Add profile to specify format specifications for different programmin…

…g languages/compiler

libr/anal/anal.c

@@ -79,6 +79,7 @@ R_API RAnal *r_anal_new() {
 	anal->sdb_meta = sdb_ns (anal->sdb, "meta", 1);
 	anal->sdb_hints = sdb_ns (anal->sdb, "hints", 1);
 	anal->sdb_types = sdb_ns (anal->sdb, "types", 1);
+	anal->sdb_fmts = sdb_ns (anal->sdb, "spec", 1);
 	anal->sdb_cc = sdb_ns (anal->sdb, "cc", 1);
 	anal->sdb_zigns = sdb_ns (anal->sdb, "zigns", 1);
 	anal->zign_path = strdup ("");

libr/anal/d/Makefile

@@ -12,6 +12,7 @@ F+= types-x86-macos-64
 F+= types-arm-ios-16
 F+= types-arm-ios-32
 F+= types-arm-ios-64
+F+= spec
 F+= cc-x86-64
 F+= cc-x86-32
 F+= cc-mips-32

libr/anal/d/meson.build

@@ -11,6 +11,7 @@ sdb_files = [
   'types-arm-ios-16',
   'types-arm-ios-32',
   'types-arm-ios-64',
+  'spec',
   'cc-x86-64',
   'cc-x86-32',
   'cc-mips-32',

libr/anal/d/spec.sdb.txt

@@ -0,0 +1,15 @@
+gcc=spec
+spec.gcc.c=char
+spec.gcc.d=int
+spec.gcc.f=float
+spec.gcc.g=double
+spec.gcc.ld=long int
+spec.gcc.li=long int
+spec.gcc.lf=double
+spec.gcc.lu=unsigned long int
+spec.gcc.llu=unsigned long long int
+spec.gcc.lli=long long int
+spec.gcc.lld=long long int
+spec.gcc.p=void *
+spec.gcc.s=const char *
+spec.gcc.u=unsigned int

libr/core/anal_tp.c

@@ -150,55 +150,26 @@ static ut64 get_addr(Sdb *trace, const char *regname, int idx) {
 	return r_num_math (NULL, sdb_const_get (trace, query, 0));
 }
 
-static RList *parse_format(char *fmt) {
+static RList *parse_format(RCore *core, char *fmt) {
 	RList *ret = r_list_new();
+	Sdb *s = core->anal->sdb_fmts;
+	const char *spec = r_config_get (core->config, "anal.spec");
+	char arr[10] = {0};
 	char *ptr = strchr (fmt, '%');
-	char *type = NULL;
+	fmt[strlen(fmt) - 1] = '\0';
 	while (ptr) {
 		ptr += 1;
-		switch(ptr[0]) {
-		case 'c':
-			type = "char";
-			break;
-		case 'f':
-			type = "float";
-			break;
-		case 'g':
-			type = "double";
-			break;
-		case 'l':
-			switch (ptr [1]) {
-			case 'f': // "%lf"
-				type = "double";
-				break;
-			case 'u': // "%lu"
-				type = "unsigned long int";
-				break;
-			case 'l':
-				if (ptr [2] == 'u') { // "%llu"
-					type = "unsigned long long int";
-				} else {
-					type = "long long int";
-				}
-				break;
-			default:
-				type = "long int";
-			}
-			break;
-		case 'p':
-			type = "void *";
-			break;
-		case 's':
-			type = "const char *";
-			break;
-		case 'u':
-			type = "unsigned int";
-			break;
-		case 'd':
-		default:
-			type = "int";
+		// strip [width] specifier
+		while (IS_DIGIT (*ptr)) { ptr++; }
+		strncpy (arr, ptr, sizeof(arr));
+		char *tmp = arr;
+		while (tmp && (IS_LOWER (*tmp) || IS_UPPER (*tmp))) { tmp++; }
+		*tmp = '\0';
+		const char *query = sdb_fmt ("spec.%s.%s", spec, arr);
+		char *type = (char *) sdb_const_get (s, query, 0);
+		if (type) {
+			r_list_append (ret, type);
 		}
-		r_list_append (ret, type);
 		ptr = strchr (ptr, '%');
 	}
 	return ret;
@@ -299,7 +270,7 @@ static void type_match(RCore *core, ut64 addr, char *fcn_name, ut64 baddr, const
 					if ((op->ptr && op->ptr != UT64_MAX) && !strcmp (name, "format")) {
 						RFlagItem *f = r_flag_get_i (core->flags, op->ptr);
 						if (f && !strncmp (f->name, "str", 3)) {
-							types = parse_format (f->realname);
+							types = parse_format (core, f->realname);
 							max += r_list_length (types);
 							format = true;
 						}

libr/core/cbin.c

@@ -823,6 +823,11 @@ static int bin_info(RCore *r, int mode) {
 	}
 	r_core_anal_type_init (r);
 	r_core_anal_cc_init (r);
+	const char *dir_prefix = r_config_get (r->config, "dir.prefix");
+	char *spath = sdb_fmt ("%s/"R2_SDB_FCNSIGN"/spec.sdb", dir_prefix);
+	if (r_file_exists (spath)) {
+		sdb_concat_by_path (r->anal->sdb_fmts, spath);
+	}
 	return true;
 }
 

libr/core/cconfig.c

@@ -2378,6 +2378,7 @@ R_API int r_core_config_init(RCore *core) {
 	SETPREF ("anal.hasnext", "false", "Continue analysis after each function");
 	SETPREF ("anal.esil", "false", "Use the new ESIL code analysis");
 	SETCB ("anal.strings", "false", &cb_analstrings, "Identify and register strings during analysis (aar only)");
+	SETPREF ("anal.spec", "gcc",  "Set profile for specifying format chars used in type analysis");
 	SETCB ("anal.vars", "true", &cb_analvars, "Analyze local variables and arguments");
 	SETPREF ("anal.vinfun", "true",  "Search values in functions (aav) (false by default to only find on non-code)");
 	SETPREF ("anal.vinfunrange", "false",  "Search values outside function ranges (requires anal.vinfun=false)\n");

libr/include/r_anal.h

@@ -653,6 +653,7 @@ typedef struct r_anal_t {
 	RAnalRange *limit;
 	RList *plugins;
 	Sdb *sdb_types;
+	Sdb *sdb_fmts;
 	Sdb *sdb_meta; // TODO: Future r_meta api
 	Sdb *sdb_zigns;