includes/normal/UtfNormalTest2.php

   1 #!/usr/bin/php
   2 <?php
   3 /**
   4  * Other tests for the unicode normalization module
   5  *
   6  * @file
   7  * @ingroup UtfNormal
   8  */
   9
  10 if( php_sapi_name() != 'cli' ) {
  11         die( "Run me from the command line please.\n" );
  12 }
  13
  14 // From http://unicode.org/Public/UNIDATA/NormalizationTest.txt
  15 $file = "NormalizationTest.txt";
  16
  17 // Anything after this character is a comment
  18 define ( 'COMMENT', '#' );
  19
  20 // Semicolons are used to separate the columns
  21 define ( 'SEPARATOR', ';' );
  22
  23 $f = fopen($file, "r");
  24
  25 /**
  26  * The following section will be used for testing different normalization methods.
  27  * - Pure PHP
  28      ~ no assertion errors
  29      ~ 6.25 minutes
  30
  31  * - php_utfnormal.so or intl extension: both are wrappers around
  32      libicu so we list the version of libicu when making the
  33      comparison
  34
  35  * - libicu Ubuntu 3.8.1-3ubuntu1.1 php 5.2.6-3ubuntu4.5
  36      ~ 2200 assertion errors
  37      ~ 5 seconds
  38          ~ output: http://paste2.org/p/921566
  39
  40  * - libicu Ubuntu 4.2.1-3 php 5.3.2-1ubuntu4.2
  41      ~ 1384 assertion errors
  42          ~ 15 seconds
  43          ~ output: http://paste2.org/p/921435
  44
  45  * - libicu Debian 4.4.1-5 php 5.3.2-1ubuntu4.2
  46      ~ no assertion errors
  47          ~ 13 seconds
  48
  49  * - Tests comparing pure PHP output with libicu output were added
  50      later and slow down the runtime.
  51  */
  52
  53 require_once("./UtfNormal.php");
  54 function normalize_form_c($c)      { return UtfNormal::toNFC($c);  }
  55 function normalize_form_d($c)      { return UtfNormal::toNFD($c);  }
  56 function normalize_form_kc($c)     { return UtfNormal::toNFKC($c); }
  57 function normalize_form_kd($c)     { return UtfNormal::toNFKD($c); }
  58
  59 /**
  60  * This set of functions is only useful if youve added a param to the
  61  * following functions to force pure PHP usage.  I decided not to
  62  * commit that code since might produce a slowdown in the UTF
  63  * normalization code just for the sake of these tests. -- hexmode
  64  */
  65 function normalize_form_c_php($c)  { return UtfNormal::toNFC($c, "php");  }
  66 function normalize_form_d_php($c)  { return UtfNormal::toNFD($c, "php");  }
  67 function normalize_form_kc_php($c) { return UtfNormal::toNFKC($c, "php"); }
  68 function normalize_form_kd_php($c) { return UtfNormal::toNFKD($c, "php"); }
  69
  70 assert_options(ASSERT_ACTIVE, 1);
  71 assert_options(ASSERT_WARNING, 0);
  72 assert_options(ASSERT_QUIET_EVAL, 1);
  73 assert_options(ASSERT_CALLBACK, 'my_assert');
  74
  75 function my_assert( $file, $line, $code ) {
  76         global $col, $lineNo;
  77         echo "Assertion that '$code' failed on line $lineNo ($col[5])\n";
  78 }
  79
  80 $count = 0;
  81 $lineNo = 0;
  82 if( $f !== false ) {
  83         while( ( $col = getRow( $f ) ) !== false ) {
  84                 $lineNo++;
  85
  86                 if(count($col) == 6) {
  87                         $count++;
  88                         if( $count % 100 === 0 ) echo "Count: $count\n";
  89                 } else {
  90                         continue;
  91                 }
  92
  93                 # verify that the pure PHP version is correct
  94                 $NFCc1  = normalize_form_c($col[0]);
  95                 $NFCc1p = normalize_form_c_php($col[0]);
  96                 assert('$NFCc1 === $NFCc1p');
  97                 $NFCc2  = normalize_form_c($col[1]);
  98                 $NFCc2p = normalize_form_c_php($col[1]);
  99                 assert('$NFCc2 === $NFCc2p');
 100                 $NFCc3  = normalize_form_c($col[2]);
 101                 $NFCc3p = normalize_form_c_php($col[2]);
 102                 assert('$NFCc3 === $NFCc3p');
 103                 $NFCc4  = normalize_form_c($col[3]);
 104                 $NFCc4p = normalize_form_c_php($col[3]);
 105                 assert('$NFCc4 === $NFCc4p');
 106                 $NFCc5  = normalize_form_c($col[4]);
 107                 $NFCc5p = normalize_form_c_php($col[4]);
 108                 assert('$NFCc5 === $NFCc5p');
 109
 110                 $NFDc1  = normalize_form_d($col[0]);
 111                 $NFDc1p = normalize_form_d_php($col[0]);
 112                 assert('$NFDc1 === $NFDc1p');
 113                 $NFDc2  = normalize_form_d($col[1]);
 114                 $NFDc2p = normalize_form_d_php($col[1]);
 115                 assert('$NFDc2 === $NFDc2p');
 116                 $NFDc3  = normalize_form_d($col[2]);
 117                 $NFDc3p = normalize_form_d_php($col[2]);
 118                 assert('$NFDc3 === $NFDc3p');
 119                 $NFDc4  = normalize_form_d($col[3]);
 120                 $NFDc4p = normalize_form_d_php($col[3]);
 121                 assert('$NFDc4 === $NFDc4p');
 122                 $NFDc5  = normalize_form_d($col[4]);
 123                 $NFDc5p = normalize_form_d_php($col[4]);
 124                 assert('$NFDc5 === $NFDc5p');
 125
 126                 $NFKDc1  = normalize_form_kd($col[0]);
 127                 $NFKDc1p = normalize_form_kd_php($col[0]);
 128                 assert('$NFKDc1 === $NFKDc1p');
 129                 $NFKDc2  = normalize_form_kd($col[1]);
 130                 $NFKDc2p = normalize_form_kd_php($col[1]);
 131                 assert('$NFKDc2 === $NFKDc2p');
 132                 $NFKDc3  = normalize_form_kd($col[2]);
 133                 $NFKDc3p = normalize_form_kd_php($col[2]);
 134                 assert('$NFKDc3 === $NFKDc3p');
 135                 $NFKDc4  = normalize_form_kd($col[3]);
 136                 $NFKDc4p = normalize_form_kd_php($col[3]);
 137                 assert('$NFKDc4 === $NFKDc4p');
 138                 $NFKDc5  = normalize_form_kd($col[4]);
 139                 $NFKDc5p = normalize_form_kd_php($col[4]);
 140                 assert('$NFKDc5 === $NFKDc5p');
 141
 142                 $NFKCc1  = normalize_form_kc($col[0]);
 143                 $NFKCc1p = normalize_form_kc_php($col[0]);
 144                 assert('$NFKCc1 === $NFKCc1p');
 145                 $NFKCc2  = normalize_form_kc($col[1]);
 146                 $NFKCc2p = normalize_form_kc_php($col[1]);
 147                 assert('$NFKCc2 === $NFKCc2p');
 148                 $NFKCc3  = normalize_form_kc($col[2]);
 149                 $NFKCc3p = normalize_form_kc_php($col[2]);
 150                 assert('$NFKCc3 === $NFKCc3p');
 151                 $NFKCc4  = normalize_form_kc($col[3]);
 152                 $NFKCc4p = normalize_form_kc_php($col[3]);
 153                 assert('$NFKCc4 === $NFKCc4p');
 154                 $NFKCc5  = normalize_form_kc($col[4]);
 155                 $NFKCc5p = normalize_form_kc_php($col[4]);
 156                 assert('$NFKCc5 === $NFKCc5p');
 157
 158                 # c2 ==  NFC(c1) ==      NFC(c2) ==      NFC(c3)
 159                 assert('$col[1] === $NFCc1');
 160                 assert('$col[1] === $NFCc2');
 161                 assert('$col[1] === $NFCc3');
 162
 163                 # c4 ==  NFC(c4) ==      NFC(c5)
 164                 assert('$col[3] === $NFCc4');
 165                 assert('$col[3] === $NFCc5');
 166
 167                 # c3 ==  NFD(c1) ==      NFD(c2) ==      NFD(c3)
 168                 assert('$col[2] === $NFDc1');
 169                 assert('$col[2] === $NFDc2');
 170                 assert('$col[2] === $NFDc3');
 171
 172                 # c5 ==  NFD(c4) ==      NFD(c5)
 173                 assert('$col[4] === $NFDc4');
 174                 assert('$col[4] === $NFDc5');
 175
 176                 # c4 == NFKC(c1) == NFKC(c2) == NFKC(c3) == NFKC(c4) == NFKC(c5)
 177                 assert('$col[3] === $NFKCc1');
 178                 assert('$col[3] === $NFKCc2');
 179                 assert('$col[3] === $NFKCc3');
 180                 assert('$col[3] === $NFKCc4');
 181                 assert('$col[3] === $NFKCc5');
 182
 183                 # c5 == NFKD(c1) == NFKD(c2) == NFKD(c3) == NFKD(c4) == NFKD(c5)
 184                 assert('$col[4] === $NFKDc1');
 185                 assert('$col[4] === $NFKDc2');
 186                 assert('$col[4] === $NFKDc3');
 187                 assert('$col[4] === $NFKDc4');
 188                 assert('$col[4] === $NFKDc5');
 189         }
 190 }
 191 echo "done.\n";
 192
 193 // Compare against http://en.wikipedia.org/wiki/UTF-8#Description
 194 function unichr($c) {
 195         if ($c <= 0x7F) {
 196                 return chr($c);
 197         } elseif ($c <= 0x7FF) {
 198                 return chr(0xC0 | $c >> 6) . chr(0x80 | $c & 0x3F);
 199         } elseif ($c <= 0xFFFF) {
 200                 return chr(0xE0 | $c >> 12) . chr(0x80 | $c >> 6 & 0x3F)
 201                         . chr(0x80 | $c & 0x3F);
 202         } elseif ($c <= 0x10FFFF) {
 203                 return chr(0xF0 | $c >> 18) . chr(0x80 | $c >> 12 & 0x3F)
 204                         . chr(0x80 | $c >> 6 & 0x3F)
 205                         . chr(0x80 | $c & 0x3F);
 206         } else {
 207                 return false;
 208         }
 209 }
 210
 211 function unistr($c) {
 212         return implode("", array_map("unichr", array_map("hexdec", explode(" ", $c))));
 213 }
 214
 215 function getRow( $f ) {
 216         $row = fgets( $f );
 217         if( $row === false ) return false;
 218         $row = rtrim($row);
 219         $pos = strpos( $row, COMMENT );
 220         $pos2 = strpos( $row, ")" );
 221         if( $pos === 0 ) return array($row);
 222         $c = "";
 223
 224         if( $pos ) {
 225                 if($pos2) $c = substr( $row, $pos2 + 2 );
 226                 else      $c = substr( $row, $pos );
 227                 $row = substr( $row, 0, $pos );
 228         }
 229
 230         $ret = array();
 231         foreach( explode( SEPARATOR, $row ) as $ent ) {
 232                 if( trim( $ent ) !== "" ) {
 233                         $ret[] = unistr($ent);
 234                 }
 235         }
 236         $ret[] = $c;
 237
 238         return $ret;
 239 }