1--TEST-- 2normalize() 3--SKIPIF-- 4<?php if( !extension_loaded( 'intl' ) ) print 'skip'; ?> 5--FILE-- 6<?php 7 8/* 9 * Try normalization and test normalization 10 * with Procedural and Object methods. 11 */ 12 13function ut_main() 14{ 15 $res_str = ''; 16 17 $forms = array( 18 Normalizer::FORM_C, 19 Normalizer::FORM_D, 20 Normalizer::FORM_KC, 21 Normalizer::FORM_KD, 22 Normalizer::NONE, 23 ); 24 25 $forms_str = array ( 26 Normalizer::FORM_C => 'UNORM_FORM_C', 27 Normalizer::FORM_D => 'UNORM_FORM_D', 28 Normalizer::FORM_KC => 'UNORM_FORM_KC', 29 Normalizer::FORM_KD => 'UNORM_FORM_KD', 30 Normalizer::NONE => 'UNORM_NONE', 31 ); 32 33 /* just make sure all the form constants are defined as in the api spec */ 34 if ( Normalizer::FORM_C != Normalizer::NFC || 35 Normalizer::FORM_D != Normalizer::NFD || 36 Normalizer::FORM_KC != Normalizer::NFKC || 37 Normalizer::FORM_KD != Normalizer::NFKD || 38 Normalizer::NONE == Normalizer::FORM_C ) { 39 40 $res_str .= "Invalid normalization form declarations!\n"; 41 } 42 43 $char_a_diaeresis = "\xC3\xA4"; // 'LATIN SMALL LETTER A WITH DIAERESIS' (U+00E4) 44 $char_a_ring = "\xC3\xA5"; // 'LATIN SMALL LETTER A WITH RING ABOVE' (U+00E5) 45 $char_o_diaeresis = "\xC3\xB6"; // 'LATIN SMALL LETTER O WITH DIAERESIS' (U+00F6) 46 47 $char_angstrom_sign = "\xE2\x84\xAB"; // 'ANGSTROM SIGN' (U+212B) 48 $char_A_ring = "\xC3\x85"; // 'LATIN CAPITAL LETTER A WITH RING ABOVE' (U+00C5) 49 50 $char_ohm_sign = "\xE2\x84\xA6"; // 'OHM SIGN' (U+2126) 51 $char_omega = "\xCE\xA9"; // 'GREEK CAPITAL LETTER OMEGA' (U+03A9) 52 53 $char_combining_ring_above = "\xCC\x8A"; // 'COMBINING RING ABOVE' (U+030A) 54 55 $char_fi_ligature = "\xEF\xAC\x81"; // 'LATIN SMALL LIGATURE FI' (U+FB01) 56 57 $char_long_s_dot = "\xE1\xBA\x9B"; // 'LATIN SMALL LETTER LONG S WITH DOT ABOVE' (U+1E9B) 58 59 $strs = array( 60 'ABC', 61 $char_a_diaeresis . '||' . $char_a_ring . '||' . $char_o_diaeresis, 62 $char_angstrom_sign . '||' . $char_A_ring . '||' . 'A' . $char_combining_ring_above, 63 $char_ohm_sign . '||' . $char_omega, 64 $char_fi_ligature, 65 $char_long_s_dot, 66 ); 67 68 foreach( $forms as $form ) 69 { 70 foreach( $strs as $str ) 71 { 72 $str_norm = ut_norm_normalize( $str, $form ); 73 $error_code = intl_get_error_code(); 74 $error_message = intl_get_error_message(); 75 76 $str_hex = urlencode($str); 77 $str_norm_hex = urlencode($str_norm); 78 $res_str .= "'$str_hex' normalized to form '{$forms_str[$form]}' is '$str_norm_hex'" 79 . "\terror info: '$error_message' ($error_code)\n" 80 . ""; 81 82 $is_norm = ut_norm_is_normalized( $str, $form ); 83 $error_code = intl_get_error_code(); 84 $error_message = intl_get_error_message(); 85 86 $res_str .= " is in form '{$forms_str[$form]}'? = " . ($is_norm ? "yes" : "no") 87 . "\terror info: '$error_message' ($error_code)\n" 88 . ""; 89 } 90 } 91 92 return $res_str; 93} 94 95include_once( 'ut_common.inc' ); 96ut_run(); 97 98?> 99--EXPECT-- 100'ABC' normalized to form 'UNORM_FORM_C' is 'ABC' error info: 'U_ZERO_ERROR' (0) 101 is in form 'UNORM_FORM_C'? = yes error info: 'U_ZERO_ERROR' (0) 102'%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' normalized to form 'UNORM_FORM_C' is '%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' error info: 'U_ZERO_ERROR' (0) 103 is in form 'UNORM_FORM_C'? = yes error info: 'U_ZERO_ERROR' (0) 104'%E2%84%AB%7C%7C%C3%85%7C%7CA%CC%8A' normalized to form 'UNORM_FORM_C' is '%C3%85%7C%7C%C3%85%7C%7C%C3%85' error info: 'U_ZERO_ERROR' (0) 105 is in form 'UNORM_FORM_C'? = no error info: 'U_ZERO_ERROR' (0) 106'%E2%84%A6%7C%7C%CE%A9' normalized to form 'UNORM_FORM_C' is '%CE%A9%7C%7C%CE%A9' error info: 'U_ZERO_ERROR' (0) 107 is in form 'UNORM_FORM_C'? = no error info: 'U_ZERO_ERROR' (0) 108'%EF%AC%81' normalized to form 'UNORM_FORM_C' is '%EF%AC%81' error info: 'U_ZERO_ERROR' (0) 109 is in form 'UNORM_FORM_C'? = yes error info: 'U_ZERO_ERROR' (0) 110'%E1%BA%9B' normalized to form 'UNORM_FORM_C' is '%E1%BA%9B' error info: 'U_ZERO_ERROR' (0) 111 is in form 'UNORM_FORM_C'? = yes error info: 'U_ZERO_ERROR' (0) 112'ABC' normalized to form 'UNORM_FORM_D' is 'ABC' error info: 'U_ZERO_ERROR' (0) 113 is in form 'UNORM_FORM_D'? = yes error info: 'U_ZERO_ERROR' (0) 114'%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' normalized to form 'UNORM_FORM_D' is 'a%CC%88%7C%7Ca%CC%8A%7C%7Co%CC%88' error info: 'U_ZERO_ERROR' (0) 115 is in form 'UNORM_FORM_D'? = no error info: 'U_ZERO_ERROR' (0) 116'%E2%84%AB%7C%7C%C3%85%7C%7CA%CC%8A' normalized to form 'UNORM_FORM_D' is 'A%CC%8A%7C%7CA%CC%8A%7C%7CA%CC%8A' error info: 'U_ZERO_ERROR' (0) 117 is in form 'UNORM_FORM_D'? = no error info: 'U_ZERO_ERROR' (0) 118'%E2%84%A6%7C%7C%CE%A9' normalized to form 'UNORM_FORM_D' is '%CE%A9%7C%7C%CE%A9' error info: 'U_ZERO_ERROR' (0) 119 is in form 'UNORM_FORM_D'? = no error info: 'U_ZERO_ERROR' (0) 120'%EF%AC%81' normalized to form 'UNORM_FORM_D' is '%EF%AC%81' error info: 'U_ZERO_ERROR' (0) 121 is in form 'UNORM_FORM_D'? = yes error info: 'U_ZERO_ERROR' (0) 122'%E1%BA%9B' normalized to form 'UNORM_FORM_D' is '%C5%BF%CC%87' error info: 'U_ZERO_ERROR' (0) 123 is in form 'UNORM_FORM_D'? = no error info: 'U_ZERO_ERROR' (0) 124'ABC' normalized to form 'UNORM_FORM_KC' is 'ABC' error info: 'U_ZERO_ERROR' (0) 125 is in form 'UNORM_FORM_KC'? = yes error info: 'U_ZERO_ERROR' (0) 126'%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' normalized to form 'UNORM_FORM_KC' is '%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' error info: 'U_ZERO_ERROR' (0) 127 is in form 'UNORM_FORM_KC'? = yes error info: 'U_ZERO_ERROR' (0) 128'%E2%84%AB%7C%7C%C3%85%7C%7CA%CC%8A' normalized to form 'UNORM_FORM_KC' is '%C3%85%7C%7C%C3%85%7C%7C%C3%85' error info: 'U_ZERO_ERROR' (0) 129 is in form 'UNORM_FORM_KC'? = no error info: 'U_ZERO_ERROR' (0) 130'%E2%84%A6%7C%7C%CE%A9' normalized to form 'UNORM_FORM_KC' is '%CE%A9%7C%7C%CE%A9' error info: 'U_ZERO_ERROR' (0) 131 is in form 'UNORM_FORM_KC'? = no error info: 'U_ZERO_ERROR' (0) 132'%EF%AC%81' normalized to form 'UNORM_FORM_KC' is 'fi' error info: 'U_ZERO_ERROR' (0) 133 is in form 'UNORM_FORM_KC'? = no error info: 'U_ZERO_ERROR' (0) 134'%E1%BA%9B' normalized to form 'UNORM_FORM_KC' is '%E1%B9%A1' error info: 'U_ZERO_ERROR' (0) 135 is in form 'UNORM_FORM_KC'? = no error info: 'U_ZERO_ERROR' (0) 136'ABC' normalized to form 'UNORM_FORM_KD' is 'ABC' error info: 'U_ZERO_ERROR' (0) 137 is in form 'UNORM_FORM_KD'? = yes error info: 'U_ZERO_ERROR' (0) 138'%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' normalized to form 'UNORM_FORM_KD' is 'a%CC%88%7C%7Ca%CC%8A%7C%7Co%CC%88' error info: 'U_ZERO_ERROR' (0) 139 is in form 'UNORM_FORM_KD'? = no error info: 'U_ZERO_ERROR' (0) 140'%E2%84%AB%7C%7C%C3%85%7C%7CA%CC%8A' normalized to form 'UNORM_FORM_KD' is 'A%CC%8A%7C%7CA%CC%8A%7C%7CA%CC%8A' error info: 'U_ZERO_ERROR' (0) 141 is in form 'UNORM_FORM_KD'? = no error info: 'U_ZERO_ERROR' (0) 142'%E2%84%A6%7C%7C%CE%A9' normalized to form 'UNORM_FORM_KD' is '%CE%A9%7C%7C%CE%A9' error info: 'U_ZERO_ERROR' (0) 143 is in form 'UNORM_FORM_KD'? = no error info: 'U_ZERO_ERROR' (0) 144'%EF%AC%81' normalized to form 'UNORM_FORM_KD' is 'fi' error info: 'U_ZERO_ERROR' (0) 145 is in form 'UNORM_FORM_KD'? = no error info: 'U_ZERO_ERROR' (0) 146'%E1%BA%9B' normalized to form 'UNORM_FORM_KD' is 's%CC%87' error info: 'U_ZERO_ERROR' (0) 147 is in form 'UNORM_FORM_KD'? = no error info: 'U_ZERO_ERROR' (0) 148'ABC' normalized to form 'UNORM_NONE' is 'ABC' error info: 'U_ZERO_ERROR' (0) 149 is in form 'UNORM_NONE'? = no error info: 'normalizer_normalize: illegal normalization form: U_ILLEGAL_ARGUMENT_ERROR' (1) 150'%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' normalized to form 'UNORM_NONE' is '%C3%A4%7C%7C%C3%A5%7C%7C%C3%B6' error info: 'U_ZERO_ERROR' (0) 151 is in form 'UNORM_NONE'? = no error info: 'normalizer_normalize: illegal normalization form: U_ILLEGAL_ARGUMENT_ERROR' (1) 152'%E2%84%AB%7C%7C%C3%85%7C%7CA%CC%8A' normalized to form 'UNORM_NONE' is '%E2%84%AB%7C%7C%C3%85%7C%7CA%CC%8A' error info: 'U_ZERO_ERROR' (0) 153 is in form 'UNORM_NONE'? = no error info: 'normalizer_normalize: illegal normalization form: U_ILLEGAL_ARGUMENT_ERROR' (1) 154'%E2%84%A6%7C%7C%CE%A9' normalized to form 'UNORM_NONE' is '%E2%84%A6%7C%7C%CE%A9' error info: 'U_ZERO_ERROR' (0) 155 is in form 'UNORM_NONE'? = no error info: 'normalizer_normalize: illegal normalization form: U_ILLEGAL_ARGUMENT_ERROR' (1) 156'%EF%AC%81' normalized to form 'UNORM_NONE' is '%EF%AC%81' error info: 'U_ZERO_ERROR' (0) 157 is in form 'UNORM_NONE'? = no error info: 'normalizer_normalize: illegal normalization form: U_ILLEGAL_ARGUMENT_ERROR' (1) 158'%E1%BA%9B' normalized to form 'UNORM_NONE' is '%E1%BA%9B' error info: 'U_ZERO_ERROR' (0) 159 is in form 'UNORM_NONE'? = no error info: 'normalizer_normalize: illegal normalization form: U_ILLEGAL_ARGUMENT_ERROR' (1) 160
