrDefault(ianaEncoding, ianaEncoding);
-
-            return new String(arrayBuffer.getBuffer(), Charset.forName(javaEncoding));
         }
 
         throw JavaScriptEngine.typeError("Argument 1 of TextDecoder.decode could not be"
                                 + " converted to any of: ArrayBufferView, ArrayBuffer.");
     }
+
+    private Charset getEncoding(final String encodingLabel) {
+        if (XUserDefinedCharset.NAME.equalsIgnoreCase(encodingLabel)) {
+            return XUserDefinedCharset.INSTANCE;
+        }
+
+        final String ianaEncoding = StandardEncodingTranslator
+                .ENCODING_TO_IANA_ENCODING.getOrDefault(encodingLabel, encodingLabel);
+        // Convert our IANA encoding names to Java charset names
+        final String javaEncoding = StandardEncodingTranslator
+                .IANA_TO_JAVA_ENCODINGS.getOrDefault(ianaEncoding, ianaEncoding);
+
+        return Charset.forName(javaEncoding);
+    }
 }
diff --git a/src/test/java/org/htmlunit/javascript/host/TextDecoderTest.java b/src/test/java/org/htmlunit/javascript/host/TextDecoderTest.java
@@ -673,7 +673,7 @@ public void encode() throws Exception {
      * @throws Exception on test failure
      */
     @Test
-    @Alerts("HtmlUnit")
+    @Alerts({"HtmlUnit", "mlU"})
     public void decode() throws Exception {
         final String html = "<html>\n"
             + "<head>\n"
@@ -686,6 +686,10 @@ public void decode() throws Exception {
             + "      var dec = new TextDecoder('utf-8');\n"
             + "      var decoded = dec.decode(encoded);\n"
             + "      log(decoded);\n"
+
+            + "      var arrayBuffer = encoded.buffer;\n"
+            + "      var typedArray = new Uint8Array(arrayBuffer, 2, 3);\n"
+            + "      log(dec.decode(typedArray));\n"
             + "    }\n"
             + "  </script>\n"
             + "</head>\n"