Utf8Test.php

Go to the documentation of this file.

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

require_once 'UtfNormalDefines.php';
require_once 'UtfNormalUtil.php';
require_once 'UtfNormal.php';
mb_internal_encoding( "utf-8" );

$verbose = false;
#$verbose = true;

$in = fopen( "UTF-8-test.txt", "rt" );
if( !$in ) {
        print "Couldn't open UTF-8-test.txt -- can't run tests.\n";
        print "If necessary, manually download this file. It can be obtained at\n";
        print "http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt\n";
        exit(-1);
}

$columns = 0;
while( false !== ( $line = fgets( $in ) ) ) {
        $matches = array();
        if( preg_match( '/^(Here come the tests:\s*)\|$/', $line, $matches ) ) {
                $columns = strpos( $line, '|' );
                break;
        }
}

if( !$columns ) {
        print "Something seems to be wrong; couldn't extract line length.\n";
        print "Check that UTF-8-test.txt was downloaded correctly from\n";
        print "http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt\n";
        exit(-1);
}

# print "$columns\n";

$ignore = array(
        # These two lines actually seem to be corrupt
        '2.1.1', '2.2.1' );

$exceptions = array(
        # Tests that should mark invalid characters due to using long
        # sequences beyond what is now considered legal.
        '2.1.5', '2.1.6', '2.2.4', '2.2.5', '2.2.6', '2.3.5',

        # Literal 0xffff, which is illegal
        '2.2.3' );

$longTests = array(
        # These tests span multiple lines
        '3.1.9', '3.2.1', '3.2.2', '3.2.3', '3.2.4', '3.2.5',
        '3.4' );

# These tests are not in proper subsections
$sectionTests = array( '3.4' );

$section = null;
$test = '';
$failed = 0;
$success = 0;
$total = 0;
while( false !== ( $line = fgets( $in ) ) ) {
        $matches = array();
        if( preg_match( '/^(\d+)\s+(.*?)\s*\|/', $line, $matches ) ) {
                $section = $matches[1];
                print $line;
                continue;
        }
        if( preg_match( '/^(\d+\.\d+\.\d+)\s*/', $line, $matches ) ) {
                $test = $matches[1];

                if( in_array( $test, $ignore ) ) {
                        continue;
                }
                if( in_array( $test, $longTests ) ) {
                        $line = fgets( $in );
                        for( $line = fgets( $in ); !preg_match( '/^\s+\|/', $line ); $line = fgets( $in ) ) {
                                testLine( $test, $line, $total, $success, $failed, $columns, $exceptions, $verbose );
                        }
                } else {
                        testLine( $test, $line, $total, $success, $failed, $columns, $exceptions, $verbose );
                }
        }
}

if( $failed ) {
        echo "\nFailed $failed tests.\n";
        echo "UTF-8 DECODER TEST FAILED\n";
        exit (-1);
}

echo "UTF-8 DECODER TEST SUCCESS!\n";
exit (0);


function testLine( $test, $line, &$total, &$success, &$failed, $columns, $exceptions, $verbose ) {
        $stripped = $line;
        UtfNormal::quickisNFCVerify( $stripped );

        $same = ( $line == $stripped );
        $len = mb_strlen( substr( $stripped, 0, strpos( $stripped, '|' ) ) );
        if( $len == 0 ) {
                $len = strlen( substr( $stripped, 0, strpos( $stripped, '|' ) ) );
        }

        $ok = $same ^ ($test >= 3 );

        $ok ^= in_array( $test, $exceptions );

        $ok &= ($columns == $len);

        $total++;
        if( $ok ) {
                $success++;
        } else {
                $failed++;
        }

        if( $verbose || !$ok ) {
                print str_replace( "\n", "$len\n", $stripped );
        }
}