Initial implementation of RAnalDataType ##anal (#13111)

libr/anal/data.c

@@ -412,3 +412,29 @@ R_API const char *r_anal_data_kind(RAnal *a, ut64 addr, const ut8 *buf, int len)
 	}
 	return "data";
 }
+
+R_API const char *r_anal_datatype_to_string(RAnalDataType t) {
+	switch (t) {
+	case R_ANAL_DATATYPE_NULL:
+		return NULL;
+	case R_ANAL_DATATYPE_ARRAY:
+		return "array";
+	case R_ANAL_DATATYPE_OBJECT: // instance
+		return "object";
+	case R_ANAL_DATATYPE_STRING:
+		return "string";
+	case R_ANAL_DATATYPE_CLASS:
+		return "class";
+	case R_ANAL_DATATYPE_BOOLEAN:
+		return "boolean";
+	case R_ANAL_DATATYPE_INT16:
+		return "int16";
+	case R_ANAL_DATATYPE_INT32:
+		return "int32";
+	case R_ANAL_DATATYPE_INT64:
+		return "int64";
+	case R_ANAL_DATATYPE_FLOAT:
+		return "float";
+	}
+	return NULL;
+}

libr/anal/p/anal_dalvik.c

@@ -139,16 +139,19 @@ static int dalvik_op(RAnal *anal, RAnalOp *op, ut64 addr, const ut8 *data, int l
 		break;
 	case 0x1a: // const-string
 		op->type = R_ANAL_OP_TYPE_MOV;
+		op->datatype = R_ANAL_DATATYPE_STRING;
 		if (len > 2) {
 			ut32 vA = data[1];
 			ut32 vB = (data[3]<<8) | data[2];
 			ut64 offset = R_ANAL_GET_OFFSET (anal, 's', vB);
 			op->ptr = offset;
+			op->refptr = 0;
 			esilprintf (op, "0x%"PFMT64x",v%d,=", offset, vA);
 		}
 		break;
 	case 0x1c: // const-class
 		op->type = R_ANAL_OP_TYPE_MOV;
+		op->datatype = R_ANAL_DATATYPE_CLASS;
 		break;
 	case 0x89: // float-to-double
 	case 0x8a: // double-to-int
@@ -174,6 +177,7 @@ static int dalvik_op(RAnal *anal, RAnalOp *op, ut64 addr, const ut8 *data, int l
 		break;
 	case 0x8f: // int-to-short
 		op->type = R_ANAL_OP_TYPE_CAST;
+		// op->datatype = R_ANAL_DATATYPE_INT32 | R_ANAL_DATATYPE_INT16;
 		{
 		ut32 vA = (data[1] & 0x0f);
 		ut32 vB = (data[1] & 0xf0) >> 4;
@@ -191,12 +195,12 @@ static int dalvik_op(RAnal *anal, RAnalOp *op, ut64 addr, const ut8 *data, int l
 	case 0x20: // instance-of
 		{
 		op->type = R_ANAL_OP_TYPE_CMP;
-
 		esilprintf (op, "%d,instanceof,%d,-,!,v%d,=", vC, vB, vA);
 		}
 		break;
 	case 0x21: // array-length
 		op->type = R_ANAL_OP_TYPE_LENGTH;
+		op->datatype = R_ANAL_DATATYPE_ARRAY;
 		break;
 	case 0x44: // aget
 	case 0x45: //aget-bool
@@ -240,9 +244,20 @@ static int dalvik_op(RAnal *anal, RAnalOp *op, ut64 addr, const ut8 *data, int l
 		}
 		break;
 	case 0x63: // sget-boolean
+		{
+		const char *vT = "-boolean";
+		op->datatype = R_ANAL_DATATYPE_BOOLEAN;
+		op->type = R_ANAL_OP_TYPE_LOAD;
+		ut32 vA = (data[1] & 0x0f);
+		ut32 vB = (data[1] & 0xf0) >> 4;
+		ut32 vC = (data[2] & 0x0f);
+		esilprintf (op, "%d,%d,sget%s,v%d,=", vC, vB, vT, vA);
+		}
+		break;
 	case 0x62: // sget-object
 		{
 		const char *vT = "-object";
+		op->datatype = R_ANAL_DATATYPE_OBJECT;
 		op->type = R_ANAL_OP_TYPE_LOAD;
 		ut32 vA = (data[1] & 0x0f);
 		ut32 vB = (data[1] & 0xf0) >> 4;

libr/core/cmd_anal.c

@@ -1544,6 +1544,13 @@ static void core_anal_bytes(RCore *core, const ut8 *buf, int len, int nops, int
 			}
 			pj_ki (pj, "size", size);
 			pj_ks (pj, "type", r_anal_optype_to_string (op.type));
+			{
+				const char *datatype = r_anal_datatype_to_string (op.datatype);
+				if (datatype) {
+					pj_ks (pj, "datatype", datatype);
+				}
+
+			}
 			pj_ks (pj, "reg", op.reg);
 			pj_ks (pj, "ireg", op.ireg);
 			pj_ki (pj, "scale", op.scale);
@@ -1625,6 +1632,10 @@ static void core_anal_bytes(RCore *core, const ut8 *buf, int len, int nops, int
 			printline ("size", "%d\n", size);
 			printline ("sign", "%s\n", r_str_bool (op.sign));
 			printline ("type", "%s\n", r_anal_optype_to_string (op.type));
+			const char *datatype = r_anal_datatype_to_string (op.datatype);
+			if (datatype) {
+				printline ("datatype", "%s\n", datatype);
+			}
 			printline ("cycles", "%d\n", op.cycles);
 			if (op.failcycles) {
 				printline ("failcycles", "%d\n", op.failcycles);

libr/include/r_anal.h

@@ -791,6 +791,19 @@ enum RAnalOpDirection {
 	R_ANAL_OP_DIR_REF = 8,
 };
 
+typedef enum r_anal_data_type_t {
+	R_ANAL_DATATYPE_NULL = 0,
+	R_ANAL_DATATYPE_ARRAY,
+	R_ANAL_DATATYPE_OBJECT, // instance
+	R_ANAL_DATATYPE_STRING,
+	R_ANAL_DATATYPE_CLASS,
+	R_ANAL_DATATYPE_BOOLEAN,
+	R_ANAL_DATATYPE_INT16,
+	R_ANAL_DATATYPE_INT32,
+	R_ANAL_DATATYPE_INT64,
+	R_ANAL_DATATYPE_FLOAT,
+} RAnalDataType;
+
 typedef struct r_anal_op_t {
 	char *mnemonic; /* mnemonic */
 	ut64 addr;      /* address */
@@ -830,6 +843,7 @@ typedef struct r_anal_op_t {
 	ut64 disp;
 	RAnalSwitchOp *switch_op;
 	RAnalHint hint;
+	RAnalDataType datatype;
 } RAnalOp;
 
 #define R_ANAL_COND_SINGLE(x) (!x->arg[1] || x->arg[0]==x->arg[1])
@@ -1287,6 +1301,7 @@ R_API RAnalType *r_anal_type_new(void);
 R_API void r_anal_type_add(RAnal *l, RAnalType *t);
 R_API RAnalType *r_anal_type_find(RAnal *a, const char* name);
 R_API void r_anal_type_list(RAnal *a, short category, short enabled);
+R_API const char *r_anal_datatype_to_string(RAnalDataType t);
 R_API RAnalType *r_anal_str_to_type(RAnal *a, const char* s);
 R_API bool r_anal_op_nonlinear(int t);
 R_API bool r_anal_op_ismemref(int t);

libr/include/r_bin.h

@@ -277,7 +277,7 @@ typedef struct r_bin_file_t {
 	struct r_bin_t *rbin;
 } RBinFile;
 
-typedef struct RBinFileOptions {
+typedef struct r_bin_file_options_t {
 	int rawstr;
 	ut64 baddr; // base address
 	ut64 laddr; // load address