1--TEST-- 2mb_convert_variables() 3--EXTENSIONS-- 4mbstring 5--INI-- 6output_handler= 7mbstring.language=Japanese 8--FILE-- 9<?php 10// TODO: Add more tests 11//$debug = true; // Uncomment this line to view error/warning/notice message in *.out file 12ini_set('include_path', __DIR__); 13include_once('common.inc'); 14 15// SJIS string (BASE64 encoded) 16$sjis = base64_decode('k/qWe4zqg2WDTINYg2eCxYK3gUIwMTIzNIJUglWCVoJXgliBQg=='); 17// JIS string (BASE64 encoded) 18$jis = base64_decode('GyRCRnxLXDhsJUYlLSU5JUgkRyQ5ISMbKEIwMTIzNBskQiM1IzYjNyM4IzkhIxsoQg=='); 19// EUC-JP string 20$euc_jp = mb_convert_encoding("日本語テキストです。01234５６７８９。", 'EUC-JP', 'UTF-8'); 21 22// Test for single scalar 23echo "== SCALAR TEST ==\n"; 24$s = $sjis; 25$encoding = mb_convert_variables('EUC-JP', 'SJIS', $s); 26print("$encoding\n"); // SJIS 27echo bin2hex($s), "\n"; // Converted to EUC-JP 28 29$s = $jis; 30$encoding = mb_convert_variables('EUC-JP', 'JIS', $s); 31print("$encoding\n"); // JIS 32echo bin2hex($s), "\n"; // Converted to EUC-JP 33 34$s = $euc_jp; 35$encoding = mb_convert_variables('SJIS', 'EUC-JP', $s); 36print("$encoding\n"); // EUC-JP 37print(base64_encode($s)."\n"); // Converted to SJIS (base64 encoded) 38 39$s = $euc_jp; 40$encoding = mb_convert_variables('JIS', 'EUC-JP', $s); 41print("$encoding\n"); // EUC-JP 42print(base64_encode($s)."\n"); // Converted to JIS (base64 encoded) 43 44// Test for multiple scalar 45$s1 = $euc_jp; 46$s2 = $euc_jp; 47$s3 = $euc_jp; 48$encoding = mb_convert_variables('EUC-JP', 'auto', $s1, $s2, $s3); 49print("$encoding\n"); // EUC-JP 50echo bin2hex("$s1$s2$s3"), "\n"; // Converted to EUC-JP 51 52// Note: Mixing encoding in array/object is not supported 53echo "== ARRAY TEST ==\n"; 54$a = array($s3, $s2, $s1); 55$aa = $a; 56$encoding = mb_convert_variables('EUC-JP', 'auto', $aa); 57print("$encoding\n"); // EUC-JP 58echo bin2hex("{$aa[0]}{$aa[1]}{$aa[2]}"), "\n"; // Converted to EUC-JP 59 60$a = array($s1, $s2, $s3); 61$aa = $a; 62$encoding = mb_convert_variables('EUC-JP', 'auto', $aa); 63print("$encoding\n"); // EUC-JP 64echo bin2hex("{$aa[0]}{$aa[1]}{$aa[2]}"), "\n"; // Converted to EUC-JP 65 66// Test for object 67echo "== OBJECT TEST ==\n"; 68class foo 69{ 70 public $s1; 71 public $s2; 72 public $s3; 73 74 function __construct() 75 { 76 global $sjis, $jis, $euc_jp; 77 78 $this->s1 = $euc_jp; 79 $this->s2 = $euc_jp; 80 $this->s3 = $euc_jp; 81 } 82} 83 84class bar 85{ 86 public $s1; 87 public $s2; 88 public $s3; 89 90 function __construct() 91 { 92 global $sjis, $jis, $euc_jp; 93 94 $this->s1 = $euc_jp; 95 $this->s2 = $euc_jp; 96 $this->s3 = $euc_jp; 97 } 98} 99 100$o = new foo; 101$oo = $o; 102$encoding = mb_convert_variables('EUC-JP', 'auto', $oo); 103print("$encoding\n"); // EUC-JP 104echo bin2hex("{$oo->s1}{$oo->s2}{$oo->s3}"), "\n"; // Converted to EUC-JP 105 106$o = new bar; 107$oo = $o; 108$encoding = mb_convert_variables('EUC-JP', 'auto', $oo); 109print("$encoding\n"); // EUC-JP 110echo bin2hex("{$oo->s1}{$oo->s2}{$oo->s3}"), "\n"; // Converted to EUC-JP 111 112// Test for scalar, array and object 113echo "== SCALAR, ARRAY AND OBJECT TEST ==\n"; 114 115$s1 = $euc_jp; 116$s2 = $euc_jp; 117$s3 = $euc_jp; 118$a = array($s1, $s2, $s3); 119$aa = $a; 120$oo = $o; 121 122$encoding = mb_convert_variables('EUC-JP', 'auto', $s1, $s2, $s3, $aa, $oo); 123print("$encoding\n"); // EUC-JP 124echo bin2hex("$s1$s2$s3"), "\n"; // Converted to EUC-JP 125echo bin2hex("{$aa[0]}{$aa[1]}{$aa[2]}"), "\n"; // Converted to EUC-JP 126echo bin2hex("{$oo->s1}{$oo->s2}{$oo->s3}"), "\n"; // Converted to EUC-JP 127 128echo "== DEEPLY NESTED OBJECT/ARRAY TEST ==\n"; 129 130class Nested 131{ 132 public $inner; 133 134 function __construct($value) 135 { 136 $this->inner = $value; 137 } 138} 139 140$deeplyNested = array(new Nested(array(new Nested(array(new Nested("BLAH")))))); 141 142$encoding = mb_convert_variables('UTF-16LE', 'UTF-8', $deeplyNested); 143echo $encoding, "\n"; 144echo bin2hex($deeplyNested[0]->inner[0]->inner[0]->inner), "\n"; 145 146echo "== INVALID STRING ENCODING TEST ==\n"; 147// Make sure both that the correct invalid encoding marker is used, 148// and that the count of illegal characters is incremented 149 150$illegalCount = mb_get_info('illegal_chars'); 151$nested = array(new Nested("\xFF")); 152mb_substitute_character(0x25); 153mb_convert_variables('UTF-16LE', 'UTF-8', $nested); 154echo bin2hex($nested[0]->inner), "\n"; 155echo "# of illegal characters detected: ", mb_get_info('illegal_chars') - $illegalCount, "\n"; 156 157$illegalCount = mb_get_info('illegal_chars'); 158$nested = array(new Nested("\xFF")); 159mb_substitute_character(0x26); 160mb_convert_variables('UTF-16LE', 'UTF-8', $nested); 161echo bin2hex($nested[0]->inner), "\n"; 162echo "# of illegal characters detected: ", mb_get_info('illegal_chars') - $illegalCount, "\n"; 163 164echo "== ENCODING AUTO-DETECTION TEST ==\n"; 165 166ini_set('mbstring.strict_detection', '1'); 167$bad_utf7 = "abc + abc"; 168var_dump(mb_convert_variables('UTF-8', 'UTF-7,UTF-8', $bad_utf7)); 169var_dump($bad_utf7); 170 171$bad_utf7imap = "abc &"; 172var_dump(mb_convert_variables('UTF-8', 'UTF7-IMAP,UTF-8', $bad_utf7imap)); 173var_dump($bad_utf7imap); 174 175?> 176--EXPECT-- 177== SCALAR TEST == 178SJIS 179c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 180JIS 181c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 182EUC-JP 183k/qWe4zqg2WDTINYg2eCxYK3gUIwMTIzNIJUglWCVoJXgliBQg== 184EUC-JP 185GyRCRnxLXDhsJUYlLSU5JUgkRyQ5ISMbKEIwMTIzNBskQiM1IzYjNyM4IzkhIxsoQg== 186EUC-JP 187c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 188== ARRAY TEST == 189EUC-JP 190c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 191EUC-JP 192c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 193== OBJECT TEST == 194EUC-JP 195c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 196EUC-JP 197c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 198== SCALAR, ARRAY AND OBJECT TEST == 199EUC-JP 200c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 201c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 202c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3c6fccbdcb8eca5c6a5ada5b9a5c8a4c7a4b9a1a33031323334a3b5a3b6a3b7a3b8a3b9a1a3 203== DEEPLY NESTED OBJECT/ARRAY TEST == 204UTF-8 20542004c0041004800 206== INVALID STRING ENCODING TEST == 2072500 208# of illegal characters detected: 1 2092600 210# of illegal characters detected: 1 211== ENCODING AUTO-DETECTION TEST == 212string(5) "UTF-8" 213string(9) "abc + abc" 214string(5) "UTF-8" 215string(5) "abc &" 216
