1--TEST-- 2Testing mb_convert_kana() function 3--EXTENSIONS-- 4mbstring 5--FILE-- 6<?php 7$zenKakuA = 'ァアィイゥウェエォオカガキギク'; 8$zenKakuB = 'グケゲコゴサザシジスズセゼソゾタ'; 9$zenKakuC = 'ダチヂッツヅテデトドナニヌネノハ'; 10$zenKakuD = 'バパヒビピフブプヘベペホボポマミ'; 11$zenKakuE = 'ムメモャヤュユョヨラリルレロヮワ'; 12$zenKakuF = 'ヰヱヲンヴヵヶヷヸヹヺ・ーヽヾ'; 13 14$hanKakuA = '｠｡｢｣､･ｦｧｨｩｪｫｬｭｮｯ'; 15$hanKakuB = 'ｰｱｲｳｴｵｶｷｸｹｺｻｼｽｾｿ'; 16$hanKakuC = 'ﾀﾁﾂﾃﾄﾅﾆﾇﾈﾉﾊﾋﾌﾍﾎﾏ'; 17$hanKakuD = 'ﾐﾑﾒﾓﾔﾕﾖﾗﾘﾙﾚﾛﾜﾝﾞﾟ'; 18 19// Convert all Zenkaku to Hankaku; no effect 20echo "'A': " . $zenKakuA . ' => ' . mb_convert_kana($zenKakuA, 'AK', 'utf-8') . "\n"; 21// Convert all Hankaku to Zenkaku; has an effect 22echo "'a': " . $zenKakuB . ' => ' . mb_convert_kana($zenKakuB, 'ak', 'utf-8') . "\n"; 23echo "'a': " . $zenKakuC . ' => ' . mb_convert_kana($zenKakuC, 'ak', 'utf-8') . "\n"; 24echo "'a': " . $zenKakuD . ' => ' . mb_convert_kana($zenKakuD, 'ak', 'utf-8') . "\n"; 25echo "'a': " . $zenKakuE . ' => ' . mb_convert_kana($zenKakuE, 'ak', 'utf-8') . "\n"; 26echo "'a': " . $zenKakuF . ' => ' . mb_convert_kana($zenKakuF, 'ak', 'utf-8') . "\n"; 27echo "\n"; 28// Convert all Zenkaku to Hankaku; has an effect 29echo "'A': " . $hanKakuA . ' => ' . mb_convert_kana($hanKakuA, 'AK', 'utf-8') . "\n"; 30echo "'A': " . $hanKakuB . ' => ' . mb_convert_kana($hanKakuB, 'AK', 'utf-8') . "\n"; 31echo "'A': " . $hanKakuC . ' => ' . mb_convert_kana($hanKakuC, 'AK', 'utf-8') . "\n"; 32echo "'A': " . $hanKakuD . ' => ' . mb_convert_kana($hanKakuD, 'AK', 'utf-8') . "\n\n"; 33 34echo "Convert ASCII letter to full-width: A -> ", bin2hex(mb_convert_kana("\x00A", 'A', 'UTF-16BE')), "\n"; 35echo "Convert ASCII letter to full-width: A -> ", bin2hex(mb_convert_kana("\x00A", 'R', 'UTF-16BE')), "\n"; 36echo "Convert ASCII numeral to full-width: 1 -> ", bin2hex(mb_convert_kana("\x001", 'N', 'UTF-16BE')), "\n"; 37echo "Convert ASCII space to full-width: ", bin2hex(mb_convert_kana("\x00 ", 'S', 'UTF-16BE')), "\n\n"; 38 39echo "Convert hankaku kana to zenkaku kana:\n"; 40echo "Using 'glue' mode:\n"; 41echo bin2hex(mb_convert_kana("\xFF\x76\xFF\x9E", 'KV', 'UTF-16BE')), "\n"; 42echo bin2hex(mb_convert_kana("\xFF\x73\xFF\x9E", 'KV', 'UTF-16BE')), "\n"; 43echo bin2hex(mb_convert_kana("\xFF\x8A\xFF\x9F", 'KV', 'UTF-16BE')), "\n"; 44echo bin2hex(mb_convert_kana("\xFF\x8A\x00A", 'KV', 'UTF-16BE')), "\n"; 45echo bin2hex(mb_convert_kana("\xFF\x76\xFF\x9E", 'HV', 'UTF-16BE')), "\n"; 46echo bin2hex(mb_convert_kana("\xFF\x8A\xFF\x9F", 'HV', 'UTF-16BE')), "\n"; 47echo bin2hex(mb_convert_kana("\xFF\x8A\x00A", 'HV', 'UTF-16BE')), "\n"; 48echo "Not using 'glue' mode:\n"; 49echo bin2hex(mb_convert_kana("\xFF\x70", 'K', 'UTF-16BE')), "\n"; 50echo bin2hex(mb_convert_kana("\xFF\x70", 'H', 'UTF-16BE')), "\n\n"; 51 52echo "Convert selected punctuation/symbols to full-width and back:\n"; 53echo bin2hex(mb_convert_kana("\x00\x5C\x00\xA5\x00\x7E\x20\x3E\x00\x27\x00\x22", 'M', 'UTF-16BE')), "\n"; 54echo bin2hex(mb_convert_kana("\xFF\xE5\xFF\x3C\xFF\xE3\x20\x3E\x20\x18\x20\x19\x20\x1C\x20\x1D", 'm', 'UTF-16BE')), "\n\n"; 55 56echo "Convert various full-width characters to half-width:\n"; 57echo bin2hex(mb_convert_kana("\xFF\x01", 'a', 'UTF-16BE')), "\n"; 58echo bin2hex(mb_convert_kana("\xFF\x21", 'r', 'UTF-16BE')), "\n"; 59echo bin2hex(mb_convert_kana("\xFF\x10", 'n', 'UTF-16BE')), "\n"; 60echo bin2hex(mb_convert_kana("\x30\x00", 's', 'UTF-16BE')), "\n"; 61echo bin2hex(mb_convert_kana("\x22\x12", 'a', 'UTF-16BE')), "\n\n"; 62 63echo "Convert full-width kana to half-width:\n"; 64echo bin2hex(mb_convert_kana("\x30\x41", 'h', 'UTF-16BE')), "\n"; 65echo bin2hex(mb_convert_kana("\x30\x01", 'h', 'UTF-16BE')), "\n"; 66echo bin2hex(mb_convert_kana("\x30\x02", 'h', 'UTF-16BE')), "\n"; 67echo bin2hex(mb_convert_kana("\x30\x0C", 'h', 'UTF-16BE')), "\n"; 68echo bin2hex(mb_convert_kana("\x30\x0D", 'h', 'UTF-16BE')), "\n"; 69echo bin2hex(mb_convert_kana("\x30\x9B", 'h', 'UTF-16BE')), "\n"; 70echo bin2hex(mb_convert_kana("\x30\x9C", 'h', 'UTF-16BE')), "\n"; 71echo bin2hex(mb_convert_kana("\x30\xFC", 'h', 'UTF-16BE')), "\n"; 72echo bin2hex(mb_convert_kana("\x30\xFB", 'h', 'UTF-16BE')), "\n"; 73echo bin2hex(mb_convert_kana("\x30\x01", 'k', 'UTF-16BE')), "\n"; 74echo bin2hex(mb_convert_kana("\x30\x02", 'k', 'UTF-16BE')), "\n"; 75echo bin2hex(mb_convert_kana("\x30\x0C", 'k', 'UTF-16BE')), "\n"; 76echo bin2hex(mb_convert_kana("\x30\x0D", 'k', 'UTF-16BE')), "\n"; 77echo bin2hex(mb_convert_kana("\x30\x9B", 'k', 'UTF-16BE')), "\n"; 78echo bin2hex(mb_convert_kana("\x30\x9C", 'k', 'UTF-16BE')), "\n"; 79echo bin2hex(mb_convert_kana("\x30\xFC", 'k', 'UTF-16BE')), "\n"; 80echo bin2hex(mb_convert_kana("\x30\xFB", 'k', 'UTF-16BE')), "\n"; 81echo bin2hex(mb_convert_kana("fooあいうエオ", "rnaskh", "UTF-8")), "\n"; 82echo "Including one which will expand to two codepoints:\n"; 83echo bin2hex(mb_convert_kana("\x30\x52", 'h', 'UTF-16BE')), "\n\n"; 84 85echo "Convert full-width hiragana to full-width katakana:\n"; 86echo bin2hex(mb_convert_kana("\x30\x41", 'C', 'UTF-16BE')), "\n"; 87echo bin2hex(mb_convert_kana("\x30\x9D", 'C', 'UTF-16BE')), "\n"; 88echo bin2hex(mb_convert_kana("\x30\x9E", 'C', 'UTF-16BE')), "\n\n"; 89 90echo "Convert full-width katakana to full-width hiragana:\n"; 91echo bin2hex(mb_convert_kana("\x30\xA1", 'c', 'UTF-16BE')), "\n"; 92echo bin2hex(mb_convert_kana("\x30\xFD", 'c', 'UTF-16BE')), "\n"; 93echo bin2hex(mb_convert_kana("\x30\xFE", 'c', 'UTF-16BE')), "\n\n"; 94 95echo bin2hex(mb_convert_kana("\x00\x00", 'A', 'UTF-16BE')), "\n"; 96echo "\n"; 97 98// Try combinations of flags which don't make sense 99function tryBadFlags($flags) { 100 try { 101 mb_convert_kana('abc', $flags, 'UTF-8'); 102 echo "BAD! mb_convert_kana should have thrown an exception for flags: $flags\n"; 103 } catch (ValueError $e) { 104 echo $e->getMessage(), "\n"; 105 } 106} 107function tryIncompatibleFlags($flag1, $flag2) { 108 tryBadFlags($flag1 . $flag2); 109 tryBadFlags($flag2 . $flag1); 110} 111 112tryIncompatibleFlags('A', 'a'); 113tryIncompatibleFlags('A', 'r'); 114tryIncompatibleFlags('A', 'n'); 115tryIncompatibleFlags('a', 'R'); 116tryIncompatibleFlags('a', 'N'); 117tryIncompatibleFlags('R', 'r'); 118tryIncompatibleFlags('N', 'n'); 119tryIncompatibleFlags('S', 's'); 120tryIncompatibleFlags('K', 'H'); 121tryIncompatibleFlags('C', 'c'); 122tryIncompatibleFlags('M', 'm'); 123tryIncompatibleFlags('h', 'C'); 124tryIncompatibleFlags('h', 'c'); 125tryIncompatibleFlags('k', 'C'); 126tryIncompatibleFlags('k', 'c'); 127 128// Try non-existent flag 129try { 130 mb_convert_kana($zenKakuA, 'Z', 'UTF-8'); 131} catch (ValueError $e) { 132 echo $e->getMessage() . "\n"; 133} 134 135// Regression test: Two codepoints collapsed into one, just one position 136// before the end of the string 137$converted = mb_convert_kana("\xb9\xde\xde", 'HV', 'JIS'); 138if ($converted !== "\x1b\$B\$2!+\x1b(B") 139 echo "Failed! Expected " . bin2hex("\x1b\$B\$2!+\x1b(B") . ", got: " . bin2hex($converted) . "\n"; 140 141// Regression test: the old implementation of mb_convert_kana would swallow 142// zero bytes in some cases 143if (mb_convert_kana("abc\x00abc", 'c', 'ASCII') !== "abc\x00abc") 144 echo "mb_convert_kana is swallowing zero bytes!\n"; 145 146?> 147--EXPECT-- 148'A': ァアィイゥウェエォオカガキギク => ァアィイゥウェエォオカガキギク 149'a': グケゲコゴサザシジスズセゼソゾタ => ｸﾞｹｹﾞｺｺﾞｻｻﾞｼｼﾞｽｽﾞｾｾﾞｿｿﾞﾀ 150'a': ダチヂッツヅテデトドナニヌネノハ => ﾀﾞﾁﾁﾞｯﾂﾂﾞﾃﾃﾞﾄﾄﾞﾅﾆﾇﾈﾉﾊ 151'a': バパヒビピフブプヘベペホボポマミ => ﾊﾞﾊﾟﾋﾋﾞﾋﾟﾌﾌﾞﾌﾟﾍﾍﾞﾍﾟﾎﾎﾞﾎﾟﾏﾐ 152'a': ムメモャヤュユョヨラリルレロヮワ => ﾑﾒﾓｬﾔｭﾕｮﾖﾗﾘﾙﾚﾛﾜﾜ 153'a': ヰヱヲンヴヵヶヷヸヹヺ・ーヽヾ => ｲｴｦﾝｳﾞヵヶヷヸヹヺ･ｰヽヾ 154 155'A': ｠｡｢｣､･ｦｧｨｩｪｫｬｭｮｯ => ｠。「」、・ヲァィゥェォャュョッ 156'A': ｰｱｲｳｴｵｶｷｸｹｺｻｼｽｾｿ => ーアイウエオカキクケコサシスセソ 157'A': ﾀﾁﾂﾃﾄﾅﾆﾇﾈﾉﾊﾋﾌﾍﾎﾏ => タチツテトナニヌネノハヒフヘホマ 158'A': ﾐﾑﾒﾓﾔﾕﾖﾗﾘﾙﾚﾛﾜﾝﾞﾟ => ミムメモヤユヨラリルレロワン゛゜ 159 160Convert ASCII letter to full-width: A -> ff21 161Convert ASCII letter to full-width: A -> ff21 162Convert ASCII numeral to full-width: 1 -> ff11 163Convert ASCII space to full-width: 3000 164 165Convert hankaku kana to zenkaku kana: 166Using 'glue' mode: 16730ac 16830f4 16930d1 17030cf0041 171304c 1723071 173306f0041 174Not using 'glue' mode: 17530fc 17630fc 177 178Convert selected punctuation/symbols to full-width and back: 179ffe5ffe5ffe3ffe32019201d 180005c005c007e007e0027002700220022 181 182Convert various full-width characters to half-width: 1830021 1840041 1850030 1860020 187002d 188 189Convert full-width kana to half-width: 190ff67 191ff64 192ff61 193ff62 194ff63 195ff9e 196ff9f 197ff70 198ff65 199ff64 200ff61 201ff62 202ff63 203ff9e 204ff9f 205ff70 206ff65 207666f6fefbdb1efbdb2efbdb3efbdb4efbdb5 208Including one which will expand to two codepoints: 209ff79ff9e 210 211Convert full-width hiragana to full-width katakana: 21230a1 21330fd 21430fe 215 216Convert full-width katakana to full-width hiragana: 2173041 218309d 219309e 220 2210000 222 223mb_convert_kana(): Argument #2 ($mode) must not combine 'A' and 'a' flags 224mb_convert_kana(): Argument #2 ($mode) must not combine 'A' and 'a' flags 225mb_convert_kana(): Argument #2 ($mode) must not combine 'A' and 'r' flags 226mb_convert_kana(): Argument #2 ($mode) must not combine 'A' and 'r' flags 227mb_convert_kana(): Argument #2 ($mode) must not combine 'A' and 'n' flags 228mb_convert_kana(): Argument #2 ($mode) must not combine 'A' and 'n' flags 229mb_convert_kana(): Argument #2 ($mode) must not combine 'R' and 'a' flags 230mb_convert_kana(): Argument #2 ($mode) must not combine 'R' and 'a' flags 231mb_convert_kana(): Argument #2 ($mode) must not combine 'N' and 'a' flags 232mb_convert_kana(): Argument #2 ($mode) must not combine 'N' and 'a' flags 233mb_convert_kana(): Argument #2 ($mode) must not combine 'R' and 'r' flags 234mb_convert_kana(): Argument #2 ($mode) must not combine 'R' and 'r' flags 235mb_convert_kana(): Argument #2 ($mode) must not combine 'N' and 'n' flags 236mb_convert_kana(): Argument #2 ($mode) must not combine 'N' and 'n' flags 237mb_convert_kana(): Argument #2 ($mode) must not combine 'S' and 's' flags 238mb_convert_kana(): Argument #2 ($mode) must not combine 'S' and 's' flags 239mb_convert_kana(): Argument #2 ($mode) must not combine 'H' and 'K' flags 240mb_convert_kana(): Argument #2 ($mode) must not combine 'H' and 'K' flags 241mb_convert_kana(): Argument #2 ($mode) must not combine 'C' and 'c' flags 242mb_convert_kana(): Argument #2 ($mode) must not combine 'C' and 'c' flags 243mb_convert_kana(): Argument #2 ($mode) must not combine 'M' and 'm' flags 244mb_convert_kana(): Argument #2 ($mode) must not combine 'M' and 'm' flags 245mb_convert_kana(): Argument #2 ($mode) must not combine 'h' and 'C' flags 246mb_convert_kana(): Argument #2 ($mode) must not combine 'h' and 'C' flags 247mb_convert_kana(): Argument #2 ($mode) must not combine 'h' and 'c' flags 248mb_convert_kana(): Argument #2 ($mode) must not combine 'h' and 'c' flags 249mb_convert_kana(): Argument #2 ($mode) must not combine 'k' and 'C' flags 250mb_convert_kana(): Argument #2 ($mode) must not combine 'k' and 'C' flags 251mb_convert_kana(): Argument #2 ($mode) must not combine 'k' and 'c' flags 252mb_convert_kana(): Argument #2 ($mode) must not combine 'k' and 'c' flags 253mb_convert_kana(): Argument #2 ($mode) contains invalid flag: 'Z' 254--CREDITS-- 255Jason Easter <easter@phpug-wuerzburg.de> 256PHPUG Würzburg <phpug-wuerzburg.de> 257Testfest 2009 2009-06-20 258
