Ticket #10165: mo.3.php

<?php


/**
 * Improved MO file reader for Wordpress.
 *
 * @author Thomas Urban <thomas.urban@toxa.de>
 */


include_once( dirname( __FILE__ ) . '/translations.php' );


class MO extends Translations
{

        var $nplurals = 2;

        var $pluralTerm = '';


        var $strOverloaded = false;



        function MO( $headers = array(), $hash = array() )
        {

                $this->headers = $headers;
                $this->entries = $hash;

                $this->strOverloaded = ( ini_get( 'mbstring.func_overload' ) & 2 ) &&
                                                                is_callable( 'mb_substr' );

                if ( $this->headers['Plural-Forms'] )
                {

                        $sep  = $this->_strpos( $this->headers['Plural-Forms'], ';' );
                        $temp = $this->_substr( $this->headers['Plural-Forms'], 0, $sep );

                        if ( preg_match( '/^nplurals\s*=\s*(\d+)$/i', trim( $temp ), $m ) )
                        {
                                $this->nplurals   = intval( $m[1] );
                                $this->pluralTerm = trim( $this->_substr( $temp, $sep + 1 ) );
                        }
                }
        }


        function _strlen( $string )
        {
                return $this->strOverloaded ? mb_strlen( $string, 'ascii' )
                                                                        : strlen( $string );
        }

        function _strpos( $haystack, $needle, $offset = null )
        {
                return $this->strOverloaded ? mb_strpos( $haystack, $needle, $offset, 'ascii' )
                                                                        : strpos( $haystack, $needle, $offset );
        }

        function _substr( $string, $offset, $length = null )
        {

                if ( $this->strOverloaded )
                        return mb_substr( $string, $offset, $length, 'ascii' );

                if ( is_null( $length ) )
                        return substr( $string, $offset );

                return substr( $string, $offset, $length );

        }

        function _str_split( $string, $chunkSize )
        {

                if ( $this->strOverloaded )
                {

                        $length = mb_strlen( $string, 'ascii' );

                        $out = array();

                        for ( $i = 0; $i < $length; $i += $chunkSize )
                                $out[] = mb_substr( $string, $i, $chunkSize, 'ascii' );

                        return $out;

                }
                else
                        return str_split( $string, $chunkSize );

        }


        /**
         * Reads provided MO file.
         *
         * @param string $filename name of MO file to read
         * @return MO
         */

        function import_from_file( $filename )
        {

                /*
                 * check cache for existing record first
                 */

                $set = wp_cache_get( $filename, 'l10n' );
                if ( is_array( $set ) )
                {

                        list( $this->headers, $this->entries ) = $set;

                        return true;

                }



                /**
                 * read header from file
                 */

                $file = fopen( $filename, 'r' );
                if ( !$file )
                        return false;

                $header = fread( $file, 28 );
                if ( $this->_strlen( $header ) != 28 )
                        return false;

                // detect endianess
                $endian = unpack( 'Nendian', $this->_substr( $header, 0, 4 ) );
                if ( $endian['endian'] == intval( hexdec( '950412de' ) ) )
                        $endian = 'N';
                else if ( $endian['endian'] == intval( hexdec( 'de120495' ) ) )
                        $endian = 'V';
                else
                        return false;

                // parse header
                $header = unpack( "{$endian}Hrevision/{$endian}Hcount/{$endian}HposOriginals/{$endian}HposTranslations/{$endian}HsizeHash/{$endian}HposHash", $this->_substr( $header, 4 ) );
                if ( !is_array( $header ) )
                        return false;

                extract( $header );

                // support revision 0 of MO format specs, only
                if ( $Hrevision != 0 )
                        return false;



                /*
                 * read index tables on originals and translations
                 */

                // read originals' index
                fseek( $file, $HposOriginals, SEEK_SET );

                $originals = fread( $file, $Hcount * 8 );
                if ( $this->_strlen( $originals ) != $Hcount * 8 )
                        return false;

                // read translations index
                fseek( $file, $HposTranslations, SEEK_SET );

                $translations = fread( $file, $Hcount * 8 );
                if ( $this->_strlen( $translations ) != $Hcount * 8 )
                        return false;

                // transform raw data into set of indices
                $originals    = $this->_str_split( $originals, 8 );
                $translations = $this->_str_split( $translations, 8 );



                /*
                 * read set of strings to separate string
                 */

                // find position of first string in file
                $HposStrings = 0x7FFFFFFF;

                for ( $i = 0; $i < $Hcount; $i++ )
                {

                        // parse index records on original and related translation
                        $o = unpack( "{$endian}length/{$endian}pos", $originals[$i] );
                        $t = unpack( "{$endian}length/{$endian}pos", $translations[$i] );

                        if ( !$o || !$t )
                                return false;

                        $originals[$i]    = $o;
                        $translations[$i] = $t;

                        $HposStrings = min( $HposStrings, $o['pos'], $t['pos'] );

                }

                // read strings expected in rest of file
                fseek( $file, $HposStrings, SEEK_SET );

                $strings = '';
                while ( !feof( $file ) )
                        $strings .= fread( $file, 4096 );

                fclose( $file );



                // collect hash records
                $hash = $header = array();

                for ( $i = 0; $i < $Hcount; $i++ )
                {

                        // adjust offset due to reading strings to separate space before
                        $originals[$i]['pos']    -= $HposStrings;
                        $translations[$i]['pos'] -= $HposStrings;

                        // extract original and translations
                        $original    = $this->_substr( $strings, $originals[$i]['pos'], $originals[$i]['length'] );
                        $translation = $this->_substr( $strings, $translations[$i]['pos'], $translations[$i]['length'] );



                        if ( $original === '' )
                        {
                                // got header --> store separately

                                $header = array();

                                foreach ( explode( "\n", $translation ) as $line )
                                {

                                        $sep = $this->_strpos( $line, ':' );
                                        if ( $sep !== false )
                                                $header[trim($this->_substr( $line, 0, $sep ))] = trim( $this->_substr( $line, $sep + 1 ));

                                }
                        }
                        else
                        {

                                // detect context in original
                                $sep = $this->_strpos( $original, "\04" );
                                if ( $sep !== false )
                                {
                                        $context  = $this->_substr( $original, 0, $sep );
                                        $original = $this->_substr( $original, $sep + 1 );
                                }
                                else
                                        $context  = null;


                                $original     = explode( "\00", $original );
                                $translation  = explode( "\00", $translation );

                                $singularFrom = array_shift( $original );
                                $singularTo   = array_shift( $translation );

                                $record = array(
                                                                'C' => $context,                // context
                                                                'S' => $singularFrom,   // singular original
                                                                'X' => $original,               // plural orignal
                                                                'T' => $singularTo,             // singular translation
                                                                'P' => $translation,    // plural translations
                                                                );

                                $key = is_null( $context ) ? $singularFrom
                                                                                   : "$context\04$singularFrom";


                                $hash[$key] = $record;

                        }
                }


                $this->headers = $header;
                $this->entries = $hash;



                /*
                 * write result to cache
                 */

                wp_cache_set( $filename, array( $header, $hash ), 'l10n' );



                return true;

        }


        /**
         * Retrieves translation of single entry.
         *
         * The provided entry is constructed manually and thus providing proper
         * key for lookup. The method returns a matching entry from internal pool
         * of translations.
         *
         * NOTE! Passing by reference is required as long as parent class is using
         *       it (obviously due to keeping things compatible with PHP4).
         *
         * @param Translate_Entry $entry entry to look up
         * @return Translate_Entry resulting entry
         */

        function translate_entry( &$entry )
        {

                if ( !isset( $this->entries[$entry->key()] ) )
                        return false;

                if ( is_array( $this->entries[$entry->key()] ) )
                {
                        // convert entry to instance of Translate_Entry on demand

                        // use internally managed record
                        $args = $this->entries[$entry->key()];

                        // add structures required by Translation_Entry
                        $args['translations'] = array( $args['T'] );
                        if ( count( $args['P'] ) )
                                $args['translations'] = array_merge( $args['translations'], array_values( $args['P'] ) );

                        if ( count( $args['P'] ) )
                                $args['plural'] = array_shift( array_keys( $args['P'] ) );

                        // temporarily transform entry
                        return new Translation_Entry( $args );

                }

                if ( $this->entries[$entry->key()] instanceof Translation_Entry )
                        return $this->entries[$entry->key()];

                return false;

        }


        function translate( $singular, $context = null )
        {

                $key = $this->key( $singular, $context );

                if ( is_array( $this->entries[$key] ) )
                        if ( is_string( $this->entries[$key]['T'] ) )
                                return $this->entries[$key]['T'];

                return $singular;

        }


        function translate_plural( $singular, $plural, $count, $context = null )
        {

                $key = $this->key( $singular, $context );

                $translated = $this->entries[$key];
                if ( is_array( $translated ) )
                {

                        $index    = $this->select_plural_form( $count );
                        $nplurals = $this->nplurals;

                        if ( ( $index >= 0 ) && ( $index < $nplurals ) )
                        {

                                if ( ( $index == 0 ) && is_string( $translated['T'] ) )
                                        // retrieve singular form
                                        return $translated['T'];

                                if ( is_string( $translated['P'][--$index] ) )
                                        // retrieve selected plural form
                                        return $translated['P'][$index];

                        }
                }


                // retrieve one of the provided original forms depending on $count
                return ( $count == 1 ) ? $singular : $plural;

        }


        function key( $singular, $context = null )
        {
                return is_null( $context ) ? $singular : "$context\04$singular";
        }


        function get_plural_forms_count()
        {
                return $this->nplurals;
        }


        function select_plural_form( $count )
        {

                $count = intval( $count );

                if ( trim( $this->pluralTerm ) === '' )
                        return ( $count === 1 ) ? 0 : 1;

                return intval( eval( 'return (' . str_replace( 'n', '$count', $this->pluralTerm ) . ');' ) );

        }
}


?>