Implement alternative string conversion functions in net/ on Android,

net/android/java/src/org/chromium/net/NetStringUtil.java

Index: net/android/java/src/org/chromium/net/NetStringUtil.java
diff --git a/net/android/java/src/org/chromium/net/NetStringUtil.java b/net/android/java/src/org/chromium/net/NetStringUtil.java
new file mode 100644
index 0000000000000000000000000000000000000000..4ab1bd9f5bd8f7eb98edaffb119fa814f8202d6e
--- /dev/null
+++ b/net/android/java/src/org/chromium/net/NetStringUtil.java
@@ -0,0 +1,89 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+package org.chromium.net;
+
+import org.chromium.base.CalledByNative;
+import org.chromium.base.JNINamespace;
+
+import java.nio.ByteBuffer;
+import java.nio.charset.Charset;
+import java.nio.charset.CharsetDecoder;
+import java.nio.charset.CodingErrorAction;
+import java.text.Normalizer;
+
+/**
+ * Utility functions for converting strings between formats when not built with
+ * icu.
+ */
+@JNINamespace("net::android")
+public class NetStringUtil {
+    /**
+     * Attempts to convert text in a given character set to a Unicode string.
+     * Returns null on failure.
+     * @param text ByteBuffer containing the character array to convert.
+     * @param charset Character set it's in encoded in.
+     * @return: Unicode string on success, null on failure.
+     */
+    @CalledByNative
+    private static String convertToUnicode(
+            ByteBuffer text,
+            String charset_name) {
+        try {
+            Charset charset = Charset.forName(charset_name);
+            CharsetDecoder decoder = charset.newDecoder();
+            // On invalid characters, this will throw an exception.
+            return decoder.decode(text).toString();
+        } catch (Exception e) {
+            return null;
+        }
+    }
+
+    /**
+     * Attempts to convert text in a given character set to a Unicode string,
+     * and normalize it.  Returns null on failure.
+     * @param text ByteBuffer containing the character array to convert.
+     * @param charset Character set it's in encoded in.
+     * @return: Unicode string on success, null on failure.
+     */
+    @CalledByNative
+    private static String convertToUnicodeAndNormalize(
+            ByteBuffer text,
+            String charset_name) {
+        String unicodeString = convertToUnicode(text, charset_name);
+        if (unicodeString == null)
+            return unicodeString;
+        return Normalizer.normalize(unicodeString, Normalizer.Form.NFC);
+    }
+
+    /**
+     * Convert text in a given character set to a Unicode string.  Any invalid
+     * characters are replaced with U+FFFD.  Returns null if the character set
+     * is not recognized.
+     * @param text ByteBuffer containing the character array to convert.
+     * @param charset Character set it's in encoded in.
+     * @return: Unicode string on success, null on failure.
+     */
+    @CalledByNative
+    private static String convertToUnicodeWithSubstitutions(
+            ByteBuffer text,
+            String charset_name) {
+        try {
+            Charset charset = Charset.forName(charset_name);
+
+            // TODO(mmenke):  Investigate if Charset.decode() can be used
+            // instead.  The question is whether it uses the proper replace
+            // character.  JDK CharsetDecoder docs say U+FFFD is the default,
+            // but Charset.decode() docs say it uses the "charset's default
+            // replacement byte array".
+            CharsetDecoder decoder = charset.newDecoder();
+            decoder.onMalformedInput(CodingErrorAction.REPLACE);
+            decoder.onUnmappableCharacter(CodingErrorAction.REPLACE);
+            decoder.replaceWith("\uFFFD");
+            return decoder.decode(text).toString();
+        } catch (Exception e) {
+            return null;
+        }
+    }
+}