Support MBCS characters in contacts and SMS messages that are stored in SIM cards.

Change-Id: I0438c5748a47512fc2a0e4b89ac85e30e3f5c6de

5 files changed, 119 insertions, 3 deletions

diff --git a/telephony/java/android/telephony/SmsMessage.java b/telephony/java/android/telephony/SmsMessage.java
index d899430..0746562 100644
--- a/telephony/java/android/telephony/SmsMessage.java
+++ b/telephony/java/android/telephony/SmsMessage.java
@@ -53,6 +53,10 @@ public class SmsMessage {
     public static final int ENCODING_7BIT = 1;
     public static final int ENCODING_8BIT = 2;
     public static final int ENCODING_16BIT = 3;
+    /**
+     * @hide This value is not defined in global standard. Only in Korea, this is used.
+     */
+    public static final int ENCODING_KSC5601 = 4;
 
     /** The maximum number of payload bytes per message */
     public static final int MAX_USER_DATA_BYTES = 140;
diff --git a/telephony/java/com/android/internal/telephony/GsmAlphabet.java b/telephony/java/com/android/internal/telephony/GsmAlphabet.java
index 75ea116..e42827f 100644
--- a/telephony/java/com/android/internal/telephony/GsmAlphabet.java
+++ b/telephony/java/com/android/internal/telephony/GsmAlphabet.java
@@ -16,10 +16,14 @@
 
 package com.android.internal.telephony;
 
+import android.text.TextUtils;
 import android.util.SparseIntArray;
 
 import android.util.Log;
 
+import java.nio.ByteBuffer;
+import java.nio.charset.Charset;
+
 /**
  * This class implements the character set mapping between
  * the GSM SMS 7-bit alphabet specified in TS 23.038 6.2.1
@@ -354,6 +358,32 @@ public class GsmAlphabet {
      */
     public static String
     gsm8BitUnpackedToString(byte[] data, int offset, int length) {
+        return gsm8BitUnpackedToString(data, offset, length, "");
+    }
+
+    /**
+     * Convert a GSM alphabet string that's stored in 8-bit unpacked
+     * format (as it often appears in SIM records) into a String
+     *
+     * Field may be padded with trailing 0xff's. The decode stops
+     * at the first 0xff encountered.
+     *
+     * Additionally, in some country(ex. Korea), there are non-ASCII or MBCS characters.
+     * If a character set is given, characters in data are treat as MBCS.
+     */
+    public static String
+    gsm8BitUnpackedToString(byte[] data, int offset, int length, String characterset) {
+        boolean isMbcs = false;
+        Charset charset = null;
+        ByteBuffer mbcsBuffer = null;
+
+        if (!TextUtils.isEmpty(characterset)
+                && !characterset.equalsIgnoreCase("us-ascii")
+                && Charset.isSupported(characterset)) {
+            isMbcs = true;
+            charset = Charset.forName(characterset);
+            mbcsBuffer = ByteBuffer.allocate(2);
+        }
         boolean prevWasEscape;
         StringBuilder ret = new StringBuilder(length);
 
@@ -379,7 +409,15 @@ public class GsmAlphabet {
                 if (prevWasEscape) {
                     ret.append((char)gsmExtendedToChar.get(c, ' '));
                 } else {
-                    ret.append((char)gsmToChar.get(c, ' '));
+                    if (!isMbcs || c < 0x80 || i + 1 >= offset + length) {
+                        ret.append((char)gsmToChar.get(c, ' '));
+                    } else {
+                        // isMbcs must be true. So both mbcsBuffer and charset are initialized.
+                        mbcsBuffer.clear();
+                        mbcsBuffer.put(data, i++, 2);
+                        mbcsBuffer.flip();
+                        ret.append(charset.decode(mbcsBuffer).toString());
+                    }
                 }
                 prevWasEscape = false;
             }
diff --git a/telephony/java/com/android/internal/telephony/IccUtils.java b/telephony/java/com/android/internal/telephony/IccUtils.java
index 005ae37..df579b0 100644
--- a/telephony/java/com/android/internal/telephony/IccUtils.java
+++ b/telephony/java/com/android/internal/telephony/IccUtils.java
@@ -16,13 +16,16 @@
 
 package com.android.internal.telephony;
 
+import android.content.res.Resources;
+import android.content.res.Resources.NotFoundException;
 import android.graphics.Bitmap;
 import android.graphics.Color;
 import android.util.Log;
 
 import com.android.internal.telephony.GsmAlphabet;
-
 import java.io.UnsupportedEncodingException;
+import java.nio.ByteBuffer;
+import java.nio.charset.Charset;
 
 /**
  * Various methods, useful for dealing with SIM data.
@@ -150,6 +153,9 @@ public class IccUtils {
      */
     public static String
     adnStringFieldToString(byte[] data, int offset, int length) {
+        if (length == 0) {
+            return "";
+        }
         if (length >= 1) {
             if (data[offset] == (byte) 0x80) {
                 int ucslen = (length - 1) / 2;
@@ -225,7 +231,15 @@ public class IccUtils {
             return ret.toString();
         }
 
-        return GsmAlphabet.gsm8BitUnpackedToString(data, offset, length);
+        Resources resource = Resources.getSystem();
+        String defaultCharset = "";
+        try {
+            defaultCharset =
+                    resource.getString(com.android.internal.R.string.gsm_alphabet_default_charset);
+        } catch (NotFoundException e) {
+            // Ignore Exception and defaultCharset is set to a empty string.
+        }
+        return GsmAlphabet.gsm8BitUnpackedToString(data, offset, length, defaultCharset.trim());
     }
 
     static int
diff --git a/telephony/java/com/android/internal/telephony/gsm/SmsMessage.java b/telephony/java/com/android/internal/telephony/gsm/SmsMessage.java
index a77484a..e24613f 100644
--- a/telephony/java/com/android/internal/telephony/gsm/SmsMessage.java
+++ b/telephony/java/com/android/internal/telephony/gsm/SmsMessage.java
@@ -33,6 +33,7 @@ import java.io.UnsupportedEncodingException;
 import static android.telephony.SmsMessage.ENCODING_7BIT;
 import static android.telephony.SmsMessage.ENCODING_8BIT;
 import static android.telephony.SmsMessage.ENCODING_16BIT;
+import static android.telephony.SmsMessage.ENCODING_KSC5601;
 import static android.telephony.SmsMessage.ENCODING_UNKNOWN;
 import static android.telephony.SmsMessage.MAX_USER_DATA_BYTES;
 import static android.telephony.SmsMessage.MAX_USER_DATA_BYTES_WITH_HEADER;
@@ -776,6 +777,27 @@ public class SmsMessage extends SmsMessageBase {
             return ret;
         }
 
+        /**
+         * Interprets the user data payload as KSC-5601 characters, and
+         * decodes them into a String.
+         *
+         * @param byteCount the number of bytes in the user data payload
+         * @return a String with the decoded characters
+         */
+        String getUserDataKSC5601(int byteCount) {
+            String ret;
+
+            try {
+                ret = new String(pdu, cur, byteCount, "KSC5601");
+            } catch (UnsupportedEncodingException ex) {
+                ret = "";
+                Log.e(LOG_TAG, "implausible UnsupportedEncodingException", ex);
+            }
+
+            cur += byteCount;
+            return ret;
+        }
+
         boolean moreDataPresent() {
             return (pdu.length > cur);
         }
@@ -1111,6 +1133,16 @@ public class SmsMessage extends SmsMessageBase {
                 Log.w(LOG_TAG, "MWI for fax, email, or other "
                         + (dataCodingScheme & 0xff));
             }
+        } else if ((dataCodingScheme & 0xC0) == 0x80) {
+            // 3GPP TS 23.038 V7.0.0 (2006-03) section 4
+            // 0x80..0xBF == Reserved coding groups
+            if (dataCodingScheme == 0x84) {
+                // This value used for KSC5601 by carriers in Korea.
+                encodingType = ENCODING_KSC5601;
+            } else {
+                Log.w(LOG_TAG, "5 - Unsupported SMS data coding scheme "
+                        + (dataCodingScheme & 0xff));
+            }
         } else {
             Log.w(LOG_TAG, "3 - Unsupported SMS data coding scheme "
                     + (dataCodingScheme & 0xff));
@@ -1135,6 +1167,10 @@ public class SmsMessage extends SmsMessageBase {
         case ENCODING_16BIT:
             messageBody = p.getUserDataUCS2(count);
             break;
+
+        case ENCODING_KSC5601:
+            messageBody = p.getUserDataKSC5601(count);
+            break;
         }
 
         if (Config.LOGV) Log.v(LOG_TAG, "SMS message body (raw): '" + messageBody + "'");
diff --git a/telephony/tests/telephonytests/src/com/android/internal/telephony/GsmAlphabetTest.java b/telephony/tests/telephonytests/src/com/android/internal/telephony/GsmAlphabetTest.java
index a6b9a2a..7011aeb 100644
--- a/telephony/tests/telephonytests/src/com/android/internal/telephony/GsmAlphabetTest.java
+++ b/telephony/tests/telephonytests/src/com/android/internal/telephony/GsmAlphabetTest.java
@@ -16,6 +16,8 @@
 
 package com.android.internal.telephony;
 
+import com.android.internal.telephony.GsmAlphabet;
+
 import junit.framework.TestCase;
 
 import android.test.suitebuilder.annotation.LargeTest;
@@ -307,4 +309,26 @@ public class GsmAlphabetTest extends TestCase {
         assertEquals("a",
                 GsmAlphabet.gsm8BitUnpackedToString(unpacked, 1, unpacked.length - 1));
     }
+
+    @SmallTest
+    public void testGsm8BitUpackedWithEuckr() throws Exception {
+        // Some feature phones in Korea store contacts as euc-kr.
+        // Test this situations.
+        byte unpacked[];
+
+        // Test general alphabet strings.
+        unpacked = IccUtils.hexStringToBytes("61626320646566FF");
+        assertEquals("abc def",
+                GsmAlphabet.gsm8BitUnpackedToString(unpacked, 0, unpacked.length, "euc-kr"));
+
+        // Test korean strings.
+        unpacked = IccUtils.hexStringToBytes("C5D7BDBAC6AEFF");
+        assertEquals("\uD14C\uC2A4\uD2B8",
+                GsmAlphabet.gsm8BitUnpackedToString(unpacked, 0, unpacked.length, "euc-kr"));
+
+        // Test gsm Extented Characters.
+        unpacked = GsmAlphabet.stringToGsm8BitPacked(sGsmExtendedChars);
+        assertEquals(sGsmExtendedChars,
+                GsmAlphabet.gsm8BitUnpackedToString(unpacked, 0, unpacked.length, "euc-kr"));
+    }
 }