php · nicolas-grekas · May 16, 2017
diff --git a/UPGRADING b/UPGRADING
@@ -56,11 +56,6 @@ PHP 7.2 UPGRADE NOTES
     parameter (assoc) is null. Previously JSON_OBJECT_AS_ARRAY was always
     ignored.
 
-- PCRE:
-  . preg_match() and other PCRE functions now distinguish between unmatched
-    subpatterns and empty matches by reporting NULL and "" (empty string),
-    respectively. Formerly, either was reported as empty string.
-
 - Session:
   . Removed register_globals related code and "!" can be used as $_SESSION key name.
   . Session is made to manage session status correctly and prevents invalid operations.
@@ -106,6 +101,9 @@ PHP 7.2 UPGRADE NOTES
 
 - PCRE:
   . Added `J` modifier for setting PCRE_DUPNAMES.
+  . Added `PREG_UNMATCHED_AS_NULL` flag to allow distinguish between unmatched
+    subpatterns and empty matches by reporting NULL and "" (empty string),
+    respectively.
 
 - Standard:
   . Simplified password hashing API updated to support Argon2i hashes when PHP is compiled with libargon2
@@ -274,6 +272,9 @@ See also: https://wiki.php.net/rfc/deprecations_php_7_2
   . IMG_EFFECT_MULTIPLY
   . IMG_BMP
 
+- PCRE
+  . PREG_UNMATCHED_AS_NULL
+
 - Standard:
   . PASSWORD_ARGON2_DEFAULT_MEMORY_COST
   . PASSWORD_ARGON2_DEFAULT_TIME_COST

diff --git a/ext/pcre/php_pcre.c b/ext/pcre/php_pcre.c
@@ -33,6 +33,7 @@
 #define PREG_PATTERN_ORDER			1
 #define PREG_SET_ORDER				2
 #define PREG_OFFSET_CAPTURE			(1<<8)
+#define PREG_UNMATCHED_AS_NULL		(1<<9)
 
 #define	PREG_SPLIT_NO_EMPTY			(1<<0)
 #define PREG_SPLIT_DELIM_CAPTURE	(1<<1)
@@ -188,6 +189,7 @@ static PHP_MINIT_FUNCTION(pcre)
 	REGISTER_LONG_CONSTANT("PREG_PATTERN_ORDER", PREG_PATTERN_ORDER, CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("PREG_SET_ORDER", PREG_SET_ORDER, CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("PREG_OFFSET_CAPTURE", PREG_OFFSET_CAPTURE, CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("PREG_UNMATCHED_AS_NULL", PREG_UNMATCHED_AS_NULL, CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("PREG_SPLIT_NO_EMPTY", PREG_SPLIT_NO_EMPTY, CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("PREG_SPLIT_DELIM_CAPTURE", PREG_SPLIT_DELIM_CAPTURE, CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("PREG_SPLIT_OFFSET_CAPTURE", PREG_SPLIT_OFFSET_CAPTURE, CONST_CS | CONST_PERSISTENT);
@@ -639,14 +641,14 @@ PHPAPI pcre* pcre_get_compiled_regex_ex(zend_string *regex, pcre_extra **extra,
 /* }}} */
 
 /* {{{ add_offset_pair */
-static inline void add_offset_pair(zval *result, char *str, int len, int offset, char *name)
+static inline void add_offset_pair(zval *result, char *str, int len, int offset, char *name, int unmatched_as_null)
 {
 	zval match_pair, tmp;
 
 	array_init_size(&match_pair, 2);
 
 	/* Add (match, offset) to the return value */
-	if (offset < 0) { /* unset substring */
+	if (unmatched_as_null && offset < 0) {
 		ZVAL_NULL(&tmp);
 	} else {
 		ZVAL_STRINGL(&tmp, str, len);
@@ -705,7 +707,7 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 {
 	zval			 result_set,		/* Holds a set of subpatterns after
 										   a global match */
-				    *match_sets = NULL;	/* An array of sets of matches for each
+					*match_sets = NULL;	/* An array of sets of matches for each
 										   subpattern after a global match */
 	pcre_extra		*extra = pce->extra;/* Holds results of studying */
 	pcre_extra		 extra_data;		/* Used locally for exec options */
@@ -720,9 +722,10 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 	char 		   **subpat_names;		/* Array for named subpatterns */
 	int				 i;
 	int				 subpats_order;		/* Order of subpattern matches */
-	int				 offset_capture;    /* Capture match offsets: yes/no */
-	unsigned char   *mark = NULL;       /* Target for MARK name */
-	zval            marks;      		/* Array of marks for PREG_PATTERN_ORDER */
+	int				 offset_capture;	/* Capture match offsets: yes/no */
+	int				 unmatched_as_null;	/* Null non-matches: yes/no */
+	unsigned char   *mark = NULL;		/* Target for MARK name */
+	zval			 marks;				/* Array of marks for PREG_PATTERN_ORDER */
 	ALLOCA_FLAG(use_heap);
 
 	ZVAL_UNDEF(&marks);
@@ -737,6 +740,7 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 
 	if (use_flags) {
 		offset_capture = flags & PREG_OFFSET_CAPTURE;
+		unmatched_as_null = flags & PREG_UNMATCHED_AS_NULL;
 
 		/*
 		 * subpats_order is pre-set to pattern mode so we change it only if
@@ -752,6 +756,7 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 		}
 	} else {
 		offset_capture = 0;
+		unmatched_as_null = 0;
 	}
 
 	/* Negative offset counts from the end of the string. */
@@ -847,11 +852,11 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 						if (offset_capture) {
 							for (i = 0; i < count; i++) {
 								add_offset_pair(&match_sets[i], (char *)stringlist[i],
-												offsets[(i<<1)+1] - offsets[i<<1], offsets[i<<1], NULL);
+												offsets[(i<<1)+1] - offsets[i<<1], offsets[i<<1], NULL, unmatched_as_null);
 							}
 						} else {
 							for (i = 0; i < count; i++) {
-								if (offsets[i<<1] < 0) { /* unset substring */
+								if (unmatched_as_null && offsets[i<<1] < 0) {
 									add_next_index_null(&match_sets[i]);
 								} else {
 									add_next_index_stringl(&match_sets[i], (char *)stringlist[i],
@@ -869,11 +874,15 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 						/*
 						 * If the number of captured subpatterns on this run is
 						 * less than the total possible number, pad the result
-						 * arrays with NULLs.
+						 * arrays with NULLs or empty strings.
 						 */
 						if (count < num_subpats) {
 							for (; i < num_subpats; i++) {
-								add_next_index_null(&match_sets[i]);
+								if (unmatched_as_null) {
+									add_next_index_null(&match_sets[i]);
+								} else {
+									add_next_index_string(&match_sets[i], "");
+								}
 							}
 						}
 					} else {
@@ -885,19 +894,19 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 							if (offset_capture) {
 								for (i = 0; i < count; i++) {
 									add_offset_pair(&result_set, (char *)stringlist[i],
-													offsets[(i<<1)+1] - offsets[i<<1], offsets[i<<1], subpat_names[i]);
+													offsets[(i<<1)+1] - offsets[i<<1], offsets[i<<1], subpat_names[i], unmatched_as_null);
 								}
 							} else {
 								for (i = 0; i < count; i++) {
 									if (subpat_names[i]) {
-									if (offsets[i<<1] < 0) { /* unset substring */
-										add_assoc_null(&result_set, subpat_names[i]);
-									} else {
+										if (unmatched_as_null && offsets[i<<1] < 0) {
+											add_assoc_null(&result_set, subpat_names[i]);
+										} else {
 											add_assoc_stringl(&result_set, subpat_names[i], (char *)stringlist[i],
-																   offsets[(i<<1)+1] - offsets[i<<1]);
-									}
+															  offsets[(i<<1)+1] - offsets[i<<1]);
+										}
 									}
-									if (offsets[i<<1] < 0) { /* unset substring */
+									if (unmatched_as_null && offsets[i<<1] < 0) {
 										add_next_index_null(&result_set);
 									} else {
 										add_next_index_stringl(&result_set, (char *)stringlist[i],
@@ -909,11 +918,11 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 							if (offset_capture) {
 								for (i = 0; i < count; i++) {
 									add_offset_pair(&result_set, (char *)stringlist[i],
-													offsets[(i<<1)+1] - offsets[i<<1], offsets[i<<1], NULL);
+													offsets[(i<<1)+1] - offsets[i<<1], offsets[i<<1], NULL, unmatched_as_null);
 								}
 							} else {
 								for (i = 0; i < count; i++) {
-									if (offsets[i<<1] < 0) { /* unset substring */
+									if (unmatched_as_null && offsets[i<<1] < 0) {
 										add_next_index_null(&result_set);
 									} else {
 										add_next_index_stringl(&result_set, (char *)stringlist[i],
@@ -936,19 +945,19 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 							for (i = 0; i < count; i++) {
 								add_offset_pair(subpats, (char *)stringlist[i],
 												offsets[(i<<1)+1] - offsets[i<<1],
-												offsets[i<<1], subpat_names[i]);
+												offsets[i<<1], subpat_names[i], unmatched_as_null);
 							}
 						} else {
 							for (i = 0; i < count; i++) {
 								if (subpat_names[i]) {
-									if (offsets[i<<1] < 0) { /* unset substring */
+									if (unmatched_as_null && offsets[i<<1] < 0) {
 										add_assoc_null(subpats, subpat_names[i]);
 									} else {
 										add_assoc_stringl(subpats, subpat_names[i], (char *)stringlist[i],
 														  offsets[(i<<1)+1] - offsets[i<<1]);
 									}
 								}
-								if (offsets[i<<1] < 0) { /* unset substring */
+								if (unmatched_as_null && offsets[i<<1] < 0) {
 									add_next_index_null(subpats);
 								} else {
 									add_next_index_stringl(subpats, (char *)stringlist[i],
@@ -961,11 +970,11 @@ PHPAPI void php_pcre_match_impl(pcre_cache_entry *pce, char *subject, int subjec
 							for (i = 0; i < count; i++) {
 								add_offset_pair(subpats, (char *)stringlist[i],
 												offsets[(i<<1)+1] - offsets[i<<1],
-												offsets[i<<1], NULL);
+												offsets[i<<1], NULL, unmatched_as_null);
 							}
 						} else {
 							for (i = 0; i < count; i++) {
-								if (offsets[i<<1] < 0) { /* unset substring */
+								if (unmatched_as_null && offsets[i<<1] < 0) {
 									add_next_index_null(subpats);
 								} else {
 									add_next_index_stringl(subpats, (char *)stringlist[i],
@@ -1869,7 +1878,7 @@ PHPAPI void php_pcre_split_impl(pcre_cache_entry *pce, char *subject, int subjec
 
 				if (offset_capture) {
 					/* Add (match, offset) pair to the return value */
-					add_offset_pair(return_value, last_match, (int)(&subject[offsets[0]]-last_match), next_offset, NULL);
+					add_offset_pair(return_value, last_match, (int)(&subject[offsets[0]]-last_match), next_offset, NULL, 0);
 				} else {
 					/* Add the piece to the return value */
 					ZVAL_STRINGL(&tmp, last_match, &subject[offsets[0]]-last_match);
@@ -1891,7 +1900,7 @@ PHPAPI void php_pcre_split_impl(pcre_cache_entry *pce, char *subject, int subjec
 					/* If we have matched a delimiter */
 					if (!no_empty || match_len > 0) {
 						if (offset_capture) {
-							add_offset_pair(return_value, &subject[offsets[i<<1]], match_len, offsets[i<<1], NULL);
+							add_offset_pair(return_value, &subject[offsets[i<<1]], match_len, offsets[i<<1], NULL, 0);
 						} else {
 							ZVAL_STRINGL(&tmp, &subject[offsets[i<<1]], match_len);
 							zend_hash_next_index_insert_new(Z_ARRVAL_P(return_value), &tmp);
@@ -1928,11 +1937,10 @@ PHPAPI void php_pcre_split_impl(pcre_cache_entry *pce, char *subject, int subjec
 
 	start_offset = (int)(last_match - subject); /* the offset might have been incremented, but without further successful matches */
 
-	if (!no_empty || start_offset < subject_len)
-	{
+	if (!no_empty || start_offset < subject_len) {
 		if (offset_capture) {
 			/* Add the last (match, offset) pair to the return value */
-			add_offset_pair(return_value, &subject[start_offset], subject_len - start_offset, start_offset, NULL);
+			add_offset_pair(return_value, &subject[start_offset], subject_len - start_offset, start_offset, NULL, 0);
 		} else {
 			/* Add the last piece to the return value */
 			ZVAL_STRINGL(&tmp, last_match, subject + subject_len - last_match);

diff --git a/ext/pcre/tests/001.phpt b/ext/pcre/tests/001.phpt
@@ -52,7 +52,7 @@ array(10) {
   [2]=>
   string(2) "06"
   [3]=>
-  NULL
+  string(0) ""
   ["month"]=>
   string(2) "12"
   [4]=>
@@ -75,7 +75,7 @@ array(10) {
   [2]=>
   string(2) "12"
   [3]=>
-  NULL
+  string(0) ""
   ["month"]=>
   string(3) "Aug"
   [4]=>

diff --git a/ext/pcre/tests/003.phpt b/ext/pcre/tests/003.phpt
@@ -58,7 +58,7 @@ array(10) {
     [0]=>
     string(2) "20"
     [1]=>
-    NULL
+    string(0) ""
   }
   ["month"]=>
   array(2) {
@@ -127,7 +127,7 @@ array(2) {
     [2]=>
     string(2) "12"
     [3]=>
-    NULL
+    string(0) ""
     ["month"]=>
     string(3) "Aug"
     [4]=>

diff --git a/ext/pcre/tests/004.phpt b/ext/pcre/tests/004.phpt
@@ -24,7 +24,7 @@ array(2) {
     [1]=>
     string(12) "unsigned int"
     [2]=>
-    NULL
+    string(0) ""
     [3]=>
     string(0) ""
     [4]=>
@@ -41,13 +41,13 @@ array(2) {
     [1]=>
     string(5) "short"
     [2]=>
-    NULL
+    string(0) ""
     [3]=>
     string(0) ""
     [4]=>
     string(1) "a"
     [5]=>
-    NULL
+    string(0) ""
     [6]=>
     string(3) ", b"
   }

diff --git a/ext/pcre/tests/bug61780.phpt b/ext/pcre/tests/bug61780.phpt
@@ -2,7 +2,7 @@
 Bug #61780 (Inconsistent PCRE captures in match results): basics
 --FILE--
 <?php
-preg_match('/(a)?([a-z]*)(\d*)/', '123', $matches);
+preg_match('/(a)?([a-z]*)(\d*)/', '123', $matches, PREG_UNMATCHED_AS_NULL);
 var_dump($matches);
 ?>
 --EXPECT--

diff --git a/ext/pcre/tests/bug61780_1.phpt b/ext/pcre/tests/bug61780_1.phpt
@@ -2,22 +2,22 @@
 Bug #61780 (Inconsistent PCRE captures in match results): numeric subpatterns
 --FILE--
 <?php
-preg_match('/(4)?(2)?\d/', '23456', $matches);
+preg_match('/(4)?(2)?\d/', '23456', $matches, PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match('/(4)?(2)?\d/', '23456', $matches, PREG_OFFSET_CAPTURE);
+preg_match('/(4)?(2)?\d/', '23456', $matches, PREG_OFFSET_CAPTURE | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(4)?(2)?\d/', '123456', $matches);
+preg_match_all('/(4)?(2)?\d/', '123456', $matches, PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(4)?(2)?\d/', '123456', $matches, PREG_OFFSET_CAPTURE);
+preg_match_all('/(4)?(2)?\d/', '123456', $matches, PREG_OFFSET_CAPTURE | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(4)?(2)?\d/', '123456', $matches, PREG_SET_ORDER);
+preg_match_all('/(4)?(2)?\d/', '123456', $matches, PREG_SET_ORDER | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(4)?(2)?\d/', '123456', $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
+preg_match_all('/(4)?(2)?\d/', '123456', $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 ?>
 --EXPECT--

diff --git a/ext/pcre/tests/bug61780_2.phpt b/ext/pcre/tests/bug61780_2.phpt
@@ -2,22 +2,22 @@
 Bug #61780 (Inconsistent PCRE captures in match results): named subpatterns
 --FILE--
 <?php
-preg_match('/(?<a>4)?(?<b>2)?\d/', '23456', $matches);
+preg_match('/(?<a>4)?(?<b>2)?\d/', '23456', $matches, PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match('/(?<a>4)?(?<b>2)?\d/', '23456', $matches, PREG_OFFSET_CAPTURE);
+preg_match('/(?<a>4)?(?<b>2)?\d/', '23456', $matches, PREG_OFFSET_CAPTURE | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches);
+preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches, PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches, PREG_OFFSET_CAPTURE);
+preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches, PREG_OFFSET_CAPTURE | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches, PREG_SET_ORDER);
+preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches, PREG_SET_ORDER | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 echo "\n\n";
-preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
+preg_match_all('/(?<a>4)?(?<b>2)?\d/', '123456', $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE | PREG_UNMATCHED_AS_NULL);
 var_export($matches);
 ?>
 --EXPECT--