moodle/lib/textlib.class.php

<?php
// This file is part of Moodle - http://moodle.org/
//
// Moodle is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// Moodle is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with Moodle.  If not, see <http://www.gnu.org/licenses/>.

/**
 * Defines string apis
 *
 * @package    core
 * @copyright  (C) 2001-3001 Eloy Lafuente (stronk7) {@link http://contiento.com}
 * @license    http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
 */

defined('MOODLE_INTERNAL') || die();

/**
 * defines string api's for manipulating strings
 *
 * This class is used to manipulate strings under Moodle 1.6 an later. As
 * utf-8 text become mandatory a pool of safe functions under this encoding
 * become necessary. The name of the methods is exactly the
 * same than their PHP originals.
 *
 * A big part of this class acts as a wrapper over the Typo3 charset library,
 * really a cool group of utilities to handle texts and encoding conversion.
 *
 * Take a look to its own copyright and license details.
 *
 * IMPORTANT Note: Typo3 libraries always expect lowercase charsets to use 100%
 * its capabilities so, don't forget to make the conversion
 * from every wrapper function!
 *
 * @package   core
 * @category  string
 * @copyright 1999 onwards Martin Dougiamas  {@link http://moodle.com}
 * @license   http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
 */
class textlib {

    /**
     * Return t3lib helper class, which is used for conversion between charsets
     *
     * @param bool $reset
     * @return t3lib_cs
     */
    protected static function typo3($reset = false) {
        static $typo3cs = null;

        if ($reset) {
            $typo3cs = null;
            return null;
        }

        if (isset($typo3cs)) {
            return $typo3cs;
        }

        global $CFG;

        // Required files
        require_once($CFG->libdir.'/typo3/class.t3lib_cs.php');
        require_once($CFG->libdir.'/typo3/class.t3lib_div.php');
        require_once($CFG->libdir.'/typo3/interface.t3lib_singleton.php');
        require_once($CFG->libdir.'/typo3/class.t3lib_l10n_locales.php');

        // do not use mbstring or recode because it may return invalid results in some corner cases
        $GLOBALS['TYPO3_CONF_VARS']['SYS']['t3lib_cs_convMethod'] = 'iconv';
        $GLOBALS['TYPO3_CONF_VARS']['SYS']['t3lib_cs_utils'] = 'iconv';

        // Tell Typo3 we are curl enabled always (mandatory since 2.0)
        $GLOBALS['TYPO3_CONF_VARS']['SYS']['curlUse'] = '1';

        // And this directory must exist to allow Typo to cache conversion
        // tables when using internal functions
        make_temp_directory('typo3temp/cs');

        // Make sure typo is using our dir permissions
        $GLOBALS['TYPO3_CONF_VARS']['BE']['folderCreateMask'] = decoct($CFG->directorypermissions);

        // Default mask for Typo
        $GLOBALS['TYPO3_CONF_VARS']['BE']['fileCreateMask'] = $CFG->directorypermissions;

        // This full path constants must be defined too, transforming backslashes
        // to forward slashed because Typo3 requires it.
        if (!defined('PATH_t3lib')) {
            define('PATH_t3lib', str_replace('\\','/',$CFG->libdir.'/typo3/'));
            define('PATH_typo3', str_replace('\\','/',$CFG->libdir.'/typo3/'));
            define('PATH_site', str_replace('\\','/',$CFG->tempdir.'/'));
            define('TYPO3_OS', stristr(PHP_OS,'win')&&!stristr(PHP_OS,'darwin')?'WIN':'');
        }

        $typo3cs = new t3lib_cs();

        return $typo3cs;
    }

    /**
     * Reset internal textlib caches.
     * @static
     */
    public static function reset_caches() {
        self::typo3(true);
    }

    /**
     * Standardise charset name
     *
     * Please note it does not mean the returned charset is actually supported.
     *
     * @static
     * @param string $charset raw charset name
     * @return string normalised lowercase charset name
     */
    public static function parse_charset($charset) {
        $charset = strtolower($charset);

        // shortcuts so that we do not have to load typo3 on every page

        if ($charset === 'utf8' or $charset === 'utf-8') {
            return 'utf-8';
        }

        if (preg_match('/^(cp|win|windows)-?(12[0-9]{2})$/', $charset, $matches)) {
            return 'windows-'.$matches[2];
        }

        if (preg_match('/^iso-8859-[0-9]+$/', $charset, $matches)) {
            return $charset;
        }

        if ($charset === 'euc-jp') {
            return 'euc-jp';
        }
        if ($charset === 'iso-2022-jp') {
            return 'iso-2022-jp';
        }
        if ($charset === 'shift-jis' or $charset === 'shift_jis') {
            return 'shift_jis';
        }
        if ($charset === 'gb2312') {
            return 'gb2312';
        }
        if ($charset === 'gb18030') {
            return 'gb18030';
        }

        // fallback to typo3
        return self::typo3()->parse_charset($charset);
    }

    /**
     * Converts the text between different encodings. It uses iconv extension with //TRANSLIT parameter,
     * falls back to typo3.
     * Returns false if fails.
     *
     * @param string $text
     * @param string $fromCS source encoding
     * @param string $toCS result encoding
     * @return string|bool converted string or false on error
     */
    public static function convert($text, $fromCS, $toCS='utf-8') {
        $fromCS = self::parse_charset($fromCS);
        $toCS   = self::parse_charset($toCS);

        $text = (string)$text; // we can work only with strings

        if ($text === '') {
            return '';
        }

        $result = iconv($fromCS, $toCS.'//TRANSLIT', $text);

        if ($result === false or $result === '') {
            // note: iconv is prone to return empty string when invalid char encountered, or false if encoding unsupported
            $oldlevel = error_reporting(E_PARSE);
            $result = self::typo3()->conv((string)$text, $fromCS, $toCS);
            error_reporting($oldlevel);
        }

        return $result;
    }

    /**
     * Multibyte safe substr() function, uses mbstring or iconv for UTF-8, falls back to typo3.
     *
     * @param string $text string to truncate
     * @param int $start negative value means from end
     * @param int $len maximum length of characters beginning from start
     * @param string $charset encoding of the text
     * @return string portion of string specified by the $start and $len
     */
    public static function substr($text, $start, $len=null, $charset='utf-8') {
        $charset = self::parse_charset($charset);

        if ($charset === 'utf-8') {
            if (function_exists('mb_substr')) {
                // this is much faster than iconv - see MDL-31142
                if ($len === null) {
                    $oldcharset = mb_internal_encoding();
                    mb_internal_encoding('UTF-8');
                    $result = mb_substr($text, $start);
                    mb_internal_encoding($oldcharset);
                    return $result;
                } else {
                    return mb_substr($text, $start, $len, 'UTF-8');
                }

            } else {
                if ($len === null) {
                    $len = iconv_strlen($text, 'UTF-8');
                }
                return iconv_substr($text, $start, $len, 'UTF-8');
            }
        }

        $oldlevel = error_reporting(E_PARSE);
        if ($len === null) {
            $result = self::typo3()->substr($charset, (string)$text, $start);
        } else {
            $result = self::typo3()->substr($charset, (string)$text, $start, $len);
        }
        error_reporting($oldlevel);

        return $result;
    }

    /**
     * Multibyte safe strlen() function, uses mbstring or iconv for UTF-8, falls back to typo3.
     *
     * @param string $text input string
     * @param string $charset encoding of the text
     * @return int number of characters
     */
    public static function strlen($text, $charset='utf-8') {
        $charset = self::parse_charset($charset);

        if ($charset === 'utf-8') {
            if (function_exists('mb_strlen')) {
                return mb_strlen($text, 'UTF-8');
            } else {
                return iconv_strlen($text, 'UTF-8');
            }
        }

        $oldlevel = error_reporting(E_PARSE);
        $result = self::typo3()->strlen($charset, (string)$text);
        error_reporting($oldlevel);

        return $result;
    }

    /**
     * Multibyte safe strtolower() function, uses mbstring, falls back to typo3.
     *
     * @param string $text input string
     * @param string $charset encoding of the text (may not work for all encodings)
     * @return string lower case text
     */
    public static function strtolower($text, $charset='utf-8') {
        $charset = self::parse_charset($charset);

        if ($charset === 'utf-8' and function_exists('mb_strtolower')) {
            return mb_strtolower($text, 'UTF-8');
        }

        $oldlevel = error_reporting(E_PARSE);
        $result = self::typo3()->conv_case($charset, (string)$text, 'toLower');
        error_reporting($oldlevel);

        return $result;
    }

    /**
     * Multibyte safe strtoupper() function, uses mbstring, falls back to typo3.
     *
     * @param string $text input string
     * @param string $charset encoding of the text (may not work for all encodings)
     * @return string upper case text
     */
    public static function strtoupper($text, $charset='utf-8') {
        $charset = self::parse_charset($charset);

        if ($charset === 'utf-8' and function_exists('mb_strtoupper')) {
            return mb_strtoupper($text, 'UTF-8');
        }

        $oldlevel = error_reporting(E_PARSE);
        $result = self::typo3()->conv_case($charset, (string)$text, 'toUpper');
        error_reporting($oldlevel);

        return $result;
    }

    /**
     * Find the position of the first occurrence of a substring in a string.
     * UTF-8 ONLY safe strpos(), uses mbstring, falls back to iconv.
     *
     * @param string $haystack the string to search in
     * @param string $needle one or more charachters to search for
     * @param int $offset offset from begining of string
     * @return int the numeric position of the first occurrence of needle in haystack.
     */
    public static function strpos($haystack, $needle, $offset=0) {
        if (function_exists('mb_strpos')) {
            return mb_strpos($haystack, $needle, $offset, 'UTF-8');
        } else {
            return iconv_strpos($haystack, $needle, $offset, 'UTF-8');
        }
    }

    /**
     * Find the position of the last occurrence of a substring in a string
     * UTF-8 ONLY safe strrpos(), uses mbstring, falls back to iconv.
     *
     * @param string $haystack the string to search in
     * @param string $needle one or more charachters to search for
     * @return int the numeric position of the last occurrence of needle in haystack
     */
    public static function strrpos($haystack, $needle) {
        if (function_exists('mb_strpos')) {
            return mb_strrpos($haystack, $needle, null, 'UTF-8');
        } else {
            return iconv_strrpos($haystack, $needle, 'UTF-8');
        }
    }

    /**
     * Try to convert upper unicode characters to plain ascii,
     * the returned string may contain unconverted unicode characters.
     *
     * @param string $text input string
     * @param string $charset encoding of the text
     * @return string converted ascii string
     */
    public static function specialtoascii($text, $charset='utf-8') {
        $charset = self::parse_charset($charset);
        $oldlevel = error_reporting(E_PARSE);
        $result = self::typo3()->specCharsToASCII($charset, (string)$text);
        error_reporting($oldlevel);
        return $result;
    }

    /**
     * Generate a correct base64 encoded header to be used in MIME mail messages.
     * This function seems to be 100% compliant with RFC1342. Credits go to:
     * paravoid (http://www.php.net/manual/en/function.mb-encode-mimeheader.php#60283).
     *
     * @param string $text input string
     * @param string $charset encoding of the text
     * @return string base64 encoded header
     */
    public static function encode_mimeheader($text, $charset='utf-8') {
        if (empty($text)) {
            return (string)$text;
        }
        // Normalize charset
        $charset = self::parse_charset($charset);
        // If the text is pure ASCII, we don't need to encode it
        if (self::convert($text, $charset, 'ascii') == $text) {
            return $text;
        }
        // Although RFC says that line feed should be \r\n, it seems that
        // some mailers double convert \r, so we are going to use \n alone
        $linefeed="\n";
        // Define start and end of every chunk
        $start = "=?$charset?B?";
        $end = "?=";
        // Accumulate results
        $encoded = '';
        // Max line length is 75 (including start and end)
        $length = 75 - strlen($start) - strlen($end);
        // Multi-byte ratio
        $multilength = self::strlen($text, $charset);
        // Detect if strlen and friends supported
        if ($multilength === false) {
            if ($charset == 'GB18030' or $charset == 'gb18030') {
                while (strlen($text)) {
                    // try to encode first 22 chars - we expect most chars are two bytes long
                    if (preg_match('/^(([\x00-\x7f])|([\x81-\xfe][\x40-\x7e])|([\x81-\xfe][\x80-\xfe])|([\x81-\xfe][\x30-\x39]..)){1,22}/m', $text, $matches)) {
                        $chunk = $matches[0];
                        $encchunk = base64_encode($chunk);
                        if (strlen($encchunk) > $length) {
                            // find first 11 chars - each char in 4 bytes - worst case scenario
                            preg_match('/^(([\x00-\x7f])|([\x81-\xfe][\x40-\x7e])|([\x81-\xfe][\x80-\xfe])|([\x81-\xfe][\x30-\x39]..)){1,11}/m', $text, $matches);
                            $chunk = $matches[0];
                            $encchunk = base64_encode($chunk);
                        }
                        $text = substr($text, strlen($chunk));
                        $encoded .= ' '.$start.$encchunk.$end.$linefeed;
                    } else {
                        break;
                    }
                }
                $encoded = trim($encoded);
                return $encoded;
            } else {
                return false;
            }
        }
        $ratio = $multilength / strlen($text);
        // Base64 ratio
        $magic = $avglength = floor(3 * $length * $ratio / 4);
        // basic infinite loop protection
        $maxiterations = strlen($text)*2;
        $iteration = 0;
        // Iterate over the string in magic chunks
        for ($i=0; $i <= $multilength; $i+=$magic) {
            if ($iteration++ > $maxiterations) {
                return false; // probably infinite loop
            }
            $magic = $avglength;
            $offset = 0;
            // Ensure the chunk fits in length, reducing magic if necessary
            do {
                $magic -= $offset;
                $chunk = self::substr($text, $i, $magic, $charset);
                $chunk = base64_encode($chunk);
                $offset++;
            } while (strlen($chunk) > $length);
            // This chunk doesn't break any multi-byte char. Use it.
            if ($chunk)
                $encoded .= ' '.$start.$chunk.$end.$linefeed;
        }
        // Strip the first space and the last linefeed
        $encoded = substr($encoded, 1, -strlen($linefeed));

        return $encoded;
    }

    /**
     * Converts all the numeric entities &#nnnn; or &#xnnn; to UTF-8
     * Original from laurynas dot butkus at gmail at:
     * http://php.net/manual/en/function.html-entity-decode.php#75153
     * with some custom mods to provide more functionality
     *
     * @param string $str input string
     * @param boolean $htmlent convert also html entities (defaults to true)
     * @return string encoded UTF-8 string
     *
     * NOTE: we could have used typo3 entities_to_utf8() here
     *       but the direct alternative used runs 400% quicker
     *       and uses 0.5Mb less memory, so, let's use it
     *       (tested against 10^6 conversions)
     */
    public static function entities_to_utf8($str, $htmlent=true) {
        static $trans_tbl; // Going to use static transliteration table

        // Replace numeric entities
        $result = preg_replace('~&#x([0-9a-f]+);~ei', 'textlib::code2utf8(hexdec("\\1"))', $str);
        $result = preg_replace('~&#([0-9]+);~e', 'textlib::code2utf8(\\1)', $result);

        // Replace literal entities (if desired)
        if ($htmlent) {
            // Generate/create $trans_tbl
            if (!isset($trans_tbl)) {
                $trans_tbl = array();
                foreach (get_html_translation_table(HTML_ENTITIES) as $val=>$key) {
                    $trans_tbl[$key] = utf8_encode($val);
                }
            }
            $result = strtr($result, $trans_tbl);
        }
        // Return utf8-ised string
        return $result;
    }

    /**
     * Converts all Unicode chars > 127 to numeric entities &#nnnn; or &#xnnn;.
     *
     * @param string $str input string
     * @param boolean $dec output decadic only number entities
     * @param boolean $nonnum remove all non-numeric entities
     * @return string converted string
     */
    public static function utf8_to_entities($str, $dec=false, $nonnum=false) {
        // Avoid some notices from Typo3 code
        $oldlevel = error_reporting(E_PARSE);
        if ($nonnum) {
            $str = self::typo3()->entities_to_utf8((string)$str, true);
        }
        $result = self::typo3()->utf8_to_entities((string)$str);
        if ($dec) {
            $result = preg_replace('/&#x([0-9a-f]+);/ie', "'&#'.hexdec('$1').';'", $result);
        }
        // Restore original debug level
        error_reporting($oldlevel);
        return $result;
    }

    /**
     * Removes the BOM from unicode string {@link http://unicode.org/faq/utf_bom.html}
     *
     * @param string $str input string
     * @return string
     */
    public static function trim_utf8_bom($str) {
        $bom = "\xef\xbb\xbf";
        if (strpos($str, $bom) === 0) {
            return substr($str, strlen($bom));
        }
        return $str;
    }

    /**
     * Returns encoding options for select boxes, utf-8 and platform encoding first
     *
     * @return array encodings
     */
    public static function get_encodings() {
        $encodings = array();
        $encodings['UTF-8'] = 'UTF-8';
        $winenc = strtoupper(get_string('localewincharset', 'langconfig'));
        if ($winenc != '') {
            $encodings[$winenc] = $winenc;
        }
        $nixenc = strtoupper(get_string('oldcharset', 'langconfig'));
        $encodings[$nixenc] = $nixenc;

        foreach (self::typo3()->synonyms as $enc) {
            $enc = strtoupper($enc);
            $encodings[$enc] = $enc;
        }
        return $encodings;
    }

    /**
     * Returns the utf8 string corresponding to the unicode value
     * (from php.net, courtesy - romans@void.lv)
     *
     * @param  int    $num one unicode value
     * @return string the UTF-8 char corresponding to the unicode value
     */
    public static function code2utf8($num) {
        if ($num < 128) {
            return chr($num);
        }
        if ($num < 2048) {
            return chr(($num >> 6) + 192) . chr(($num & 63) + 128);
        }
        if ($num < 65536) {
            return chr(($num >> 12) + 224) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
        }
        if ($num < 2097152) {
            return chr(($num >> 18) + 240) . chr((($num >> 12) & 63) + 128) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
        }
        return '';
    }

    /**
     * Makes first letter of each word capital - words must be separated by spaces.
     * Use with care, this function does not work properly in many locales!!!
     *
     * @param string $text input string
     * @return string
     */
    public static function strtotitle($text) {
        if (empty($text)) {
            return $text;
        }

        if (function_exists('mb_convert_case')) {
            return mb_convert_case($text, MB_CASE_TITLE, 'UTF-8');
        }

        $text = self::strtolower($text);
        $words = explode(' ', $text);
        foreach ($words as $i=>$word) {
            $length = self::strlen($word);
            if (!$length) {
                continue;

            } else if ($length == 1) {
                $words[$i] = self::strtoupper($word);

            } else {
                $letter = self::substr($word, 0, 1);
                $letter = self::strtoupper($letter);
                $rest   = self::substr($word, 1);
                $words[$i] = $letter.$rest;
            }
        }
        return implode(' ', $words);
    }

    /**
     * Locale aware sorting, the key associations are kept, values are sorted alphabetically.
     *
     * @param array $arr array to be sorted (reference)
     * @param int $sortflag One of Collator::SORT_REGULAR, Collator::SORT_NUMERIC, Collator::SORT_STRING
     * @return void modifies parameter
     */
    public static function asort(array &$arr, $sortflag = null) {
        debugging('textlib::asort has been superseeded by collatorlib::asort please upgrade your code to use that', DEBUG_DEVELOPER);
        collatorlib::asort($arr, $sortflag);
    }
}


/**
 * A collator class with static methods that can be used for sorting.
 *
 * @package   core
 * @copyright 2011 Sam Hemelryk
 *            2012 Petr Skoda
 * @license   http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
 */
class collatorlib {
    /** @const compare items using general PHP comparison, equivalent to Collator::SORT_REGULAR, this may bot be locale aware! */
    const SORT_REGULAR = 0;

    /** @const compare items as strings, equivalent to Collator::SORT_STRING */
    const SORT_STRING = 1;

    /** @const compare items as numbers, equivalent to Collator::SORT_NUMERIC */
    const SORT_NUMERIC = 2;

    /** @const compare items like natsort(), equivalent to SORT_NATURAL */
    const SORT_NATURAL = 6;

    /** @const do not ignore case when sorting, use bitwise "|" with SORT_NATURAL or SORT_STRING, equivalent to Collator::UPPER_FIRST */
    const CASE_SENSITIVE = 64;

    /** @var Collator|false|null **/
    protected static $collator = null;

    /** @var string|null The locale that was used in instantiating the current collator **/
    protected static $locale = null;

    /**
     * Prevent class instances, all methods are static.
     */
    private function __construct() {
    }

    /**
     * Ensures that a collator is available and created
     *
     * @return bool Returns true if collation is available and ready
     */
    protected static function ensure_collator_available() {
        $locale = get_string('locale', 'langconfig');
        if (is_null(self::$collator) || $locale != self::$locale) {
            self::$collator = false;
            self::$locale = $locale;
            if (class_exists('Collator', false)) {
                $collator = new Collator($locale);
                if (!empty($collator) && $collator instanceof Collator) {
                    // Check for non fatal error messages. This has to be done immediately
                    // after instantiation as any further calls to collation will cause
                    // it to reset to 0 again (or another error code if one occurred)
                    $errorcode = $collator->getErrorCode();
                    $errormessage = $collator->getErrorMessage();
                    // Check for an error code, 0 means no error occurred
                    if ($errorcode !== 0) {
                        // Get the actual locale being used, e.g. en, he, zh
                        $localeinuse = $collator->getLocale(Locale::ACTUAL_LOCALE);
                        // Check for the common fallback warning error codes. If this occurred
                        // there is normally little to worry about:
                        // - U_USING_DEFAULT_WARNING (127)  - default fallback locale used (pt => UCA)
                        // - U_USING_FALLBACK_WARNING (128) - fallback locale used (de_CH => de)
                        // (UCA: Unicode Collation Algorithm http://unicode.org/reports/tr10/)
                        if ($errorcode === -127 || $errorcode === -128) {
                            // Check if the locale in use is UCA default one ('root') or
                            // if it is anything like the locale we asked for
                            if ($localeinuse !== 'root' && strpos($locale, $localeinuse) !== 0) {
                                // The locale we asked for is completely different to the locale
                                // we have received, let the user know via debugging
                                debugging('Invalid locale: "' . $locale . '", with warning (not fatal) "' . $errormessage .
                                    '", falling back to "' . $collator->getLocale(Locale::VALID_LOCALE) . '"');
                            } else {
                                // Nothing to do here, this is expected!
                                // The Moodle locale setting isn't what the collator expected but
                                // it is smart enough to match the first characters of our locale
                                // to find the correct locale or to use UCA collation
                            }
                        } else {
                            // We've received some other sort of non fatal warning - let the
                            // user know about it via debugging.
                            debugging('Problem with locale: "' . $locale . '", with message "' . $errormessage .
                                '", falling back to "' . $collator->getLocale(Locale::VALID_LOCALE) . '"');
                        }
                    }
                    // Store the collator object now that we can be sure it is in a workable condition
                    self::$collator = $collator;
                } else {
                    // Fatal error while trying to instantiate the collator... something went wrong
                    debugging('Error instantiating collator for locale: "' . $locale . '", with error [' .
                        intl_get_error_code() . '] ' . intl_get_error_message($collator));
                }
            }
        }
        return (self::$collator instanceof Collator);
    }

    /**
     * Restore array contents keeping new keys.
     * @static
     * @param array $arr
     * @param array $original
     * @return void modifies $arr
     */
    protected static function restore_array(array &$arr, array &$original) {
        foreach ($arr as $key => $ignored) {
            $arr[$key] = $original[$key];
        }
    }

    /**
     * Normalise numbers in strings for natural sorting comparisons.
     * @static
     * @param string $string
     * @return string string with normalised numbers
     */
    protected static function naturalise($string) {
        return preg_replace_callback('/[0-9]+/', array('collatorlib', 'callback_naturalise'), $string);
    }

    /**
     * @internal
     * @static
     * @param array $matches
     * @return string
     */
    public static function callback_naturalise($matches) {
        return str_pad($matches[0], 20, '0', STR_PAD_LEFT);
    }

    /**
     * Locale aware sorting, the key associations are kept, values are sorted alphabetically.
     *
     * @param array $arr array to be sorted (reference)
     * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
     *      optionally "|" collatorlib::CASE_SENSITIVE
     * @return bool True on success
     */
    public static function asort(array &$arr, $sortflag = collatorlib::SORT_STRING) {
        if (empty($arr)) {
            // nothing to do
            return true;
        }

        $original = null;

        $casesensitive = (bool)($sortflag & collatorlib::CASE_SENSITIVE);
        $sortflag = ($sortflag & ~collatorlib::CASE_SENSITIVE);
        if ($sortflag != collatorlib::SORT_NATURAL and $sortflag != collatorlib::SORT_STRING) {
            $casesensitive = false;
        }

        if (self::ensure_collator_available()) {
            if ($sortflag == collatorlib::SORT_NUMERIC) {
                $flag = Collator::SORT_NUMERIC;

            } else if ($sortflag == collatorlib::SORT_REGULAR) {
                $flag = Collator::SORT_REGULAR;

            } else {
                $flag = Collator::SORT_STRING;
            }

            if ($sortflag == collatorlib::SORT_NATURAL) {
                $original = $arr;
                if ($sortflag == collatorlib::SORT_NATURAL) {
                    foreach ($arr as $key => $value) {
                        $arr[$key] = self::naturalise((string)$value);
                    }
                }
            }
            if ($casesensitive) {
                self::$collator->setAttribute(Collator::CASE_FIRST, Collator::UPPER_FIRST);
            } else {
                self::$collator->setAttribute(Collator::CASE_FIRST, Collator::OFF);
            }
            $result = self::$collator->asort($arr, $flag);
            if ($original) {
                self::restore_array($arr, $original);
            }
            return $result;
        }

        // try some fallback that works at least for English

        if ($sortflag == collatorlib::SORT_NUMERIC) {
            return asort($arr, SORT_NUMERIC);

        } else if ($sortflag == collatorlib::SORT_REGULAR) {
            return asort($arr, SORT_REGULAR);
        }

        if (!$casesensitive) {
            $original = $arr;
            foreach ($arr as $key => $value) {
                $arr[$key] = textlib::strtolower($value);
            }
        }

        if ($sortflag == collatorlib::SORT_NATURAL) {
            $result = natsort($arr);

        } else {
            $result = asort($arr, SORT_LOCALE_STRING);
        }

        if ($original) {
            self::restore_array($arr, $original);
        }

        return $result;
    }

    /**
     * Locale aware sort of objects by a property in common to all objects
     *
     * @param array $objects An array of objects to sort (handled by reference)
     * @param string $property The property to use for comparison
     * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
     *      optionally "|" collatorlib::CASE_SENSITIVE
     * @return bool True on success
     */
    public static function asort_objects_by_property(array &$objects, $property, $sortflag = collatorlib::SORT_STRING) {
        $original = $objects;
        foreach ($objects as $key => $object) {
            $objects[$key] = $object->$property;
        }
        $result = self::asort($objects, $sortflag);
        self::restore_array($objects, $original);
        return $result;
    }

    /**
     * Locale aware sort of objects by a method in common to all objects
     *
     * @param array $objects An array of objects to sort (handled by reference)
     * @param string $method The method to call to generate a value for comparison
     * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
     *      optionally "|" collatorlib::CASE_SENSITIVE
     * @return bool True on success
     */
    public static function asort_objects_by_method(array &$objects, $method, $sortflag = collatorlib::SORT_STRING) {
        $original = $objects;
        foreach ($objects as $key => $object) {
            $objects[$key] = $object->{$method}();
        }
        $result = self::asort($objects, $sortflag);
        self::restore_array($objects, $original);
        return $result;
    }

    /**
     * Locale aware sorting, the key associations are kept, keys are sorted alphabetically.
     *
     * @param array $arr array to be sorted (reference)
     * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
     *      optionally "|" collatorlib::CASE_SENSITIVE
     * @return bool True on success
     */
    public static function ksort(array &$arr, $sortflag = collatorlib::SORT_STRING) {
        $keys = array_keys($arr);
        if (!self::asort($keys, $sortflag)) {
            return false;
        }
        // This is a bit slow, but we need to keep the references
        $original = $arr;
        $arr = array(); // Surprisingly this does not break references outside
        foreach ($keys as $key) {
            $arr[$key] = $original[$key];
        }

        return true;
    }
}