Attempted patch to fix marshaling of symbolic names we store as java.lang.String

patch.diff

diff --git a/core/src/main/java/org/jruby/runtime/marshal/MarshalStream.java b/core/src/main/java/org/jruby/runtime/marshal/MarshalStream.java
index 4b14a54..b1b7a9a 100644
--- a/core/src/main/java/org/jruby/runtime/marshal/MarshalStream.java
+++ b/core/src/main/java/org/jruby/runtime/marshal/MarshalStream.java
@@ -46,6 +46,7 @@ import org.jruby.RubyArray;
 import org.jruby.RubyBignum;
 import org.jruby.RubyBoolean;
 import org.jruby.RubyClass;
+import org.jruby.RubyEncoding;
 import org.jruby.RubyFixnum;
 import org.jruby.RubyFloat;
 import org.jruby.RubyHash;
@@ -136,6 +137,11 @@ public class MarshalStream extends FilterOutputStream {
         return fixnum.getLongValue() <= RubyFixnum.MAX_MARSHAL_FIXNUM && fixnum.getLongValue() >= RubyFixnum.MIN_MARSHAL_FIXNUM;
     }
 
+    private void dumpName(String name) throws IOException {
+        // forcing this always to UTF-8 is not right, but we don't have original source's encoding at this point
+        dumpObject(runtime.newSymbol(name, UTF8Encoding.INSTANCE));
+    }
+
     private void writeAndRegisterSymbol(String sym) throws IOException {
         if (cache.isSymbolRegistered(sym)) {
             cache.writeSymbolLink(this, sym);
@@ -330,7 +336,7 @@ public class MarshalStream extends FilterOutputStream {
         registerLinkTarget(value);
         write(TYPE_USRMARSHAL);
         RubyClass metaclass = value.getMetaClass().getRealClass();
-        writeAndRegisterSymbol(metaclass.getName());
+        dumpName(metaclass.getName());
 
         IRubyObject marshaled;
         if (method != null) {
@@ -372,7 +378,7 @@ public class MarshalStream extends FilterOutputStream {
         write(TYPE_USERDEF);
         RubyClass metaclass = value.getMetaClass().getRealClass();
 
-        writeAndRegisterSymbol(metaclass.getName());
+        dumpName(metaclass.getName());
 
         writeString(marshaled.getByteList());
 
@@ -392,7 +398,7 @@ public class MarshalStream extends FilterOutputStream {
         }
         
         // w_symbol
-        writeAndRegisterSymbol(type.getName());
+        dumpName(type.getName());
     }
     
     public void dumpVariablesWithEncoding(List<Variable<Object>> vars, IRubyObject obj) throws IOException {
@@ -414,7 +420,7 @@ public class MarshalStream extends FilterOutputStream {
     private void dumpVariablesShared(List<Variable<Object>> vars) throws IOException {
         for (Variable<Object> var : vars) {
             if (var.getValue() instanceof IRubyObject) {
-                writeAndRegisterSymbol(var.getName());
+                dumpName(var.getName());
                 dumpObject((IRubyObject)var.getValue());
             }
         }
@@ -473,9 +479,9 @@ public class MarshalStream extends FilterOutputStream {
     }
 
     public void writeString(String value) throws IOException {
-        writeInt(value.length());
-        // FIXME: should preserve unicode?
-        out.write(RubyString.stringToBytes(value));
+        byte[] bytes = value.getBytes(RubyEncoding.UTF8);
+        writeInt(bytes.length);
+        out.write(bytes);
     }
 
     public void writeString(ByteList value) throws IOException {
diff --git a/core/src/main/java/org/jruby/runtime/marshal/UnmarshalStream.java b/core/src/main/java/org/jruby/runtime/marshal/UnmarshalStream.java
index d731ac7..ba715ca 100644
--- a/core/src/main/java/org/jruby/runtime/marshal/UnmarshalStream.java
+++ b/core/src/main/java/org/jruby/runtime/marshal/UnmarshalStream.java
@@ -320,6 +320,23 @@ public class UnmarshalStream extends InputStream {
         return new ByteList(buffer,false);
     }
 
+    public String unmarshalSymbol() throws IOException {
+        int length = unmarshalInt();
+        byte[] buffer = new byte[length];
+
+        int readLength = 0;
+        while (readLength < length) {
+            int read = inputStream.read(buffer, readLength, length - readLength);
+
+            if (read == -1) {
+                throw getRuntime().newArgumentError("marshal data too short");
+            }
+            readLength += read;
+        }
+
+        return new String(buffer, RubyEncoding.UTF8);
+    }
+
     public int unmarshalInt() throws IOException {
         int c = readSignedByte();
         if (c == 0) {