1--TEST--
2Test mb_ereg() function : usage variations - match special characters
3--SKIPIF--
4<?php
5extension_loaded('mbstring') or die('skip');
6function_exists('mb_ereg') or die("skip mb_ereg() is not available in this build");
7?>
8--FILE--
9<?php
10/* Prototype  : int mb_ereg(string $pattern, string $string [, array $registers])
11 * Description: Regular expression match for multibyte string
12 * Source code: ext/mbstring/php_mbregex.c
13 */
14
15/*
16 * Test how mb_ereg() matches special characters for $pattern
17 */
18
19echo "*** Testing mb_ereg() : usage variations ***\n";
20
21if(mb_regex_encoding('utf-8') == true) {
22	echo "Regex encoding set to utf-8\n";
23} else {
24	echo "Could not set regex encoding to utf-8\n";
25}
26
27$regex_char = array ('\w+' => b'\w+',
28                     '\W+' => b'\W+',
29                     '\s+' => b'\s+',
30                     '\S+' => b'\S+',
31                     '\d+' => b'\d+',
32                     '\D+' => b'\D+',
33                     '\b' =>  b'\b',
34                     '\B' =>  b'\B');
35
36$string_ascii = b'This is an English string. 0123456789.';
37$string_mb = base64_decode('5pel5pys6Kqe44OG44Kt44K544OI44Gn44GZ44CCMDEyMzTvvJXvvJbvvJfvvJjvvJnjgII=');
38
39foreach ($regex_char as $displayChar => $char) {
40	echo "\n--** Pattern is: $displayChar **--\n";
41	if (@$regs_ascii || @$regs_mb) {
42		$regs_ascii = null;
43		$regs_mb = null;
44	}
45	echo "-- ASCII String: --\n";
46	var_dump(mb_ereg($char, $string_ascii, $regs_ascii));
47	base64_encode_var_dump($regs_ascii);
48
49	echo "-- Multibyte String: --\n";
50	var_dump(mb_ereg($char, $string_mb, $regs_mb));
51	base64_encode_var_dump($regs_mb);
52
53}
54
55/**
56 * replicate a var dump of an array but outputted string values are base64 encoded
57 *
58 * @param array $regs
59 */
60function base64_encode_var_dump($regs) {
61	if ($regs) {
62		echo "array(" . count($regs) . ") {\n";
63		foreach ($regs as $key => $value) {
64			echo "  [$key]=>\n  ";
65			if (is_string($value)) {
66				var_dump(base64_encode($value));
67			} else {
68				var_dump($value);
69			}
70		}
71		echo "}\n";
72	} else {
73		echo "NULL\n";
74	}
75}
76
77echo "Done";
78
79?>
80--EXPECT--
81*** Testing mb_ereg() : usage variations ***
82Regex encoding set to utf-8
83
84--** Pattern is: \w+ **--
85-- ASCII String: --
86int(4)
87array(1) {
88  [0]=>
89  string(8) "VGhpcw=="
90}
91-- Multibyte String: --
92int(27)
93array(1) {
94  [0]=>
95  string(36) "5pel5pys6Kqe44OG44Kt44K544OI44Gn44GZ"
96}
97
98--** Pattern is: \W+ **--
99-- ASCII String: --
100int(1)
101array(1) {
102  [0]=>
103  string(4) "IA=="
104}
105-- Multibyte String: --
106int(3)
107array(1) {
108  [0]=>
109  string(4) "44CC"
110}
111
112--** Pattern is: \s+ **--
113-- ASCII String: --
114int(1)
115array(1) {
116  [0]=>
117  string(4) "IA=="
118}
119-- Multibyte String: --
120bool(false)
121NULL
122
123--** Pattern is: \S+ **--
124-- ASCII String: --
125int(4)
126array(1) {
127  [0]=>
128  string(8) "VGhpcw=="
129}
130-- Multibyte String: --
131int(53)
132array(1) {
133  [0]=>
134  string(72) "5pel5pys6Kqe44OG44Kt44K544OI44Gn44GZ44CCMDEyMzTvvJXvvJbvvJfvvJjvvJnjgII="
135}
136
137--** Pattern is: \d+ **--
138-- ASCII String: --
139int(10)
140array(1) {
141  [0]=>
142  string(16) "MDEyMzQ1Njc4OQ=="
143}
144-- Multibyte String: --
145int(20)
146array(1) {
147  [0]=>
148  string(28) "MDEyMzTvvJXvvJbvvJfvvJjvvJk="
149}
150
151--** Pattern is: \D+ **--
152-- ASCII String: --
153int(27)
154array(1) {
155  [0]=>
156  string(36) "VGhpcyBpcyBhbiBFbmdsaXNoIHN0cmluZy4g"
157}
158-- Multibyte String: --
159int(30)
160array(1) {
161  [0]=>
162  string(40) "5pel5pys6Kqe44OG44Kt44K544OI44Gn44GZ44CC"
163}
164
165--** Pattern is: \b **--
166-- ASCII String: --
167int(1)
168array(1) {
169  [0]=>
170  bool(false)
171}
172-- Multibyte String: --
173int(1)
174array(1) {
175  [0]=>
176  bool(false)
177}
178
179--** Pattern is: \B **--
180-- ASCII String: --
181int(1)
182array(1) {
183  [0]=>
184  bool(false)
185}
186-- Multibyte String: --
187int(1)
188array(1) {
189  [0]=>
190  bool(false)
191}
192Done