jruby · Mar 13, 2015 · Mar 13, 2015 · Mar 13, 2015 · Mar 16, 2015
Showing with 52 additions and 10 deletions.

+28 −3 src/org/jcodings/Encoding.java

+18 −2 src/org/jcodings/EncodingDB.java

+2 −2 src/org/jcodings/EncodingList.java

+1 −1 src/org/jcodings/MultiByteEncoding.java

+3 −2 src/org/jcodings/transcode/TranscodeFunctions.java
diff --git a/src/org/jcodings/Encoding.java b/src/org/jcodings/Encoding.java
@@ -33,12 +33,13 @@ public abstract class Encoding implements Cloneable {
     private static int count;
 
     protected final int minLength, maxLength;
-    protected final boolean isFixedWidth, isSingleByte, isDummy, isAsciiCompatible;
+    protected final boolean isFixedWidth, isSingleByte, isAsciiCompatible;
 
     protected byte[]name;
     protected int hashCode;
     private int index;
     protected Charset charset = null;
+    protected boolean isDummy;
 
     protected Encoding(String name, int minLength, int maxLength, boolean isDummy) {
         setName(name);
@@ -67,6 +68,10 @@ protected final void setName(byte[]name) {
         this.hashCode = BytesHash.hashCode(this.name, 0, this.name.length);
     }
 
+    protected final void setDummy(boolean dummy) {
+        this.isDummy = dummy;
+    }
+
     @Override
     public final String toString() {
         return new String(name);
@@ -120,9 +125,14 @@ public String getCharsetName() {
     }
 
     public Encoding replicate(byte[]name) {
+        return replicate(name, false);
+    }
+
+    public Encoding replicate(byte[]name, boolean dummy) {
         try {
             Encoding clone = (Encoding)clone();
             clone.setName(name);
+            clone.setDummy(dummy);
             clone.index = count++;
             return clone;
         } catch (CloneNotSupportedException cnse){
@@ -535,18 +545,33 @@ public final boolean isFixedWidth() {
     public static final byte NEW_LINE = (byte)0x0a;
 
     public static Encoding load(String name) {
-        String encClassName = "org.jcodings.specific." + name + "Encoding";
+        return loadDummyOrEncoding(name, false);
+    }
 
+    public static Encoding loadForDummy(String name) {
+        return loadDummyOrEncoding(name, true);
+    }
+
+    private static Encoding loadDummyOrEncoding(String name, boolean dummy) {
+        String encClassName = "org.jcodings.specific." + name + "Encoding";
         Class<?> encClass;
         try {
             encClass = Class.forName(encClassName);
         } catch (ClassNotFoundException cnfe) {
             throw new InternalException(ErrorMessages.ERR_ENCODING_CLASS_DEF_NOT_FOUND, encClassName);
         }
 
+        if (dummy) {
+            try {
+                return (Encoding) encClass.getField("DUMMY").get(encClass);
+            } catch (Exception e) {
+                // try to fall back on normal instance
+            }
+        }
+
         try {
             return (Encoding)encClass.getField("INSTANCE").get(encClass);
-        } catch (Exception e) {
+        } catch (Exception e2) {
             throw new InternalException(ErrorMessages.ERR_ENCODING_LOAD_ERROR, encClassName);
         }
     }

diff --git a/src/org/jcodings/EncodingDB.java b/src/org/jcodings/EncodingDB.java
@@ -58,6 +58,11 @@ private Entry (byte[]name, String encodingClass, Entry base, boolean isDummy) {
             this(name, ascii.encodingClass, ascii, true);
         }
 
+        // dummy with base
+        Entry(byte[]name, Entry base, boolean dummy) {
+            this(name, base.encodingClass, base, dummy);
+        }
+
         @Override
         public int hashCode() {
             return encodingClass.hashCode();
@@ -73,7 +78,7 @@ public Encoding getEncoding() {
                     encoding = Encoding.load(encodingClass);
                 } else {
                     if (isDummy) {
-                        encoding = ASCIIEncoding.DUMMY.replicate(name);
+                        encoding = Encoding.loadForDummy(encodingClass).replicate(name, isDummy);
                     } else {
                         encoding = Encoding.load(encodingClass).replicate(name);
                     }
@@ -166,12 +171,16 @@ public static void alias(String alias, String original) {
     }
 
     public static void replicate(String replica, String original) {
+        replicate(replica, original, false);
+    }
+
+    private static void replicate(String replica, String original, boolean dummy) {
         byte[]origBytes = original.getBytes();
         Entry originalEntry = encodings.get(origBytes);
         if (originalEntry == null) throw new InternalException(ErrorMessages.ERR_NO_SUCH_ENCODNG, original);
         byte[]replicaBytes = replica.getBytes();
         if (encodings.get(replicaBytes) != null) throw new InternalException(ErrorMessages.ERR_ENCODING_REPLICA_ALREADY_REGISTERED, replica);
-        encodings.putDirect(replicaBytes, new Entry(replicaBytes, originalEntry));
+        encodings.putDirect(replicaBytes, new Entry(replicaBytes, originalEntry, dummy));
     }
 
     public static void set_base(String name, String original) {
@@ -188,6 +197,10 @@ public static void dummy(String name) {
         dummy(name.getBytes());
     }
 
+    public static void dummyUnicode(String replica) {
+        replicate(replica, replica + "BE", true);
+    }
+
     static {
         for (int i = 0; i < builtin.length / 2; i++) {
             declare(builtin[i << 1], builtin[(i << 1) + 1]);
@@ -213,6 +226,9 @@ public static void dummy(String name) {
             case 'D':
                 dummy(enc[1]);
                 break;
+            case 'U': // ENC_DUMMY_UNICODE from encdb.c
+                dummyUnicode(enc[1]);
+                break;
             default:
                 Thread.dumpStack();
                 throw new InternalException("Unknown flag: " + enc[0].charAt(0));

diff --git a/src/org/jcodings/EncodingList.java b/src/org/jcodings/EncodingList.java
@@ -133,8 +133,8 @@ final class EncodingList {
         {"R", "UTF8-MAC", "UTF-8"},
         {"A", "UTF-8-MAC", "UTF8-MAC"},
         {"A", "UTF-8-HFS", "UTF8-MAC" /* Emacs 23.2 */},
-        {"R", "UTF-16", "UTF-16BE"},
-        {"R", "UTF-32", "UTF-32BE"},
+        {"U", "UTF-16"},
+        {"U", "UTF-32"},
         {"A", "UCS-2BE", "UTF-16BE"},
         {"A", "UCS-4BE", "UTF-32BE"},
         {"A", "UCS-4LE", "UTF-32LE"},

diff --git a/src/org/jcodings/MultiByteEncoding.java b/src/org/jcodings/MultiByteEncoding.java
@@ -187,7 +187,7 @@ protected final int mb4CodeToMbc(int code, byte[]bytes, int p) {
     }
 
     protected final boolean mb2IsCodeCType(int code, int ctype) {
-        if (code < 128) {
+        if ((code & 0xFFFFFFFFL) < 128) { // unsigned 32-bit
             return isCodeCTypeInternal(code, ctype); // configured with ascii
         } else {
             if (isWordGraphPrint(ctype)) {

diff --git a/src/org/jcodings/transcode/TranscodeFunctions.java b/src/org/jcodings/transcode/TranscodeFunctions.java
@@ -165,16 +165,17 @@ public static int funSiFromUTF32(byte[] statep, byte[] s, int sStart, int l) {
         switch (sp[0]) {
             case 0:
                 s3 = s[sStart+3] & 0xFF;
-                if (s0 == 0 && s1 == 0 && s2 == 0xFE && s3 == 0xEE) {
+                if (s0 == 0 && s1 == 0 && s2 == 0xFE && s3 == 0xFF) {
                     sp[0] = BE;
                     return TranscodingInstruction.ZERObt;
                 } else if (s0 == 0xFF && s1 == 0xFE && s2 == 0 && s3 == 0) {
                     sp[0] = LE;
                     return TranscodingInstruction.ZERObt;
                 }
+                break;
             case BE:
                 if (s0 == 0 && ((0 < s1 && s1 <= 0x10)) ||
-                        (s1 == 0 && (s2 < 0xD8 && 0xDF < s2))) {
+                        (s1 == 0 && (s2 < 0xD8 || 0xDF < s2))) {
                     return TranscodingInstruction.FUNso;
                 }
                 break;