Add stringtoint() and inttostring(). (#1767)

* Add stringtoint() and inttostring().

Add math.stringtoint() and math.inttostring() which let you convert between an
integer and a string representation of it, and back.

* Rename functions and use a stack allocation and snprintf()

* Use errno.h, not sys/errno.h

docs/modules/math.rst

@@ -157,11 +157,54 @@ file and create signatures based on those results.
 .. c:function:: mode(offset, size)
 
     .. versionadded:: 4.2.0
-    
+
     Returns the most common byte, starting at *offset* and looking at the next
     *size* bytes. When scanning a
     running process the *offset* argument should be a virtual address within
     the process address space. The returned value is a float.
     *offset* and *size* are optional; if left empty, the complete file is searched.
 
     *Example: math.mode(0, filesize) == 0xFF*
+
+.. c:function:: to_string(int)
+
+    .. versionadded:: 4.3.0
+
+    Convert the given integer to a string. Note: integers in YARA are signed.
+
+    *Example: math.to_string(10) == "10"*
+    *Example: math.to_string(-1) == "-1"*
+
+.. c:function:: to_string(int, base)
+
+    .. versionadded:: 4.3.0
+
+    Convert the given integer to a string in the given base. Supported bases are
+    10, 8 and 16. Note: integers in YARA are signed.
+
+    *Example: math.to_string(32, 16) == "20"*
+    *Example: math.to_string(-1, 16) == "ffffffffffffffff"*
+
+.. c:function:: to_int(string)
+
+    .. versionadded:: 4.3.0
+
+    Convert the given string to a signed integer. If the string starts with "0x"
+    it is treated as base 16. If the string starts with "0" it is treated base
+    8. Leading '+' or '-' is also supported.
+
+    *Example: math.to_int("1234") == 1234*
+    *Example: math.to_int("-10") == -10*
+    *Example: math.to_int("-010" == -8*
+
+.. c:function:: to_int(string, base)
+
+    .. versionadded:: 4.3.0
+
+    Convert the given string, interpreted with the given base, to a signed
+    integer. Base must be 0 or between 2 and 32 inclusive. If it is zero then
+    the string will be intrepreted as base 16 if it starts with "0x" or as base
+    8 if it starts with "0". Leading '+' or '-' is also supported.
+
+    *Example: math.to_int("011", 8) == "9"*
+    *Example: math.to_int("-011", 0) == "-9"*

libyara/modules/math/math.c

@@ -27,6 +27,8 @@ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
 
+#include <stdlib.h>
+#include <errno.h>
 #include <math.h>
 #include <yara/mem.h>
 #include <yara/modules.h>
@@ -35,6 +37,9 @@ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #define MODULE_NAME math
 
 #define PI 3.141592653589793
+// This is more than enough space to hold the maximum signed 64bit integer as a
+// string in decimal, hex or octal, including the sign and NULL terminator.
+#define INT64_MAX_STRING 30
 
 // log2 is not defined by math.h in VC++
 
@@ -721,6 +726,53 @@ define_function(mode_global)
   return_integer(most_common);
 }
 
+define_function(to_string)
+{
+  int64_t i = integer_argument(1);
+  char str[INT64_MAX_STRING];
+  snprintf(str, INT64_MAX_STRING, "%lld", i);
+  return_string(&str);
+}
+
+define_function(to_string_base)
+{
+  int64_t i = integer_argument(1);
+  int64_t base = integer_argument(2);
+  char str[INT64_MAX_STRING];
+  char *fmt;
+  switch (base)
+  {
+  case 10:
+    fmt = "%lld";
+    break;
+  case 8:
+    fmt = "%llo";
+    break;
+  case 16:
+    fmt = "%llx";
+    break;
+  default:
+    return_string(YR_UNDEFINED);
+  }
+  snprintf(str, INT64_MAX_STRING, fmt, i);
+  return_string(&str);
+}
+
+define_function(to_int)
+{
+  char* s = string_argument(1);
+  int64_t result = strtoll(s, NULL, 0);
+  return_integer(result == 0 && errno ? YR_UNDEFINED : result);
+}
+
+define_function(to_int_base)
+{
+  char* s = string_argument(1);
+  int64_t base = integer_argument(2);
+  int64_t result = strtoll(s, NULL, base);
+  return_integer(result == 0 && errno ? YR_UNDEFINED : result);
+}
+
 begin_declarations
   declare_float("MEAN_BYTES");
   declare_function("in_range", "fff", "i", in_range);
@@ -744,6 +796,10 @@ begin_declarations
   declare_function("percentage", "i", "f", percentage_global);
   declare_function("mode", "ii", "i", mode_range);
   declare_function("mode", "", "i", mode_global);
+  declare_function("to_string", "i", "s", to_string);
+  declare_function("to_string", "ii", "s", to_string_base);
+  declare_function("to_int", "s", "i", to_int);
+  declare_function("to_int", "si", "i", to_int_base);
 end_declarations
 
 int module_initialize(YR_MODULE* module)

tests/test-math.c

@@ -223,6 +223,140 @@ int main(int argc, char** argv)
       }",
       "123ABCDEF123456987DE");
 
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_string(1234) == \"1234\" \
+      }",
+      NULL);
+
+  // We use signed integers by default if no base is specified.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_string(-1) == \"-1\" \
+      }",
+      NULL);
+
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_string(32, 16) == \"20\" \
+      }",
+      NULL);
+
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_string(32, 8) == \"40\" \
+      }",
+      NULL);
+
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_string(32, 10) == \"32\" \
+      }",
+      NULL);
+
+  // Base 10 is always a signed integer, all other bases are unsigned.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_string(-1, 10) == \"-1\" and \
+          math.to_string(-1, 16) == \"ffffffffffffffff\" and \
+          math.to_string(-1, 8) == \"1777777777777777777777\" \
+      }",
+      NULL);
+
+  // Passing a base that is not 10, 8 or 16 will result in UNDEFINED.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          not defined(math.to_string(32, 9)) \
+      }",
+      NULL);
+
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\"1234\") == 1234 \
+      }",
+      NULL);
+
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\"-1\") == -1 \
+      }",
+      NULL);
+
+  // Leading spaces and + are allowed.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\" +1\") == 1 \
+      }",
+      NULL);
+
+  // Strings can be prefixed with 0x and will be interpreted as hexadecimal.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\"0x10\") == 16 \
+      }",
+      NULL);
+
+  // Strings prefixed with 0 will be interpreted as octal.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\"010\") == 8 \
+      }",
+      NULL);
+
+  // Strings that are only partially converted are still fine.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\"10A20\") == 10 \
+      }",
+      NULL);
+
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\"10\", 8) == 8 \
+      }",
+      NULL);
+
+  // Base 0 is a special case that tries to interpret the string by prefix, or
+  // default to decimal. We aren't doing anything special to get this, it is
+  // part of strtoll by default.
+  assert_true_rule(
+      "import \"math\" \
+      rule test { \
+        condition: \
+          math.to_int(\"010\", 0) == 8 and \
+          math.to_int(\"0x10\", 0) == 16 and \
+          math.to_int(\"10\", 0) == 10 \
+      }",
+      NULL);
+
   yr_finalize();
 
   YR_DEBUG_FPRINTF(