UtfNormalMemStress.php

Go to the documentation of this file.

<?php
if ( PHP_SAPI != 'cli' ) {
    die( "Run me from the command line please.\n" );
}

if ( isset( $_SERVER['argv'] ) && in_array( '--icu', $_SERVER['argv'] ) ) {
    dl( 'php_utfnormal.so' );
}

require_once 'UtfNormalDefines.php';
require_once 'UtfNormalUtil.php';
require_once 'UtfNormal.php';

define( 'BENCH_CYCLES', 1 );
define( 'BIGSIZE', 1024 * 1024 * 10 ); // 10m
ini_set( 'memory_limit', BIGSIZE + 120 * 1024 * 1024 );

$testfiles = array(
    'testdata/washington.txt' => 'English text',
    'testdata/berlin.txt' => 'German text',
    'testdata/bulgakov.txt' => 'Russian text',
    'testdata/tokyo.txt' => 'Japanese text',
    'testdata/young.txt' => 'Korean text'
);
$normalizer = new UtfNormal;
UtfNormal::loadData();
foreach ( $testfiles as $file => $desc ) {
    benchmarkTest( $normalizer, $file, $desc );
}

# -------

function benchmarkTest( &$u, $filename, $desc ) {
    print "Testing $filename ($desc)...\n";
    $data = file_get_contents( $filename );
    $all = $data;
    while ( strlen( $all ) < BIGSIZE ) {
        $all .= $all;
    }
    $data = $all;
    echo "Data is " . strlen( $data ) . " bytes.\n";
    $forms = array(
        'quickIsNFCVerify',
        'cleanUp',
    );

    foreach ( $forms as $form ) {
        if ( is_array( $form ) ) {
            $str = $data;
            foreach ( $form as $step ) {
                $str = benchmarkForm( $u, $str, $step );
            }
        } else {
            benchmarkForm( $u, $data, $form );
        }
    }
}

function benchmarkForm( &$u, &$data, $form ) {
    #$start = microtime( true );
    for ( $i = 0; $i < BENCH_CYCLES; $i++ ) {
        $start = microtime( true );
        $out = $u->$form( $data, UtfNormal::$utfCanonicalDecomp );
        $deltas[] = ( microtime( true ) - $start );
    }
    #$delta = (microtime( true ) - $start) / BENCH_CYCLES;
    sort( $deltas );
    $delta = $deltas[0]; # Take shortest time

    $rate = intval( strlen( $data ) / $delta );
    $same = ( 0 == strcmp( $data, $out ) );

    printf( " %20s %6.1fms %12s bytes/s (%s)\n",
        $form,
        $delta * 1000.0,
        number_format( $rate ),
        ( $same ? 'no change' : 'changed' ) );

    return $out;
}