[apparmor] [patch 05/16] cleanup/fix escape sequences in the backend and add support for \d

[john.johansen at canonical.com]

the octal escape sequence was broken, so that short escapes \0, \00 \xa,
didn't work and actually resulted in some encoding bugs.

Also we were missing support for the decimal # conversion \d123

Signed-off-by: John Johansen <john.johansen at canonical.com>

---
 parser/libapparmor_re/parse.y |  122 +++++++++++++++++++++++++++++-------------
 1 file changed, 85 insertions(+), 37 deletions(-)

--- 2.9-test.orig/parser/libapparmor_re/parse.y
+++ 2.9-test/parser/libapparmor_re/parse.y
@@ -156,29 +156,100 @@
 %%
 
 
+int isodigit(char c)
+{
+	return (c >= '0' && c <= '7') ? 1 : 0;
+}
+
 int octdigit(char c)
 {
-	if (c >= '0' && c <= '7')
+	if (isodigit(c))
+		return c - '0';
+	return -1;
+}
+
+int decdigit(char c)
+{
+	if (isdigit(c))
 		return c - '0';
 	return -1;
 }
 
 int hexdigit(char c)
 {
-	if (c >= '0' && c <= '9')
+	if (isdigit(c))
 		return c - '0';
 	else if (c >= 'A' && c <= 'F')
 		return 10 + c - 'A';
 	else if (c >= 'a' && c <= 'f')
 		return 10 + c - 'a';
-	else
-		return -1;
+	return -1;
 }
 
-int regex_lex(YYSTYPE *val, const char **pos)
+/* take the largest sequence of character that is in range of 0-255
+ * - will consume at most 3 digits.
+ * - if a 3 digit sequence has value greater than 255 it will take
+ *   the first 2 digits.
+ *   eg. for the sequence \444, \44 will be taken as the octal number
+ *       and the third 4 will not be consumed (ie pos will point to it)
+ */
+int convert_octseq(const char **pos)
 {
-	int c;
+	int val, c;
+
+	for (val = 0; (c = octdigit(**pos)) >= 0; (*pos)++) {
+		int tmp = (val << 3) + c;
+		if (tmp > 255)
+			break;
+		val = tmp;
+	}
+
+	return val;
+}
+
+/* take the largest sequence of character that is in range of 0-255
+ * - will consume at most 3 digits.
+ * - if a 3 digit sequence has value greater than 255 it will take
+ *   the first 2 digits.
+ *   eg. for the sequence \d444, \d44 will be taken as the decimal number
+ *       and the third 4 will not be consumed (ie pos will point to it)
+ */
+int convert_decseq(const char **pos)
+{
+	int val, c;
+
+	for (val = 0; (c = decdigit(**pos)) >= 0; (*pos)++) {
+		int tmp = (val * 10) + c;
+		if (tmp > 255)
+			break;
+		val = tmp;
+	}
+
+	return val;
+}
 
+/* take the largest sequence of character that is in range of 0-255
+ * - will consume at most 2 digits.
+ * - if there is not two valid hex characters only the first will be taken
+ *   eg. for the sequence \x4z, \x4 will be taken as the hex number
+ */
+int convert_hexseq(const char **pos)
+{
+	int val, c;
+
+	for (val = 0; (c = hexdigit(**pos)) >= 0; (*pos)++) {
+		int tmp = (val << 4) + c;
+		if (tmp > 255)
+			break;
+		val = tmp;
+	}
+
+	return val;
+}
+
+
+int regex_lex(YYSTYPE *val, const char **pos)
+{
 	val->c = **pos;
 	switch(*(*pos)++) {
 	case '\0':
@@ -191,6 +262,11 @@
 
 	case '\\':
 		val->c = **pos;
+		if (isodigit(**pos)) {
+			val->c = convert_octseq(pos);
+			break;
+		}
+
 		switch(*(*pos)++) {
 		case '\0':
 			(*pos)--;
@@ -198,55 +274,27 @@
 		case '\\':
 			val->c = '\\';
 			break;
-
-		case '0':
-			val->c = 0;
-			if ((c = octdigit(**pos)) >= 0) {
-				val->c = c;
-				(*pos)++;
-			}
-			if ((c = octdigit(**pos)) >= 0) {
-				val->c = (val->c << 3) + c;
-				(*pos)++;
-			}
-			if ((c = octdigit(**pos)) >= 0) {
-				val->c = (val->c << 3) + c;
-				(*pos)++;
-			}
+		case 'd':
+			val->c = convert_decseq(pos);
 			break;
-
 		case 'x':
-			val->c = 0;
-			if ((c = hexdigit(**pos)) >= 0) {
-				val->c = c;
-				(*pos)++;
-			}
-			if ((c = hexdigit(**pos)) >= 0) {
-				val->c = (val->c << 4) + c;
-				(*pos)++;
-			}
+			val->c = convert_hexseq(pos);
 			break;
-
 		case 'a':
 			val->c = '\a';
 			break;
-
 		case 'e':
 			val->c = 033  /* ESC */;
 			break;
-
 		case 'f':
 			val->c = '\f';
 			break;
-
 		case 'n':
 			val->c = '\n';
 			break;
-
 		case 'r':
 			val->c = '\r';
 			break;
-
 		case 't':
 			val->c = '\t';
 			break;