Allow empty needles in mb_strpos and mb_strstr function family.

MBstring analogous implementation to 6d578482a933be7597b686b59a935b316161d251

Closes GH-4977

diff --git a/ext/mbstring/libmbfl/mbfl/mbfilter.c b/ext/mbstring/libmbfl/mbfl/mbfilter.c
index 1cb6d28e7b4e0c1bd3bdc538cd5363bf2a06fe72..6b0aff2a4e04a4a41ee12990d1bf7abd460767a0 100644 (file)
--- a/ext/mbstring/libmbfl/mbfl/mbfilter.c
+++ b/ext/mbstring/libmbfl/mbfl/mbfilter.c
@@ -656,7 +656,7 @@ filter_count_output(int c, void *data)
 }
 
 size_t
-mbfl_strlen(mbfl_string *string)
+mbfl_strlen(const mbfl_string *string)
 {
        size_t len, n, k;
        unsigned char *p;
@@ -855,13 +855,29 @@ mbfl_strpos(
                needle_u8 = needle;
        }
 
-       if (needle_u8->len < 1) {
-               result = (size_t) -8;
+       result = (size_t) -1;
+       if (haystack_u8->len < needle_u8->len) {
                goto out;
        }
 
-       result = (size_t) -1;
-       if (haystack_u8->len < needle_u8->len) {
+       if (needle_u8->len == 0) {
+               size_t haystack_length = mbfl_strlen(haystack_u8);
+               /* Check if offset is out of bound */
+               if (
+                       (offset > 0 && offset > haystack_length)
+                       || (offset < 0 && -offset > haystack_length)
+               ) {
+                       result = -16;
+                       goto out;
+               }
+
+               if (offset < 0) {
+                       result = haystack_length + offset;
+               } else if (reverse) {
+                       result = haystack_length;
+               } else {
+                       result = (size_t) offset;
+               }
                goto out;
        }
 

diff --git a/ext/mbstring/libmbfl/mbfl/mbfilter.h b/ext/mbstring/libmbfl/mbfl/mbfilter.h
index 0966e2df44b57bd41e25506af9fffb4eb293733e..edba946f72087bd5db88b2c56e698195b1d4b6c6 100644 (file)
--- a/ext/mbstring/libmbfl/mbfl/mbfilter.h
+++ b/ext/mbstring/libmbfl/mbfl/mbfilter.h
@@ -193,7 +193,7 @@ static inline int mbfl_is_error(size_t len) {
  * strlen
  */
 MBFLAPI extern size_t
-mbfl_strlen(mbfl_string *string);
+mbfl_strlen(const mbfl_string *string);
 
 /*
  * oddlen

diff --git a/ext/mbstring/mbstring.c b/ext/mbstring/mbstring.c
index a0dbb3a30ea9eeddadd2abf6585a5eaf2af13685..1615dc2178cf3e912c5e58a440ed8be80c3e12af 100644 (file)
--- a/ext/mbstring/mbstring.c
+++ b/ext/mbstring/mbstring.c
@@ -2110,11 +2110,6 @@ PHP_FUNCTION(mb_strpos)
                }
        }
 
-       if (needle.len == 0) {
-               php_error_docref(NULL, E_WARNING, "Empty delimiter");
-               RETURN_FALSE;
-       }
-
        n = mbfl_strpos(&haystack, &needle, offset, reverse);
        if (!mbfl_is_error(n)) {
                RETVAL_LONG(n);
@@ -2189,11 +2184,6 @@ PHP_FUNCTION(mb_stripos)
                RETURN_THROWS();
        }
 
-       if (needle.len == 0) {
-               php_error_docref(NULL, E_WARNING, "Empty delimiter");
-               RETURN_FALSE;
-       }
-
        n = php_mb_stripos(0, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, offset, from_encoding);
 
        if (!mbfl_is_error(n)) {
@@ -2246,11 +2236,6 @@ PHP_FUNCTION(mb_strstr)
                RETURN_FALSE;
        }
 
-       if (needle.len == 0) {
-               php_error_docref(NULL, E_WARNING, "Empty delimiter");
-               RETURN_FALSE;
-       }
-
        n = mbfl_strpos(&haystack, &needle, 0, 0);
        if (!mbfl_is_error(n)) {
                if (part) {
@@ -2350,11 +2335,6 @@ PHP_FUNCTION(mb_stristr)
                RETURN_FALSE;
        }
 
-       if (!needle.len) {
-               php_error_docref(NULL, E_WARNING, "Empty delimiter");
-               RETURN_FALSE;
-       }
-
        n = php_mb_stripos(0, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, 0, from_encoding);
        if (mbfl_is_error(n)) {
                RETURN_FALSE;
@@ -4849,10 +4829,6 @@ MBSTRING_API size_t php_mb_stripos(int mode, const char *old_haystack, size_t ol
                        break;
                }
 
-               if (needle.len == 0) {
-                       break;
-               }
-
                if (offset != 0) {
                        size_t haystack_char_len = mbfl_strlen(&haystack);
 

diff --git a/ext/mbstring/tests/mb_stripos_empty_needle.phpt b/ext/mbstring/tests/mb_stripos_empty_needle.phpt
new file mode 100644 (file)
index 0000000..31e21f1
--- /dev/null
+++ b/ext/mbstring/tests/mb_stripos_empty_needle.phpt
@@ -0,0 +1,86 @@
+--TEST--
+Test mb_stripos() function :  with empty needle
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_stripos') or die("skip mb_stripos() is not available in this build");
+?>
+--FILE--
+<?php
+
+mb_internal_encoding('UTF-8');
+
+$string_ascii = 'abc def';
+// Japanese string in UTF-8
+$string_mb = "日本語テキストです。01234５６７８９。";
+
+echo "\n-- ASCII string without offset --\n";
+var_dump(mb_stripos($string_ascii, ''));
+
+echo "\n-- ASCII string with in range positive offset --\n";
+var_dump(mb_stripos($string_ascii, '', 2));
+
+echo "\n-- ASCII string with in range negative offset --\n";
+var_dump(mb_stripos($string_ascii, '', -2));
+
+echo "\n-- ASCII string with out of bound positive offset --\n";
+var_dump(mb_stripos($string_ascii, '', 150));
+
+echo "\n-- ASCII string with out of bound negative offset --\n";
+var_dump(mb_stripos($string_ascii, '', -150));
+
+
+echo "\n-- Multi-byte string without offset --\n";
+var_dump(mb_stripos($string_mb, ''));
+
+echo "\n-- Multi-byte string with in range positive offset --\n";
+var_dump(mb_stripos($string_mb, '', 2));
+
+echo "\n-- Multi-byte string with in range negative offset --\n";
+var_dump(mb_stripos($string_mb, '', -2));
+
+echo "\n-- Multi-byte string with out of bound positive offset --\n";
+var_dump(mb_stripos($string_mb, '', 150));
+
+echo "\n-- Multi-byte string with out of bound negative offset --\n";
+var_dump(mb_stripos($string_mb, '', -150));
+
+?>
+--EXPECTF--
+-- ASCII string without offset --
+int(0)
+
+-- ASCII string with in range positive offset --
+int(2)
+
+-- ASCII string with in range negative offset --
+int(5)
+
+-- ASCII string with out of bound positive offset --
+
+Warning: mb_stripos(): Offset not contained in string in %s on line %d
+bool(false)
+
+-- ASCII string with out of bound negative offset --
+
+Warning: mb_stripos(): Offset not contained in string in %s on line %d
+bool(false)
+
+-- Multi-byte string without offset --
+int(0)
+
+-- Multi-byte string with in range positive offset --
+int(2)
+
+-- Multi-byte string with in range negative offset --
+int(19)
+
+-- Multi-byte string with out of bound positive offset --
+
+Warning: mb_stripos(): Offset not contained in string in %s on line %d
+bool(false)
+
+-- Multi-byte string with out of bound negative offset --
+
+Warning: mb_stripos(): Offset not contained in string in %s on line %d
+bool(false)

diff --git a/ext/mbstring/tests/mb_stristr_empty_needle.phpt b/ext/mbstring/tests/mb_stristr_empty_needle.phpt
new file mode 100644 (file)
index 0000000..60850a9
--- /dev/null
+++ b/ext/mbstring/tests/mb_stristr_empty_needle.phpt
@@ -0,0 +1,37 @@
+--TEST--
+Test mb_stristr() function : with empty needle
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_stristr') or die("skip mb_stristr() is not available in this build");
+?>
+--FILE--
+<?php
+
+mb_internal_encoding('UTF-8');
+
+$string_ascii = 'abc def';
+// Japanese string in UTF-8
+$string_mb = "日本語テキストです。01234５６７８９。";
+
+echo "\n-- ASCII string --\n";
+var_dump(mb_stristr($string_ascii, '', false, 'ISO-8859-1'));
+var_dump(mb_stristr($string_ascii, ''));
+var_dump(mb_stristr($string_ascii, '', true));
+
+echo "\n-- Multibyte string --\n";
+var_dump(mb_stristr($string_mb, ''));
+var_dump(mb_stristr($string_mb, '', false, 'utf-8'));
+var_dump(mb_stristr($string_mb, '', true));
+
+?>
+--EXPECT--
+-- ASCII string --
+string(7) "abc def"
+string(7) "abc def"
+string(0) ""
+
+-- Multibyte string --
+string(53) "日本語テキストです。01234５６７８９。"
+string(53) "日本語テキストです。01234５６７８９。"
+string(0) ""

diff --git a/ext/mbstring/tests/mb_strpos_empty_needle.phpt b/ext/mbstring/tests/mb_strpos_empty_needle.phpt
new file mode 100644 (file)
index 0000000..3161264
--- /dev/null
+++ b/ext/mbstring/tests/mb_strpos_empty_needle.phpt
@@ -0,0 +1,86 @@
+--TEST--
+Test mb_strpos() function : with empty needle
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_strpos') or die("skip mb_strpos() is not available in this build");
+?>
+--FILE--
+<?php
+
+mb_internal_encoding('UTF-8');
+
+$string_ascii = 'abc def';
+// Japanese string in UTF-8
+$string_mb = "日本語テキストです。01234５６７８９。";
+
+echo "\n-- ASCII string without offset --\n";
+var_dump(mb_strpos($string_ascii, ''));
+
+echo "\n-- ASCII string with in range positive offset --\n";
+var_dump(mb_strpos($string_ascii, '', 2));
+
+echo "\n-- ASCII string with in range negative offset --\n";
+var_dump(mb_strpos($string_ascii, '', -2));
+
+echo "\n-- ASCII string with out of bound positive offset --\n";
+var_dump(mb_strpos($string_ascii, '', 15));
+
+echo "\n-- ASCII string with out of bound negative offset --\n";
+var_dump(mb_strpos($string_ascii, '', -15));
+
+
+echo "\n-- Multi-byte string without offset --\n";
+var_dump(mb_strpos($string_mb, ''));
+
+echo "\n-- Multi-byte string with in range positive offset --\n";
+var_dump(mb_strpos($string_mb, '', 2));
+
+echo "\n-- Multi-byte string with in range negative offset --\n";
+var_dump(mb_strpos($string_mb, '', -2));
+
+echo "\n-- Multi-byte string with out of bound positive offset --\n";
+var_dump(mb_strpos($string_mb, '', 150));
+
+echo "\n-- Multi-byte string with out of bound negative offset --\n";
+var_dump(mb_strpos($string_mb, '', -150));
+
+?>
+--EXPECTF--
+-- ASCII string without offset --
+int(0)
+
+-- ASCII string with in range positive offset --
+int(2)
+
+-- ASCII string with in range negative offset --
+int(5)
+
+-- ASCII string with out of bound positive offset --
+
+Warning: mb_strpos(): Offset not contained in string in %s on line %d
+bool(false)
+
+-- ASCII string with out of bound negative offset --
+
+Warning: mb_strpos(): Offset not contained in string in %s on line %d
+bool(false)
+
+-- Multi-byte string without offset --
+int(0)
+
+-- Multi-byte string with in range positive offset --
+int(2)
+
+-- Multi-byte string with in range negative offset --
+int(19)
+
+-- Multi-byte string with out of bound positive offset --
+
+Warning: mb_strpos(): Offset not contained in string in %s on line %d
+bool(false)
+
+-- Multi-byte string with out of bound negative offset --
+
+Warning: mb_strpos(): Offset not contained in string in %s on line %d
+bool(false)

diff --git a/ext/mbstring/tests/mb_strrchr_empty_needle.phpt b/ext/mbstring/tests/mb_strrchr_empty_needle.phpt
new file mode 100644 (file)
index 0000000..1efacc9
--- /dev/null
+++ b/ext/mbstring/tests/mb_strrchr_empty_needle.phpt
@@ -0,0 +1,37 @@
+--TEST--
+Test mb_strrchr() function : with empty needle
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_strrchr') or die("skip mb_strrchr() is not available in this build");
+?>
+--FILE--
+<?php
+
+mb_internal_encoding('UTF-8');
+
+$string_ascii = 'abc def';
+// Japanese string in UTF-8
+$string_mb = "日本語テキストです。01234５６７８９。";
+
+echo "\n-- ASCII string --\n";
+var_dump(bin2hex(mb_strrchr($string_ascii, '', false, 'ISO-8859-1')));
+var_dump(bin2hex(mb_strrchr($string_ascii, '')));
+var_dump(bin2hex(mb_strrchr($string_ascii, '', true)));
+
+echo "\n-- Multibyte string --\n";
+var_dump(bin2hex(mb_strrchr($string_mb, '')));
+var_dump(bin2hex(mb_strrchr($string_mb, '', false, 'utf-8')));
+var_dump(bin2hex(mb_strrchr($string_mb, '', true)));
+
+?>
+--EXPECT--
+-- ASCII string --
+string(0) ""
+string(0) ""
+string(0) ""
+
+-- Multibyte string --
+string(0) ""
+string(0) ""
+string(0) ""

diff --git a/ext/mbstring/tests/mb_strripos_empty_needle.phpt b/ext/mbstring/tests/mb_strripos_empty_needle.phpt
new file mode 100644 (file)
index 0000000..2eaf8cb
--- /dev/null
+++ b/ext/mbstring/tests/mb_strripos_empty_needle.phpt
@@ -0,0 +1,86 @@
+--TEST--
+Test mb_strripos() function : with empty needle
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_strripos') or die("skip mb_strripos() is not available in this build");
+?>
+--FILE--
+<?php
+
+mb_internal_encoding('UTF-8');
+
+$string_ascii = 'abc def';
+// Japanese string in UTF-8
+$string_mb = "日本語テキストです。01234５６７８９。";
+
+echo "\n-- ASCII string without offset --\n";
+var_dump(mb_strripos($string_ascii, ''));
+
+echo "\n-- ASCII string with in range positive offset --\n";
+var_dump(mb_strripos($string_ascii, '', 2));
+
+echo "\n-- ASCII string with in range negative offset --\n";
+var_dump(mb_strripos($string_ascii, '', -2));
+
+echo "\n-- ASCII string with out of bound positive offset --\n";
+var_dump(mb_strripos($string_ascii, '', 15));
+
+echo "\n-- ASCII string with out of bound negative offset --\n";
+var_dump(mb_strripos($string_ascii, '', -15));
+
+
+echo "\n-- Multi-byte string without offset --\n";
+var_dump(mb_strripos($string_mb, ''));
+
+echo "\n-- Multi-byte string with in range positive offset --\n";
+var_dump(mb_strripos($string_mb, '', 2));
+
+echo "\n-- Multi-byte string with in range negative offset --\n";
+var_dump(mb_strripos($string_mb, '', -2));
+
+echo "\n-- Multi-byte string with out of bound positive offset --\n";
+var_dump(mb_strripos($string_mb, '', 150));
+
+echo "\n-- Multi-byte string with out of bound negative offset --\n";
+var_dump(mb_strripos($string_mb, '', -150));
+
+?>
+--EXPECTF--
+-- ASCII string without offset --
+int(7)
+
+-- ASCII string with in range positive offset --
+int(7)
+
+-- ASCII string with in range negative offset --
+int(5)
+
+-- ASCII string with out of bound positive offset --
+
+Warning: mb_strripos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)
+
+-- ASCII string with out of bound negative offset --
+
+Warning: mb_strripos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)
+
+-- Multi-byte string without offset --
+int(21)
+
+-- Multi-byte string with in range positive offset --
+int(21)
+
+-- Multi-byte string with in range negative offset --
+int(19)
+
+-- Multi-byte string with out of bound positive offset --
+
+Warning: mb_strripos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)
+
+-- Multi-byte string with out of bound negative offset --
+
+Warning: mb_strripos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)

diff --git a/ext/mbstring/tests/mb_strrpos_empty_needle.phpt b/ext/mbstring/tests/mb_strrpos_empty_needle.phpt
new file mode 100644 (file)
index 0000000..a56c9c3
--- /dev/null
+++ b/ext/mbstring/tests/mb_strrpos_empty_needle.phpt
@@ -0,0 +1,86 @@
+--TEST--
+Test mb_strrpos() function :  with empty needle
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_strrpos') or die("skip mb_strrpos() is not available in this build");
+?>
+--FILE--
+<?php
+
+mb_internal_encoding('UTF-8');
+
+$string_ascii = 'abc def';
+// Japanese string in UTF-8
+$string_mb = "日本語テキストです。01234５６７８９。";
+
+echo "\n-- ASCII string without offset --\n";
+var_dump(mb_strrpos($string_ascii, ''));
+
+echo "\n-- ASCII string with in range positive offset --\n";
+var_dump(mb_strrpos($string_ascii, '', 2));
+
+echo "\n-- ASCII string with in range negative offset --\n";
+var_dump(mb_strrpos($string_ascii, '', -2));
+
+echo "\n-- ASCII string with out of bound positive offset --\n";
+var_dump(mb_strrpos($string_ascii, '', 15));
+
+echo "\n-- ASCII string with out of bound negative offset --\n";
+var_dump(mb_strrpos($string_ascii, '', -15));
+
+
+echo "\n-- Multi-byte string without offset --\n";
+var_dump(mb_strrpos($string_mb, ''));
+
+echo "\n-- Multi-byte string with in range positive offset --\n";
+var_dump(mb_strrpos($string_mb, '', 2));
+
+echo "\n-- Multi-byte string with in range negative offset --\n";
+var_dump(mb_strrpos($string_mb, '', -2));
+
+echo "\n-- Multi-byte string with out of bound positive offset --\n";
+var_dump(mb_strrpos($string_mb, '', 150));
+
+echo "\n-- Multi-byte string with out of bound negative offset --\n";
+var_dump(mb_strrpos($string_mb, '', -150));
+
+?>
+--EXPECTF--
+-- ASCII string without offset --
+int(7)
+
+-- ASCII string with in range positive offset --
+int(7)
+
+-- ASCII string with in range negative offset --
+int(5)
+
+-- ASCII string with out of bound positive offset --
+
+Warning: mb_strrpos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)
+
+-- ASCII string with out of bound negative offset --
+
+Warning: mb_strrpos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)
+
+-- Multi-byte string without offset --
+int(21)
+
+-- Multi-byte string with in range positive offset --
+int(21)
+
+-- Multi-byte string with in range negative offset --
+int(19)
+
+-- Multi-byte string with out of bound positive offset --
+
+Warning: mb_strrpos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)
+
+-- Multi-byte string with out of bound negative offset --
+
+Warning: mb_strrpos(): Offset is greater than the length of haystack string in %s on line %d
+bool(false)

diff --git a/ext/mbstring/tests/mb_strstr_empty_needle.phpt b/ext/mbstring/tests/mb_strstr_empty_needle.phpt
new file mode 100644 (file)
index 0000000..fd72a77
--- /dev/null
+++ b/ext/mbstring/tests/mb_strstr_empty_needle.phpt
@@ -0,0 +1,37 @@
+--TEST--
+Test mb_strstr() function : with empty needle
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip');
+function_exists('mb_strstr') or die("skip mb_strstr() is not available in this build");
+?>
+--FILE--
+<?php
+
+mb_internal_encoding('UTF-8');
+
+$string_ascii = 'abc def';
+// Japanese string in UTF-8
+$string_mb = "日本語テキストです。01234５６７８９。";
+
+echo "\n-- ASCII string --\n";
+var_dump(mb_strstr($string_ascii, '', false, 'ISO-8859-1'));
+var_dump(mb_strstr($string_ascii, ''));
+var_dump(mb_strstr($string_ascii, '', true));
+
+echo "\n-- Multibyte string --\n";
+var_dump(mb_strstr($string_mb, ''));
+var_dump(mb_strstr($string_mb, '', false, 'utf-8'));
+var_dump(mb_strstr($string_mb, '', true));
+
+?>
+--EXPECT--
+-- ASCII string --
+string(7) "abc def"
+string(7) "abc def"
+string(0) ""
+
+-- Multibyte string --
+string(53) "日本語テキストです。01234５６７８９。"
+string(53) "日本語テキストです。01234５６７８９。"
+string(0) ""