From b03c6164a49aec8632b5b530eb12dc0b931bef60 Mon Sep 17 00:00:00 2001
From: Thomas Punt <tpunt@hotmail.co.uk>
Date: Tue, 29 Dec 2015 22:19:33 +0000
Subject: [PATCH] Improve implementation

Use estrndup() over strndup()
Free allocated memory
Only duplicate yytext when neccesary
---
 Zend/zend_language_scanner.l | 81 ++++++++++++++++++++++++++++++------
 1 file changed, 69 insertions(+), 12 deletions(-)

diff --git a/Zend/zend_language_scanner.l b/Zend/zend_language_scanner.l
index 1b9a5b3b13b6b..1c023649450be 100644
--- a/Zend/zend_language_scanner.l
+++ b/Zend/zend_language_scanner.l
@@ -112,6 +112,13 @@ do {																			\
 	} \
 }
 
+#define CONTAINS_UNDERSCORES(n, len, res) \
+{ \
+	int i; \
+	for (i = 0; i < len && n[i] != '_'; ++i); \
+	res = i != len; \
+}
+
 #define STRIP_UNDERSCORES(n, len) \
 { \
 	int i, old_len = len; \
@@ -1617,10 +1624,15 @@ NEWLINE ("\r"|"\n"|"\r\n")
 
 <ST_IN_SCRIPTING>{BNUM} {
 	/* The +/- 2 skips "0b" */
-	int len = yyleng - 2;
-	char *end, *bin = strndup(yytext + 2, len);
+	int len = yyleng - 2, contains_underscores;
+	char *end, *bin = yytext + 2;
 
-	STRIP_UNDERSCORES(bin, len)
+	CONTAINS_UNDERSCORES(bin, len, contains_underscores)
+
+	if (contains_underscores) {
+		bin = estrndup(bin, len);
+		STRIP_UNDERSCORES(bin, len)
+	}
 
 	/* Skip any leading 0s */
 	while (*bin == '0') {
@@ -1636,20 +1648,31 @@ NEWLINE ("\r"|"\n"|"\r\n")
 			ZVAL_LONG(zendlval, ZEND_STRTOL(bin, &end, 2));
 			ZEND_ASSERT(!errno && end == yytext + yyleng);
 		}
+		if (contains_underscores) {
+			efree(bin);
+		}
 		RETURN_TOKEN(T_LNUMBER);
 	} else {
 		ZVAL_DOUBLE(zendlval, zend_bin_strtod(bin, (const char **)&end));
 		/* errno isn't checked since we allow HUGE_VAL/INF overflow */
 		ZEND_ASSERT(end == yytext + yyleng);
+		if (contains_underscores) {
+			efree(bin);
+		}
 		RETURN_TOKEN(T_DNUMBER);
 	}
 }
 
 <ST_IN_SCRIPTING>{LNUM} {
-	int len = yyleng;
-	char *end, *lnum = strndup(yytext, yyleng);
+	int len = yyleng, contains_underscores;
+	char *end, *lnum = yytext;
+
+	CONTAINS_UNDERSCORES(lnum, len, contains_underscores)
 
-	STRIP_UNDERSCORES(lnum, len)
+	if (contains_underscores) {
+		lnum = estrndup(lnum, len);
+		STRIP_UNDERSCORES(lnum, len)
+	}
 
 	if (len < MAX_LENGTH_OF_LONG - 1) { /* Won't overflow */
 		errno = 0;
@@ -1660,6 +1683,9 @@ NEWLINE ("\r"|"\n"|"\r\n")
 		if (end != lnum + len) {
 			zend_throw_exception(zend_ce_parse_error, "Invalid numeric literal", 0);
 			ZVAL_UNDEF(zendlval);
+			if (contains_underscores) {
+				efree(lnum);
+			}
 			RETURN_TOKEN(T_LNUMBER);
 		}
 	} else {
@@ -1678,28 +1704,45 @@ NEWLINE ("\r"|"\n"|"\r\n")
 				zend_throw_exception(zend_ce_parse_error,
 					"Invalid numeric literal", 0);
 				ZVAL_UNDEF(zendlval);
+				if (contains_underscores) {
+					efree(lnum);
+				}
 				RETURN_TOKEN(T_DNUMBER);
 			}
 			ZEND_ASSERT(!errno);
+			if (contains_underscores) {
+				efree(lnum);
+			}
 			RETURN_TOKEN(T_DNUMBER);
 		}
 		/* Also not an assert for the same reason */
 		if (end != lnum + len) {
 			zend_throw_exception(zend_ce_parse_error, "Invalid numeric literal", 0);
 			ZVAL_UNDEF(zendlval);
+			if (contains_underscores) {
+				efree(lnum);
+			}
 			RETURN_TOKEN(T_DNUMBER);
 		}
 	}
 	ZEND_ASSERT(!errno);
+	if (contains_underscores) {
+		efree(lnum);
+	}
 	RETURN_TOKEN(T_LNUMBER);
 }
 
 <ST_IN_SCRIPTING>{HNUM} {
 	/* The +/- 2 skips "0x" */
-	int len = yyleng - 2;
-	char *end, *hex = strndup(yytext + 2, len);
+	int len = yyleng - 2, contains_underscores;
+	char *end, *hex = yytext + 2;
 
-	STRIP_UNDERSCORES(hex, len)
+	CONTAINS_UNDERSCORES(hex, len, contains_underscores)
+
+	if (contains_underscores) {
+		hex = estrndup(hex, len);
+		STRIP_UNDERSCORES(hex, len)
+	}
 
 	/* Skip any leading 0s */
 	while (*hex == '0') {
@@ -1715,11 +1758,17 @@ NEWLINE ("\r"|"\n"|"\r\n")
 			ZVAL_LONG(zendlval, ZEND_STRTOL(hex, &end, 16));
 			ZEND_ASSERT(!errno && end == hex + len);
 		}
+		if (contains_underscores) {
+			efree(hex);
+		}
 		RETURN_TOKEN(T_LNUMBER);
 	} else {
 		ZVAL_DOUBLE(zendlval, zend_hex_strtod(hex, (const char **)&end));
 		/* errno isn't checked since we allow HUGE_VAL/INF overflow */
 		ZEND_ASSERT(end == hex + len);
+		if (contains_underscores) {
+			efree(hex);
+		}
 		RETURN_TOKEN(T_DNUMBER);
 	}
 }
@@ -1747,14 +1796,22 @@ string:
 
 <ST_IN_SCRIPTING>{DNUM}|{EXPONENT_DNUM} {
 	const char *end;
-	int len = yyleng;
-	char *dnum = strndup(yytext, yyleng);
+	int len = yyleng, contains_underscores;
+	char *dnum = yytext;
 
-	STRIP_UNDERSCORES(dnum, len)
+	CONTAINS_UNDERSCORES(dnum, len, contains_underscores)
+
+	if (contains_underscores) {
+		dnum = estrndup(dnum, len);
+		STRIP_UNDERSCORES(dnum, len)
+	}
 
 	ZVAL_DOUBLE(zendlval, zend_strtod(dnum, &end));
 	/* errno isn't checked since we allow HUGE_VAL/INF overflow */
 	ZEND_ASSERT(end == dnum + len);
+	if (contains_underscores) {
+		efree(dnum);
+	}
 	RETURN_TOKEN(T_DNUMBER);
 }