Utf8CaseGenerate.php

Go to the documentation of this file.

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

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

$in = fopen("UnicodeData.txt", "rt" );
if( !$in ) {
        print "Can't open UnicodeData.txt for reading.\n";
        print "If necessary, fetch this file from the internet:\n";
        print "http://www.unicode.org/Public/UNIDATA/UnicodeData.txt\n";
        exit(-1);
}
$wikiUpperChars = array();
$wikiLowerChars = array();

print "Reading character definitions...\n";
while( false !== ($line = fgets( $in ) ) ) {
        $columns = explode(';', $line);
        $codepoint = $columns[0];
        $name = $columns[1];
        $simpleUpper = $columns[12];
        $simpleLower = $columns[13];
        
        $source = codepointToUtf8( hexdec( $codepoint ) );
        if( $simpleUpper ) {
                $wikiUpperChars[$source] = codepointToUtf8( hexdec( $simpleUpper ) );
        }
        if( $simpleLower ) {
                $wikiLowerChars[$source] = codepointToUtf8( hexdec( $simpleLower ) );
        }
}
fclose( $in );

$out = fopen("Utf8Case.php", "wt");
if( $out ) {
        $outUpperChars = escapeArray( $wikiUpperChars );
        $outLowerChars = escapeArray( $wikiLowerChars );
        $outdata = "<" . "?php
\$wikiUpperChars = $outUpperChars;

\$wikiLowerChars = $outLowerChars;\n";
        fputs( $out, $outdata );
        fclose( $out );
        print "Wrote out Utf8Case.php\n";
} else {
        print "Can't create file Utf8Case.php\n";
        exit(-1);
}


function escapeArray( $arr ) {
        return "array(\n" .
                implode( ",\n",
                        array_map( "escapeLine",
                                array_keys( $arr ),
                                array_values( $arr ) ) ) .
                "\n)";
}

function escapeLine( $key, $val ) {
        $encKey = escapeSingleString( $key );
        $encVal = escapeSingleString( $val );
        return "\t'$encKey' => '$encVal'";
}