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