mb_split() can now handle empty matches like preg_split() does.

Author: moriyoshi

NEWS

@@ -5,6 +5,9 @@ PHP                                                                        NEWS
 - CLI server:
   . Fixed bug #64128 (buit-in web server is broken on ppc64). (Remi)
 
+- Mbstring:
+  . mb_split() can now handle empty matches like preg_split() does. (Moriyoshi)
+
 ?? ??? 2012, PHP 5.4.12
 
 - Core:

ext/mbstring/php_mbregex.c

@@ -1055,7 +1055,7 @@ PHP_FUNCTION(mb_split)
 	php_mb_regex_t *re;
 	OnigRegion *regs = NULL;
 	char *string;
-	OnigUChar *pos;
+	OnigUChar *pos, *chunk_pos;
 	int string_len;
 
 	int n, err;
@@ -1065,8 +1065,8 @@ PHP_FUNCTION(mb_split)
 		RETURN_FALSE;
 	} 
 
-	if (count == 0) {
-		count = 1;
+	if (count > 0) {
+		count--;
 	}
 
 	/* create regex pattern buffer */
@@ -1076,31 +1076,30 @@ PHP_FUNCTION(mb_split)
 
 	array_init(return_value);
 
-	pos = (OnigUChar *)string;
+	chunk_pos = pos = (OnigUChar *)string;
 	err = 0;
 	regs = onig_region_new();
 	/* churn through str, generating array entries as we go */
-	while ((--count != 0) &&
-		   (err = onig_search(re, (OnigUChar *)string, (OnigUChar *)(string + string_len), pos, (OnigUChar *)(string + string_len), regs, 0)) >= 0) {
-		if (regs->beg[0] == regs->end[0]) {
-			php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty regular expression");
+	while (count != 0 && (pos - (OnigUChar *)string) < string_len) {
+		int beg, end;
+		err = onig_search(re, (OnigUChar *)string, (OnigUChar *)(string + string_len), pos, (OnigUChar *)(string + string_len), regs, 0);
+		if (err < 0) {
 			break;
 		}
-
+		beg = regs->beg[0], end = regs->end[0];
 		/* add it to the array */
-		if (regs->beg[0] < string_len && regs->beg[0] >= (pos - (OnigUChar *)string)) {
-			add_next_index_stringl(return_value, (char *)pos, ((OnigUChar *)(string + regs->beg[0]) - pos), 1);
+		if ((pos - (OnigUChar *)string) < end) {
+			if (beg < string_len && beg >= (chunk_pos - (OnigUChar *)string)) {
+				add_next_index_stringl(return_value, (char *)chunk_pos, ((OnigUChar *)(string + beg) - chunk_pos), 1);
+				--count;
+			} else {
+				err = -2;
+				break;
+			}
+			/* point at our new starting point */
+			chunk_pos = pos = (OnigUChar *)string + end;
 		} else {
-			err = -2;
-			break;
-		}
-		/* point at our new starting point */
-		n = regs->end[0];
-		if ((pos - (OnigUChar *)string) < n) {
-			pos = (OnigUChar *)string + n;
-		}
-		if (count < 0) {
-			count = 0;
+			pos++;
 		}
 		onig_region_free(regs, 0);
 	}
@@ -1117,9 +1116,9 @@ PHP_FUNCTION(mb_split)
 	}
 
 	/* otherwise we just have one last element to add to the array */
-	n = ((OnigUChar *)(string + string_len) - pos);
+	n = ((OnigUChar *)(string + string_len) - chunk_pos);
 	if (n > 0) {
-		add_next_index_stringl(return_value, (char *)pos, n, 1);
+		add_next_index_stringl(return_value, (char *)chunk_pos, n, 1);
 	} else {
 		add_next_index_stringl(return_value, "", 0, 1);
 	}

ext/mbstring/tests/mb_split_empty_match.phpt

@@ -0,0 +1,23 @@
+--TEST--
+mb_split() empty match
+--
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_split') or die("skip mb_split() is not available in this build");
+?>
+--FILE--
+<?php
+mb_regex_set_options('m');
+var_dump(mb_split('^', "a\nb\nc"));
+--EXPECT--
+array(3) {
+  [0]=>
+  string(2) "a
+"
+  [1]=>
+  string(2) "b
+"
+  [2]=>
+  string(1) "c"
+}

ext/mbstring/tests/mb_split_variation1.phpt

@@ -156,16 +156,12 @@ array(1) {
 }
 
 -- Iteration 10 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
 }
 
 -- Iteration 11 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
@@ -178,8 +174,6 @@ array(1) {
 }
 
 -- Iteration 13 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
@@ -192,24 +186,18 @@ array(1) {
 }
 
 -- Iteration 15 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
 }
 
 -- Iteration 16 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
 }
 
 -- Iteration 17 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
@@ -240,16 +228,12 @@ array(1) {
 }
 
 -- Iteration 22 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
 }
 
 -- Iteration 23 --
-
-Warning: mb_split(): Empty regular expression in %s on line %d
 array(1) {
   [0]=>
   string(13) "a b c d e f g"
@@ -259,4 +243,4 @@ array(1) {
 
 Warning: mb_split() expects parameter 1 to be string, resource given in %s on line %d
 bool(false)
-Done
+Done