MediaWiki  REL1_22
UtfNormalBench.php
Go to the documentation of this file.
00001 <?php
00027 if( PHP_SAPI != 'cli' ) {
00028     die( "Run me from the command line please.\n" );
00029 }
00030 
00031 if( isset( $_SERVER['argv'] ) && in_array( '--icu', $_SERVER['argv'] ) ) {
00032     dl( 'php_utfnormal.so' );
00033 }
00034 
00035 require_once 'UtfNormalDefines.php';
00036 require_once 'UtfNormalUtil.php';
00037 require_once 'UtfNormal.php';
00038 
00039 define( 'BENCH_CYCLES', 5 );
00040 
00041 $testfiles = array(
00042     'testdata/washington.txt' => 'English text',
00043     'testdata/berlin.txt' => 'German text',
00044     'testdata/bulgakov.txt' => 'Russian text',
00045     'testdata/tokyo.txt' => 'Japanese text',
00046     'testdata/young.txt' => 'Korean text'
00047 );
00048 $normalizer = new UtfNormal;
00049 UtfNormal::loadData();
00050 foreach( $testfiles as $file => $desc ) {
00051     benchmarkTest( $normalizer, $file, $desc );
00052 }
00053 
00054 # -------
00055 
00056 function benchmarkTest( &$u, $filename, $desc ) {
00057     print "Testing $filename ($desc)...\n";
00058     $data = file_get_contents( $filename );
00059     $forms = array(
00060 #       'placebo',
00061         'cleanUp',
00062         'toNFC',
00063 #       'toNFKC',
00064 #       'toNFD', 'toNFKD',
00065         'NFC',
00066 #       'NFKC',
00067 #       'NFD', 'NFKD',
00068         array( 'fastDecompose', 'fastCombiningSort', 'fastCompose' ),
00069 #       'quickIsNFC', 'quickIsNFCVerify',
00070         );
00071     foreach( $forms as $form ) {
00072         if( is_array( $form ) ) {
00073             $str = $data;
00074             foreach( $form as $step ) {
00075                 $str = benchmarkForm( $u, $str, $step );
00076             }
00077         } else {
00078             benchmarkForm( $u, $data, $form );
00079         }
00080     }
00081 }
00082 
00083 function benchTime() {
00084     $st = explode( ' ', microtime() );
00085     return (float)$st[0] + (float)$st[1];
00086 }
00087 
00088 function benchmarkForm( &$u, &$data, $form ) {
00089     #$start = benchTime();
00090     for( $i = 0; $i < BENCH_CYCLES; $i++ ) {
00091         $start = benchTime();
00092         $out = $u->$form( $data, UtfNormal::$utfCanonicalDecomp );
00093         $deltas[] = (benchTime() - $start);
00094     }
00095     #$delta = (benchTime() - $start) / BENCH_CYCLES;
00096     sort( $deltas );
00097     $delta = $deltas[0]; # Take shortest time
00098 
00099     $rate = intval( strlen( $data ) / $delta );
00100     $same = (0 == strcmp( $data, $out ) );
00101 
00102     printf( " %20s %6.1fms %12s bytes/s (%s)\n",
00103         $form,
00104         $delta*1000.0,
00105         number_format( $rate ),
00106         ($same ? 'no change' : 'changed' ) );
00107     return $out;
00108 }