MediaWiki
REL1_22
|
00001 <?php 00028 if( PHP_SAPI != 'cli' ) { 00029 die( "Run me from the command line please.\n" ); 00030 } 00031 00032 $verbose = true; 00033 #define( 'PRETTY_UTF8', true ); 00034 00035 if( defined( 'PRETTY_UTF8' ) ) { 00036 function pretty( $string ) { 00037 return strtoupper( bin2hex( $string ) ); 00038 } 00039 } else { 00044 function pretty( $string ) { 00045 return strtoupper( utf8ToHexSequence( $string ) ); 00046 } 00047 } 00048 00049 if( isset( $_SERVER['argv'] ) && in_array( '--icu', $_SERVER['argv'] ) ) { 00050 dl( 'php_utfnormal.so' ); 00051 } 00052 00053 require_once 'UtfNormalDefines.php'; 00054 require_once 'UtfNormalUtil.php'; 00055 require_once 'UtfNormal.php'; 00056 00057 $in = fopen("NormalizationTest.txt", "rt"); 00058 if( !$in ) { 00059 print "Couldn't open NormalizationTest.txt -- can't run tests.\n"; 00060 print "If necessary, manually download this file. It can be obtained at\n"; 00061 print "http://www.unicode.org/Public/UNIDATA/NormalizationTest.txt"; 00062 exit(-1); 00063 } 00064 00065 $normalizer = new UtfNormal; 00066 00067 $total = 0; 00068 $success = 0; 00069 $failure = 0; 00070 $ok = true; 00071 $testedChars = array(); 00072 while( false !== ( $line = fgets( $in ) ) ) { 00073 list( $data, $comment ) = explode( '#', $line ); 00074 if( $data === '' ) continue; 00075 $matches = array(); 00076 if( preg_match( '/@Part([\d])/', $data, $matches ) ) { 00077 if( $matches[1] > 0 ) { 00078 $ok = reportResults( $total, $success, $failure ) && $ok; 00079 } 00080 print "Part {$matches[1]}: $comment"; 00081 continue; 00082 } 00083 00084 $columns = array_map( "hexSequenceToUtf8", explode( ";", $data ) ); 00085 array_unshift( $columns, '' ); 00086 00087 $testedChars[$columns[1]] = true; 00088 $total++; 00089 if( testNormals( $normalizer, $columns, $comment, $verbose ) ) { 00090 $success++; 00091 } else { 00092 $failure++; 00093 # print "FAILED: $comment"; 00094 } 00095 if( $total % 100 == 0 ) print "$total "; 00096 } 00097 fclose( $in ); 00098 00099 $ok = reportResults( $total, $success, $failure ) && $ok; 00100 00101 $in = fopen("UnicodeData.txt", "rt" ); 00102 if( !$in ) { 00103 print "Can't open UnicodeData.txt for reading.\n"; 00104 print "If necessary, fetch this file from the internet:\n"; 00105 print "http://www.unicode.org/Public/UNIDATA/UnicodeData.txt\n"; 00106 exit(-1); 00107 } 00108 print "Now testing invariants...\n"; 00109 while( false !== ($line = fgets( $in ) ) ) { 00110 $cols = explode( ';', $line ); 00111 $char = codepointToUtf8( hexdec( $cols[0] ) ); 00112 $desc = $cols[0] . ": " . $cols[1]; 00113 if( $char < "\x20" || $char >= UTF8_SURROGATE_FIRST && $char <= UTF8_SURROGATE_LAST ) { 00114 # Can't check NULL with the ICU plugin, as null bytes fail in C land. 00115 # Skip other control characters, as we strip them for XML safety. 00116 # Surrogates are illegal on their own or in UTF-8, ignore. 00117 continue; 00118 } 00119 if( empty( $testedChars[$char] ) ) { 00120 $total++; 00121 if( testInvariant( $normalizer, $char, $desc, $verbose ) ) { 00122 $success++; 00123 } else { 00124 $failure++; 00125 } 00126 if( $total % 100 == 0 ) print "$total "; 00127 } 00128 } 00129 fclose( $in ); 00130 00131 $ok = reportResults( $total, $success, $failure ) && $ok; 00132 00133 if( $ok ) { 00134 print "TEST SUCCEEDED!\n"; 00135 exit(0); 00136 } else { 00137 print "TEST FAILED!\n"; 00138 exit(-1); 00139 } 00140 00141 ## ------ 00142 00143 function reportResults( &$total, &$success, &$failure ) { 00144 $percSucc = intval( $success * 100 / $total ); 00145 $percFail = intval( $failure * 100 / $total ); 00146 print "\n"; 00147 print "$success tests successful ($percSucc%)\n"; 00148 print "$failure tests failed ($percFail%)\n\n"; 00149 $ok = ($success > 0 && $failure == 0); 00150 $total = 0; 00151 $success = 0; 00152 $failure = 0; 00153 return $ok; 00154 } 00155 00156 function testNormals( &$u, $c, $comment, $verbose, $reportFailure = false ) { 00157 $result = testNFC( $u, $c, $comment, $reportFailure ); 00158 $result = testNFD( $u, $c, $comment, $reportFailure ) && $result; 00159 $result = testNFKC( $u, $c, $comment, $reportFailure ) && $result; 00160 $result = testNFKD( $u, $c, $comment, $reportFailure ) && $result; 00161 $result = testCleanUp( $u, $c, $comment, $reportFailure ) && $result; 00162 00163 if( $verbose && !$result && !$reportFailure ) { 00164 print $comment; 00165 testNormals( $u, $c, $comment, $verbose, true ); 00166 } 00167 return $result; 00168 } 00169 00170 function verbosify( $a, $b, $col, $form, $verbose ) { 00171 #$result = ($a === $b); 00172 $result = (strcmp( $a, $b ) == 0); 00173 if( $verbose ) { 00174 $aa = pretty( $a ); 00175 $bb = pretty( $b ); 00176 $ok = $result ? "succeed" : " failed"; 00177 $eq = $result ? "==" : "!="; 00178 print " $ok $form c$col '$aa' $eq '$bb'\n"; 00179 } 00180 return $result; 00181 } 00182 00183 function testNFC( &$u, $c, $comment, $verbose ) { 00184 $result = verbosify( $c[2], $u->toNFC( $c[1] ), 1, 'NFC', $verbose ); 00185 $result = verbosify( $c[2], $u->toNFC( $c[2] ), 2, 'NFC', $verbose ) && $result; 00186 $result = verbosify( $c[2], $u->toNFC( $c[3] ), 3, 'NFC', $verbose ) && $result; 00187 $result = verbosify( $c[4], $u->toNFC( $c[4] ), 4, 'NFC', $verbose ) && $result; 00188 $result = verbosify( $c[4], $u->toNFC( $c[5] ), 5, 'NFC', $verbose ) && $result; 00189 return $result; 00190 } 00191 00192 function testCleanUp( &$u, $c, $comment, $verbose ) { 00193 $x = $c[1]; 00194 $result = verbosify( $c[2], $u->cleanUp( $x ), 1, 'cleanUp', $verbose ); 00195 $x = $c[2]; 00196 $result = verbosify( $c[2], $u->cleanUp( $x ), 2, 'cleanUp', $verbose ) && $result; 00197 $x = $c[3]; 00198 $result = verbosify( $c[2], $u->cleanUp( $x ), 3, 'cleanUp', $verbose ) && $result; 00199 $x = $c[4]; 00200 $result = verbosify( $c[4], $u->cleanUp( $x ), 4, 'cleanUp', $verbose ) && $result; 00201 $x = $c[5]; 00202 $result = verbosify( $c[4], $u->cleanUp( $x ), 5, 'cleanUp', $verbose ) && $result; 00203 return $result; 00204 } 00205 00206 function testNFD( &$u, $c, $comment, $verbose ) { 00207 $result = verbosify( $c[3], $u->toNFD( $c[1] ), 1, 'NFD', $verbose ); 00208 $result = verbosify( $c[3], $u->toNFD( $c[2] ), 2, 'NFD', $verbose ) && $result; 00209 $result = verbosify( $c[3], $u->toNFD( $c[3] ), 3, 'NFD', $verbose ) && $result; 00210 $result = verbosify( $c[5], $u->toNFD( $c[4] ), 4, 'NFD', $verbose ) && $result; 00211 $result = verbosify( $c[5], $u->toNFD( $c[5] ), 5, 'NFD', $verbose ) && $result; 00212 return $result; 00213 } 00214 00215 function testNFKC( &$u, $c, $comment, $verbose ) { 00216 $result = verbosify( $c[4], $u->toNFKC( $c[1] ), 1, 'NFKC', $verbose ); 00217 $result = verbosify( $c[4], $u->toNFKC( $c[2] ), 2, 'NFKC', $verbose ) && $result; 00218 $result = verbosify( $c[4], $u->toNFKC( $c[3] ), 3, 'NFKC', $verbose ) && $result; 00219 $result = verbosify( $c[4], $u->toNFKC( $c[4] ), 4, 'NFKC', $verbose ) && $result; 00220 $result = verbosify( $c[4], $u->toNFKC( $c[5] ), 5, 'NFKC', $verbose ) && $result; 00221 return $result; 00222 } 00223 00224 function testNFKD( &$u, $c, $comment, $verbose ) { 00225 $result = verbosify( $c[5], $u->toNFKD( $c[1] ), 1, 'NFKD', $verbose ); 00226 $result = verbosify( $c[5], $u->toNFKD( $c[2] ), 2, 'NFKD', $verbose ) && $result; 00227 $result = verbosify( $c[5], $u->toNFKD( $c[3] ), 3, 'NFKD', $verbose ) && $result; 00228 $result = verbosify( $c[5], $u->toNFKD( $c[4] ), 4, 'NFKD', $verbose ) && $result; 00229 $result = verbosify( $c[5], $u->toNFKD( $c[5] ), 5, 'NFKD', $verbose ) && $result; 00230 return $result; 00231 } 00232 00233 function testInvariant( &$u, $char, $desc, $verbose, $reportFailure = false ) { 00234 $result = verbosify( $char, $u->toNFC( $char ), 1, 'NFC', $reportFailure ); 00235 $result = verbosify( $char, $u->toNFD( $char ), 1, 'NFD', $reportFailure ) && $result; 00236 $result = verbosify( $char, $u->toNFKC( $char ), 1, 'NFKC', $reportFailure ) && $result; 00237 $result = verbosify( $char, $u->toNFKD( $char ), 1, 'NFKD', $reportFailure ) && $result; 00238 $result = verbosify( $char, $u->cleanUp( $char ), 1, 'cleanUp', $reportFailure ) && $result; 00239 00240 if( $verbose && !$result && !$reportFailure ) { 00241 print $desc; 00242 testInvariant( $u, $char, $desc, $verbose, true ); 00243 } 00244 return $result; 00245 }