Add support for Format41c (type references)

6 files changed, 38 insertions, 8 deletions

diff --git a/smali/src/main/antlr3/org/jf/smali/smaliLexer.g b/smali/src/main/antlr3/org/jf/smali/smaliLexer.g
index c07970ea..07c03167 100644
--- a/smali/src/main/antlr3/org/jf/smali/smaliLexer.g
+++ b/smali/src/main/antlr3/org/jf/smali/smaliLexer.g
@@ -690,6 +690,11 @@ INSTRUCTION_FORMAT3rms_METHOD
 	:	'invoke-virtual-quick/range'
 	|	'invoke-super-quick/range';
 
+INSTRUCTION_FORMAT41c_TYPE
+	:	'check-cast/jumbo'
+	|	'new-instance/jumbo'
+	|	'const-class/jumbo';
+
 INSTRUCTION_FORMAT51l
 	:	'const-wide';
 
diff --git a/smali/src/main/antlr3/org/jf/smali/smaliParser.g b/smali/src/main/antlr3/org/jf/smali/smaliParser.g
index e69e4111..a4fe4598 100644
--- a/smali/src/main/antlr3/org/jf/smali/smaliParser.g
+++ b/smali/src/main/antlr3/org/jf/smali/smaliParser.g
@@ -117,6 +117,7 @@ tokens {
 	I_STATEMENT_FORMAT35c_TYPE;
 	I_STATEMENT_FORMAT3rc_METHOD;
 	I_STATEMENT_FORMAT3rc_TYPE;
+	I_STATEMENT_FORMAT41c_TYPE;
 	I_STATEMENT_FORMAT51l;
 	I_STATEMENT_ARRAY_DATA;
 	I_STATEMENT_PACKED_SWITCH;
@@ -703,13 +704,14 @@ instruction returns [int size]
 			}
 		}
 		-> ^(I_STATEMENT_FORMAT20bc INSTRUCTION_FORMAT20bc VERIFICATION_ERROR_TYPE verification_error_reference)
+		//TODO: check if dalvik has a jumbo version of throw-verification-error
 	|	//e.g. goto/16 endloop:
 		INSTRUCTION_FORMAT20t label_ref_or_offset {$size = Format.Format20t.size;}
 		-> ^(I_STATEMENT_FORMAT20t[$start, "I_STATEMENT_FORMAT20t"] INSTRUCTION_FORMAT20t label_ref_or_offset)
-	|	//e.g. sget-object v0 java/lang/System/out LJava/io/PrintStream;
+	|	//e.g. sget-object v0, java/lang/System/out LJava/io/PrintStream;
 		INSTRUCTION_FORMAT21c_FIELD REGISTER COMMA fully_qualified_field {$size = Format.Format21c.size;}
 		-> ^(I_STATEMENT_FORMAT21c_FIELD[$start, "I_STATEMENT_FORMAT21c_FIELD"] INSTRUCTION_FORMAT21c_FIELD REGISTER fully_qualified_field)
-	|	//e.g. sget-object-volatile v0 java/lang/System/out LJava/io/PrintStream;
+	|	//e.g. sget-object-volatile v0, java/lang/System/out LJava/io/PrintStream;
 		INSTRUCTION_FORMAT21c_FIELD_ODEX REGISTER COMMA fully_qualified_field {$size = Format.Format21c.size;}
 		{
 			if (!allowOdex) {
@@ -717,10 +719,10 @@ instruction returns [int size]
 			}
 		}
 		-> ^(I_STATEMENT_FORMAT21c_FIELD[$start, "I_STATEMENT_FORMAT21c_FIELD"] INSTRUCTION_FORMAT21c_FIELD_ODEX REGISTER fully_qualified_field)
-	|	//e.g. const-string v1 "Hello World!"
+	|	//e.g. const-string v1, "Hello World!"
 		INSTRUCTION_FORMAT21c_STRING REGISTER COMMA STRING_LITERAL {$size = Format.Format21c.size;}
 		-> ^(I_STATEMENT_FORMAT21c_STRING[$start, "I_STATEMENT_FORMAT21c_STRING"] INSTRUCTION_FORMAT21c_STRING REGISTER STRING_LITERAL)
-	|	//e.g. const-class v2 Lorg/jf/HelloWorld2/HelloWorld2;
+	|	//e.g. const-class v2, Lorg/jf/HelloWorld2/HelloWorld2;
 		INSTRUCTION_FORMAT21c_TYPE REGISTER COMMA reference_type_descriptor {$size = Format.Format21c.size;}
 		-> ^(I_STATEMENT_FORMAT21c_TYPE[$start, "I_STATEMENT_FORMAT21c"] INSTRUCTION_FORMAT21c_TYPE REGISTER reference_type_descriptor)
 	|	//e.g. const/high16 v1, 1234
@@ -833,6 +835,9 @@ instruction returns [int size]
 		{
 			throwOdexedInstructionException(input, $INSTRUCTION_FORMAT3rms_METHOD.text);
 		}
+	|	//e.g. const-class/jumbo v2, Lorg/jf/HelloWorld2/HelloWorld2;
+		INSTRUCTION_FORMAT41c_TYPE REGISTER COMMA reference_type_descriptor {$size = Format.Format41c.size;}
+		-> ^(I_STATEMENT_FORMAT41c_TYPE[$start, "I_STATEMENT_FORMAT41c"] INSTRUCTION_FORMAT41c_TYPE REGISTER reference_type_descriptor)
 	|	//e.g. const-wide v0, 5000000000L
 		INSTRUCTION_FORMAT51l REGISTER COMMA fixed_literal {$size = Format.Format51l.size;}
 		-> ^(I_STATEMENT_FORMAT51l[$start, "I_STATEMENT_FORMAT51l"] INSTRUCTION_FORMAT51l REGISTER fixed_literal)
diff --git a/smali/src/main/antlr3/org/jf/smali/smaliTreeWalker.g b/smali/src/main/antlr3/org/jf/smali/smaliTreeWalker.g
index b6542ca3..30a714dc 100644
--- a/smali/src/main/antlr3/org/jf/smali/smaliTreeWalker.g
+++ b/smali/src/main/antlr3/org/jf/smali/smaliTreeWalker.g
@@ -998,7 +998,7 @@ instruction[int totalMethodRegisters, int methodParameterRegisters, List<Instruc
 
 			$instructions.add(new Instruction20t(opcode, addressOffset));
 		}
-	|	//e.g. sget_object v0 java/lang/System/out LJava/io/PrintStream;
+	|	//e.g. sget_object v0, java/lang/System/out LJava/io/PrintStream;
 		^(I_STATEMENT_FORMAT21c_FIELD inst=(INSTRUCTION_FORMAT21c_FIELD | INSTRUCTION_FORMAT21c_FIELD_ODEX) REGISTER fully_qualified_field)
 		{
 			Opcode opcode = Opcode.getOpcodeByName($inst.text);
@@ -1008,7 +1008,7 @@ instruction[int totalMethodRegisters, int methodParameterRegisters, List<Instruc
 
 			$instructions.add(new Instruction21c(opcode, regA, fieldIdItem));
 		}
-	|	//e.g. const-string v1 "Hello World!"
+	|	//e.g. const-string v1, "Hello World!"
 		^(I_STATEMENT_FORMAT21c_STRING INSTRUCTION_FORMAT21c_STRING REGISTER string_literal)
 		{
 			Opcode opcode = Opcode.getOpcodeByName($INSTRUCTION_FORMAT21c_STRING.text);
@@ -1018,7 +1018,7 @@ instruction[int totalMethodRegisters, int methodParameterRegisters, List<Instruc
 
 			instructions.add(new Instruction21c(opcode, regA, stringIdItem));
 		}
-	|	//e.g. const-class v2 org/jf/HelloWorld2/HelloWorld2
+	|	//e.g. const-class v2, org/jf/HelloWorld2/HelloWorld2
 		^(I_STATEMENT_FORMAT21c_TYPE INSTRUCTION_FORMAT21c_TYPE REGISTER reference_type_descriptor)
 		{
 			Opcode opcode = Opcode.getOpcodeByName($INSTRUCTION_FORMAT21c_TYPE.text);
@@ -1074,7 +1074,7 @@ instruction[int totalMethodRegisters, int methodParameterRegisters, List<Instruc
 
 			$instructions.add(new Instruction22b(opcode, regA, regB, (byte)litC));
 		}
-	|	//e.g. iput-object v1 v0 org/jf/HelloWorld2/HelloWorld2.helloWorld Ljava/lang/String;
+	|	//e.g. iput-object v1, v0, org/jf/HelloWorld2/HelloWorld2.helloWorld Ljava/lang/String;
 		^(I_STATEMENT_FORMAT22c_FIELD inst=(INSTRUCTION_FORMAT22c_FIELD | INSTRUCTION_FORMAT22c_FIELD_ODEX) registerA=REGISTER registerB=REGISTER fully_qualified_field)
 		{
 			Opcode opcode = Opcode.getOpcodeByName($inst.text);
@@ -1249,6 +1249,16 @@ instruction[int totalMethodRegisters, int methodParameterRegisters, List<Instruc
 
 			$instructions.add(new Instruction3rc(opcode, (short)registerCount, startRegister, typeIdItem));
 		}
+	|	//e.g. const-class/jumbo v2, org/jf/HelloWorld2/HelloWorld2
+		^(I_STATEMENT_FORMAT41c_TYPE INSTRUCTION_FORMAT41c_TYPE REGISTER reference_type_descriptor)
+		{
+			Opcode opcode = Opcode.getOpcodeByName($INSTRUCTION_FORMAT41c_TYPE.text);
+			int regA = parseRegister_short($REGISTER.text, $totalMethodRegisters, $methodParameterRegisters);
+
+			TypeIdItem typeIdItem = $reference_type_descriptor.type;
+
+			$instructions.add(new Instruction41c(opcode, regA, typeIdItem));
+		}
 	|	//e.g. const-wide v0, 5000000000L
 		^(I_STATEMENT_FORMAT51l INSTRUCTION_FORMAT51l REGISTER fixed_64bit_literal)
 		{
diff --git a/smali/src/main/jflex/smaliLexer.flex b/smali/src/main/jflex/smaliLexer.flex
index 8697b8d7..2cc83e1a 100644
--- a/smali/src/main/jflex/smaliLexer.flex
+++ b/smali/src/main/jflex/smaliLexer.flex
@@ -562,6 +562,10 @@ Type = {PrimitiveType} | {ClassDescriptor} | {ArrayDescriptor}
         return newToken(INSTRUCTION_FORMAT3rms_METHOD);
     }
 
+    "check-cast/jumbo" | "new-instance/jumbo" | "const-class/jumbo" {
+        return newToken(INSTRUCTION_FORMAT41c_TYPE);
+    }
+
     "const-wide" {
         return newToken(INSTRUCTION_FORMAT51l);
     }
diff --git a/smali/src/test/resources/LexerTest/InstructionTest.smali b/smali/src/test/resources/LexerTest/InstructionTest.smali
index 7a7c3352..433b0017 100644
--- a/smali/src/test/resources/LexerTest/InstructionTest.smali
+++ b/smali/src/test/resources/LexerTest/InstructionTest.smali
@@ -222,4 +222,7 @@ filled-new-array/range
 execute-inline/range
 invoke-virtual-quick/range
 invoke-super-quick/range
+check-cast/jumbo
+new-instance/jumbo
+const-class/jumbo
 const-wide
\ No newline at end of file
diff --git a/smali/src/test/resources/LexerTest/InstructionTest.tokens b/smali/src/test/resources/LexerTest/InstructionTest.tokens
index 23e48956..435e7cfa 100644
--- a/smali/src/test/resources/LexerTest/InstructionTest.tokens
+++ b/smali/src/test/resources/LexerTest/InstructionTest.tokens
@@ -222,4 +222,7 @@ INSTRUCTION_FORMAT3rc_TYPE("filled-new-array/range")
 INSTRUCTION_FORMAT3rmi_METHOD("execute-inline/range")
 INSTRUCTION_FORMAT3rms_METHOD("invoke-virtual-quick/range")
 INSTRUCTION_FORMAT3rms_METHOD("invoke-super-quick/range")
+INSTRUCTION_FORMAT41c_TYPE("check-cast/jumbo")
+INSTRUCTION_FORMAT41c_TYPE("new-instance/jumbo")
+INSTRUCTION_FORMAT41c_TYPE("const-class/jumbo")
 INSTRUCTION_FORMAT51l("const-wide")
\ No newline at end of file