Tweak how array descriptors are parsed

This is needed to allow smalidea to have an outer array type element to
represent the type as a whole, and an inner class type element for the
actual element type

smali/src/main/antlr3/smaliParser.g

@@ -39,7 +39,7 @@ tokens {
   ANNOTATION_DIRECTIVE;
   ANNOTATION_VISIBILITY;
   ARRAY_DATA_DIRECTIVE;
-  ARRAY_DESCRIPTOR;
+  ARRAY_TYPE_PREFIX;
   ARROW;
   BOOL_LITERAL;
   BYTE_LITERAL;
@@ -369,16 +369,12 @@ import org.jf.dexlib2.Opcodes;
       case '[':
       {
         int i = typeStartIndex;
-            while (str.charAt(++i) == '[');
+        while (str.charAt(++i) == '[');
 
-            if (str.charAt(i++) == 'L') {
+        token.setType(ARRAY_TYPE_PREFIX);
-                while (str.charAt(i++) != ';');
+        token.setText(str.substring(typeStartIndex, i));
-        }
+        token.setStopIndex(baseToken.getStartIndex() + i - 1);
-
+        break;
-            token.setType(ARRAY_DESCRIPTOR);
-            token.setText(str.substring(typeStartIndex, i));
-            token.setStopIndex(baseToken.getStartIndex() + i - 1);
-            break;
       }
       default:
         throw new RuntimeException(String.format("Invalid character '\%c' in param list \"\%s\" at position \%d", str.charAt(typeStartIndex), str, typeStartIndex));
@@ -591,20 +587,23 @@ param_list
   | PARAM_LIST_OR_ID_START PRIMITIVE_TYPE* PARAM_LIST_OR_ID_END -> PRIMITIVE_TYPE*
   | nonvoid_type_descriptor*;
 
+array_descriptor
+  : ARRAY_TYPE_PREFIX (PRIMITIVE_TYPE | CLASS_DESCRIPTOR);
+
 type_descriptor
   : VOID_TYPE
   | PRIMITIVE_TYPE
   | CLASS_DESCRIPTOR
-  | ARRAY_DESCRIPTOR;
+  | array_descriptor;
 
 nonvoid_type_descriptor
   : PRIMITIVE_TYPE
   | CLASS_DESCRIPTOR
-  | ARRAY_DESCRIPTOR;
+  | array_descriptor;
 
 reference_type_descriptor
   : CLASS_DESCRIPTOR
-  | ARRAY_DESCRIPTOR;
+  | array_descriptor;
 
 integer_literal
   : POSITIVE_INTEGER_LITERAL -> INTEGER_LITERAL[$POSITIVE_INTEGER_LITERAL]

smali/src/main/antlr3/smaliTreeWalker.g

@@ -1146,20 +1146,20 @@ insn_sparse_switch_directive
       $method::methodBuilder.addInstruction(new BuilderSparseSwitchPayload($sparse_switch_elements.elements));
     };
 
+array_descriptor returns [String type]
+  : ARRAY_TYPE_PREFIX ( PRIMITIVE_TYPE { $type = $ARRAY_TYPE_PREFIX.text + $PRIMITIVE_TYPE.text; }
+                      | CLASS_DESCRIPTOR { $type = $ARRAY_TYPE_PREFIX.text + $CLASS_DESCRIPTOR.text; });
+
 nonvoid_type_descriptor returns [String type]
-  : (PRIMITIVE_TYPE
+  : (PRIMITIVE_TYPE { $type = $text; }
-  | CLASS_DESCRIPTOR
+  | CLASS_DESCRIPTOR { $type = $text; }
-  | ARRAY_DESCRIPTOR)
+  | array_descriptor { $type = $array_descriptor.type; })
-  {
+  ;
-    $type = $start.getText();
-  };
 
 reference_type_descriptor returns [String type]
-  : (CLASS_DESCRIPTOR
+  : (CLASS_DESCRIPTOR { $type = $text; }
-  | ARRAY_DESCRIPTOR)
+  | array_descriptor { $type = $array_descriptor.type; })
-  {
+  ;
-    $type = $start.getText();
-  };
 
 type_descriptor returns [String type]
   : VOID_TYPE {$type = "V";}

smali/src/main/jflex/smaliLexer.flex

@@ -231,13 +231,14 @@ PrimitiveType = [ZBSCIJFD]
 
 ClassDescriptor = L ({SimpleName} "/")* {SimpleName} ;
 
-ArrayDescriptor = "[" + ({PrimitiveType} | {ClassDescriptor})
+ArrayPrefix = "["+
 
-Type = {PrimitiveType} | {ClassDescriptor} | {ArrayDescriptor}
+Type = {PrimitiveType} | {ClassDescriptor} | {ArrayPrefix} ({ClassDescriptor} | {PrimitiveType})
 
 
 %state PARAM_LIST_OR_ID
 %state PARAM_LIST
+%state ARRAY_DESCRIPTOR
 %state STRING
 %state CHAR
 
@@ -314,7 +315,7 @@ Type = {PrimitiveType} | {ClassDescriptor} | {ArrayDescriptor}
 <PARAM_LIST> {
     {PrimitiveType} { return newToken(PRIMITIVE_TYPE); }
     {ClassDescriptor} { return newToken(CLASS_DESCRIPTOR); }
-    {ArrayDescriptor} { return newToken(ARRAY_DESCRIPTOR); }
+    {ArrayPrefix} { return newToken(ARRAY_TYPE_PREFIX); }
     [^] { yypushback(1); yybegin(YYINITIAL); return newToken(PARAM_LIST_END); }
     <<EOF>> { yybegin(YYINITIAL); return newToken(PARAM_LIST_END); }
 }
@@ -611,12 +612,24 @@ Type = {PrimitiveType} | {ClassDescriptor} | {ArrayDescriptor}
     }
 }
 
+<ARRAY_DESCRIPTOR> {
+   {PrimitiveType} { yybegin(YYINITIAL); return newToken(PRIMITIVE_TYPE); }
+   {ClassDescriptor} { yybegin(YYINITIAL); return newToken(CLASS_DESCRIPTOR); }
+   [^] { yypushback(1); yybegin(YYINITIAL); }
+}
+
 /*Types*/
 <YYINITIAL> {
     {PrimitiveType} { return newToken(PRIMITIVE_TYPE); }
     V { return newToken(VOID_TYPE); }
     {ClassDescriptor} { return newToken(CLASS_DESCRIPTOR); }
-    {ArrayDescriptor} { return newToken(ARRAY_DESCRIPTOR); }
+
+    // we have to drop into a separate state so that we don't parse something like
+    // "[I->" as "[" followed by "I-" as a SIMPLE_NAME
+    {ArrayPrefix} {
+      yybegin(ARRAY_DESCRIPTOR);
+      return newToken(ARRAY_TYPE_PREFIX);
+    }
 
     {PrimitiveType} {PrimitiveType}+ {
         yypushback(yylength());

smali/src/test/resources/LexerTest/RealSmaliFileTest.tokens

@@ -184,7 +184,8 @@ REGISTER("v0")
 COMMA(",")
 REGISTER("v0")
 COMMA(",")
-ARRAY_DESCRIPTOR("[I")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("I")
 INSTRUCTION_FORMAT31t("fill-array-data")
 REGISTER("v0")
 COMMA(",")
@@ -197,7 +198,8 @@ CLASS_DESCRIPTOR("Lcom/android/internal/os/BatteryStatsImpl;")
 ARROW("->")
 SIMPLE_NAME("PROC_WAKELOCKS_FORMAT")
 COLON(":")
-ARRAY_DESCRIPTOR("[I")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("I")
 LINE_DIRECTIVE(".line")
 POSITIVE_INTEGER_LITERAL("3495")
 INSTRUCTION_FORMAT21c_TYPE("new-instance")

smali/src/test/resources/LexerTest/TypeAndIdentifierTest.smali

@@ -23,6 +23,8 @@ LI/I/I;
 [D
 [Ljava/lang/String;
 [LI/I/I;
+[[LI/I/I;
+[[I
 
 IIIII
 ZBSCIJFD
@@ -49,4 +51,6 @@ L[Ljava/lang/String;
 
 <linit>
 
-III
+III
+
+[I->clone()Ljava/lang/Object;

smali/src/test/resources/LexerTest/TypeAndIdentifierTest.tokens

@@ -13,16 +13,30 @@ CLASS_DESCRIPTOR("LI;")
 CLASS_DESCRIPTOR("LV;")
 CLASS_DESCRIPTOR("LI/I/I;")
 
-ARRAY_DESCRIPTOR("[Z")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[B")
+PRIMITIVE_TYPE("Z")
-ARRAY_DESCRIPTOR("[S")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[C")
+PRIMITIVE_TYPE("B")
-ARRAY_DESCRIPTOR("[I")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[J")
+PRIMITIVE_TYPE("S")
-ARRAY_DESCRIPTOR("[F")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[D")
+PRIMITIVE_TYPE("C")
-ARRAY_DESCRIPTOR("[Ljava/lang/String;")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[LI/I/I;")
+PRIMITIVE_TYPE("I")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("J")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("F")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("D")
+ARRAY_TYPE_PREFIX("[")
+CLASS_DESCRIPTOR("Ljava/lang/String;")
+ARRAY_TYPE_PREFIX("[")
+CLASS_DESCRIPTOR("LI/I/I;")
+ARRAY_TYPE_PREFIX("[[")
+CLASS_DESCRIPTOR("LI/I/I;")
+ARRAY_TYPE_PREFIX("[[")
+PRIMITIVE_TYPE("I")
 
 PARAM_LIST_OR_ID_START("")
 PRIMITIVE_TYPE("I")
@@ -46,8 +60,10 @@ PARAM_LIST_OR_ID_END("")
 PARAM_LIST_START("")
 PRIMITIVE_TYPE("I")
 CLASS_DESCRIPTOR("La;")
-ARRAY_DESCRIPTOR("[La;")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[I")
+CLASS_DESCRIPTOR("La;")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("I")
 PARAM_LIST_END("")
 
 PARAM_LIST_START("")
@@ -56,19 +72,26 @@ CLASS_DESCRIPTOR("Ljava/lang/String;")
 PARAM_LIST_END("")
 
 PARAM_LIST_START("")
-ARRAY_DESCRIPTOR("[I")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[I")
+PRIMITIVE_TYPE("I")
-ARRAY_DESCRIPTOR("[I")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("I")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("I")
 PARAM_LIST_END("")
 
 PARAM_LIST_START("")
-ARRAY_DESCRIPTOR("[I")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[Z")
+PRIMITIVE_TYPE("I")
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("Z")
 PARAM_LIST_END("")
 
 PARAM_LIST_START("")
-ARRAY_DESCRIPTOR("[I")
+ARRAY_TYPE_PREFIX("[")
-ARRAY_DESCRIPTOR("[Ljava/lang/String;")
+PRIMITIVE_TYPE("I")
+ARRAY_TYPE_PREFIX("[")
+CLASS_DESCRIPTOR("Ljava/lang/String;")
 PARAM_LIST_END("")
 
 MEMBER_NAME("<init>")
@@ -79,12 +102,12 @@ MEMBER_NAME("<init->")
 SIMPLE_NAME("Ljava") INVALID_TOKEN("/") SIMPLE_NAME("lang") INVALID_TOKEN("/") SIMPLE_NAME("String")
 SIMPLE_NAME("L") INVALID_TOKEN(";")
 SIMPLE_NAME("LI")
-SIMPLE_NAME("L") ARRAY_DESCRIPTOR("[Ljava/lang/String;")
+SIMPLE_NAME("L") ARRAY_TYPE_PREFIX("[") CLASS_DESCRIPTOR("Ljava/lang/String;")
 
-INVALID_TOKEN("[")
+ARRAY_TYPE_PREFIX("[")
-INVALID_TOKEN("[") VOID_TYPE("V")
+ARRAY_TYPE_PREFIX("[") VOID_TYPE("V")
-INVALID_TOKEN("[") SIMPLE_NAME("java") INVALID_TOKEN("/") SIMPLE_NAME("lang") INVALID_TOKEN("/") SIMPLE_NAME("String") INVALID_TOKEN(";")
+ARRAY_TYPE_PREFIX("[") SIMPLE_NAME("java") INVALID_TOKEN("/") SIMPLE_NAME("lang") INVALID_TOKEN("/") SIMPLE_NAME("String") INVALID_TOKEN(";")
-INVALID_TOKEN("[") INVALID_TOKEN(";")
+ARRAY_TYPE_PREFIX("[") INVALID_TOKEN(";")
 
 MEMBER_NAME("<linit>")
 
@@ -92,4 +115,12 @@ PARAM_LIST_OR_ID_START("")
 PRIMITIVE_TYPE("I")
 PRIMITIVE_TYPE("I")
 PRIMITIVE_TYPE("I")
-PARAM_LIST_OR_ID_END("")
+PARAM_LIST_OR_ID_END("")
+
+ARRAY_TYPE_PREFIX("[")
+PRIMITIVE_TYPE("I")
+ARROW("->")
+SIMPLE_NAME("clone")
+OPEN_PAREN("(")
+CLOSE_PAREN(")")
+CLASS_DESCRIPTOR("Ljava/lang/Object;")

smalidea/src/main/antlr3/smalideaParser.g

@@ -347,7 +347,7 @@ class_descriptor
 
 array_descriptor
   @init { Marker marker = mark(); }
-  : ARRAY_DESCRIPTOR;
+  : ARRAY_TYPE_PREFIX (primitive_type | class_descriptor);
   finally { marker.done(SmaliElementTypes.ARRAY_TYPE); }
 
 void_type

smalidea/src/main/java/org/jf/smalidea/SmaliTokens.java

@@ -50,7 +50,7 @@ public class SmaliTokens {
     @SuppressWarnings({"UnusedDeclaration"}) public static IElementType ANNOTATION_DIRECTIVE;
     @SuppressWarnings({"UnusedDeclaration"}) public static IElementType ANNOTATION_VISIBILITY;
     @SuppressWarnings({"UnusedDeclaration"}) public static IElementType ARRAY_DATA_DIRECTIVE;
-    @SuppressWarnings({"UnusedDeclaration"}) public static IElementType ARRAY_DESCRIPTOR;
+    @SuppressWarnings({"UnusedDeclaration"}) public static IElementType ARRAY_TYPE_PREFIX;
     @SuppressWarnings({"UnusedDeclaration"}) public static IElementType ARROW;
     @SuppressWarnings({"UnusedDeclaration"}) public static IElementType BOOL_LITERAL;
     @SuppressWarnings({"UnusedDeclaration"}) public static IElementType BYTE_LITERAL;
@@ -169,7 +169,7 @@ public class SmaliTokens {
         tokenColors.put("ANNOTATION_DIRECTIVE", SmaliHighlightingColors.DIRECTIVE);
         tokenColors.put("ANNOTATION_VISIBILITY", SmaliHighlightingColors.ACCESS);
         tokenColors.put("ARRAY_DATA_DIRECTIVE", SmaliHighlightingColors.DIRECTIVE);
-        tokenColors.put("ARRAY_DESCRIPTOR", SmaliHighlightingColors.TYPE);
+        tokenColors.put("ARRAY_TYPE_PREFIX", SmaliHighlightingColors.TYPE);
         tokenColors.put("ARROW", SmaliHighlightingColors.ARROW);
         tokenColors.put("BOOL_LITERAL", SmaliHighlightingColors.LITERAL);
         tokenColors.put("BYTE_LITERAL", SmaliHighlightingColors.NUMBER);

smalidea/src/test/java/org/jf/smalidea/SmaliLexerTest.java

@@ -72,7 +72,8 @@ public class SmaliLexerTest extends LexerTestCase {
                 "WHITE_SPACE (' ')\n" +
                 "SIMPLE_NAME ('main')\n" +
                 "OPEN_PAREN ('(')\n" +
-                "ARRAY_DESCRIPTOR ('[Ljava/lang/String;')\n" +
+                "ARRAY_TYPE_PREFIX ('[')\n" +
+                "CLASS_DESCRIPTOR ('Ljava/lang/String;')\n" +
                 "CLOSE_PAREN (')')\n" +
                 "VOID_TYPE ('V')\n" +
                 "WHITE_SPACE ('\\n    ')\n" +