generated from jric11/baseProject
87 lines
4.3 KiB
PHP
87 lines
4.3 KiB
PHP
<?php
|
|
|
|
declare(strict_types=1);
|
|
|
|
/**
|
|
* Pattern.php
|
|
*
|
|
* @since 2011-05-23
|
|
* @category Library
|
|
* @package UnicodeData
|
|
* @author Nicola Asuni <info@tecnick.com>
|
|
* @copyright 2011-2026 Nicola Asuni - Tecnick.com LTD
|
|
* @license https://www.gnu.org/copyleft/lesser.html GNU-LGPL v3 (see LICENSE)
|
|
* @link https://github.com/tecnickcom/tc-lib-unicode-data
|
|
*
|
|
* This file is part of tc-lib-unicode-data software library.
|
|
*/
|
|
|
|
namespace Com\Tecnick\Unicode\Data;
|
|
|
|
/**
|
|
* Com\Tecnick\Unicode\Data\Pattern
|
|
*
|
|
* Regular expressions matching directional text.
|
|
*
|
|
* Generated by tools/generate.php from the Unicode Character Database 17.0.0.
|
|
* Do not edit manually.
|
|
*
|
|
* @since 2011-05-23
|
|
* @category Library
|
|
* @package UnicodeData
|
|
* @author Nicola Asuni <info@tecnick.com>
|
|
* @copyright 2011-2026 Nicola Asuni - Tecnick.com LTD
|
|
* @license https://www.gnu.org/copyleft/lesser.html GNU-LGPL v3 (see LICENSE)
|
|
* @link https://github.com/tecnickcom/tc-lib-unicode-data
|
|
*/
|
|
|
|
class Pattern
|
|
{
|
|
/**
|
|
* Pattern matching the strong right-to-left characters (Bidi_Class=R) and the
|
|
* right-to-left explicit formatting characters (RLE, RLO, RLI).
|
|
* Arabic characters are matched by the ARABIC pattern instead.
|
|
* The subject string must be valid UTF-8.
|
|
*/
|
|
public const RTL =
|
|
'/['
|
|
. '\x{05BE}\x{05C0}\x{05C3}\x{05C6}\x{05D0}-\x{05EA}\x{05EF}-\x{05F4}\x{07C0}-\x{07EA}'
|
|
. '\x{07F4}\x{07F5}\x{07FA}\x{07FE}-\x{0815}\x{081A}\x{0824}\x{0828}\x{0830}-\x{083E}'
|
|
. '\x{0840}-\x{0858}\x{085E}\x{200F}\x{202B}\x{202E}\x{2067}\x{FB1D}\x{FB1F}-\x{FB28}'
|
|
. '\x{FB2A}-\x{FB36}\x{FB38}-\x{FB3C}\x{FB3E}\x{FB40}\x{FB41}\x{FB43}\x{FB44}\x{FB46}-\x{FB4F}'
|
|
. '\x{10800}-\x{10805}\x{10808}\x{1080A}-\x{10835}\x{10837}\x{10838}\x{1083C}'
|
|
. '\x{1083F}-\x{10855}\x{10857}-\x{1089E}\x{108A7}-\x{108AF}\x{108E0}-\x{108F2}'
|
|
. '\x{108F4}\x{108F5}\x{108FB}-\x{1091B}\x{10920}-\x{10939}\x{1093F}-\x{10959}'
|
|
. '\x{10980}-\x{109B7}\x{109BC}-\x{109CF}\x{109D2}-\x{10A00}\x{10A10}-\x{10A13}'
|
|
. '\x{10A15}-\x{10A17}\x{10A19}-\x{10A35}\x{10A40}-\x{10A48}\x{10A50}-\x{10A58}'
|
|
. '\x{10A60}-\x{10A9F}\x{10AC0}-\x{10AE4}\x{10AEB}-\x{10AF6}\x{10B00}-\x{10B35}'
|
|
. '\x{10B40}-\x{10B55}\x{10B58}-\x{10B72}\x{10B78}-\x{10B91}\x{10B99}-\x{10B9C}'
|
|
. '\x{10BA9}-\x{10BAF}\x{10C00}-\x{10C48}\x{10C80}-\x{10CB2}\x{10CC0}-\x{10CF2}'
|
|
. '\x{10CFA}-\x{10CFF}\x{10D4A}-\x{10D65}\x{10D6F}-\x{10D85}\x{10D8E}\x{10D8F}'
|
|
. '\x{10E80}-\x{10EA9}\x{10EAD}\x{10EB0}\x{10EB1}\x{10F00}-\x{10F27}\x{10F70}-\x{10F81}'
|
|
. '\x{10F86}-\x{10F89}\x{10FB0}-\x{10FCB}\x{10FE0}-\x{10FF6}\x{1E800}-\x{1E8C4}'
|
|
. '\x{1E8C7}-\x{1E8CF}\x{1E900}-\x{1E943}\x{1E94B}\x{1E950}-\x{1E959}\x{1E95E}\x{1E95F}'
|
|
. ']/u';
|
|
|
|
/**
|
|
* Pattern matching the Arabic characters (Bidi_Class=AL and Bidi_Class=AN).
|
|
* The subject string must be valid UTF-8.
|
|
*/
|
|
public const ARABIC =
|
|
'/['
|
|
. '\x{0600}-\x{0605}\x{0608}\x{060B}\x{060D}\x{061B}-\x{064A}\x{0660}-\x{0669}\x{066B}-\x{066F}'
|
|
. '\x{0671}-\x{06D5}\x{06DD}\x{06E5}\x{06E6}\x{06EE}\x{06EF}\x{06FA}-\x{070D}\x{070F}\x{0710}'
|
|
. '\x{0712}-\x{072F}\x{074D}-\x{07A5}\x{07B1}\x{0860}-\x{086A}\x{0870}-\x{0891}'
|
|
. '\x{08A0}-\x{08C9}\x{08E2}\x{FB50}-\x{FBC2}\x{FBD3}-\x{FD3D}\x{FD50}-\x{FD8F}'
|
|
. '\x{FD92}-\x{FDC7}\x{FDF0}-\x{FDFC}\x{FE70}-\x{FE74}\x{FE76}-\x{FEFC}\x{10D00}-\x{10D23}'
|
|
. '\x{10D30}-\x{10D39}\x{10D40}-\x{10D49}\x{10E60}-\x{10E7E}\x{10EC2}-\x{10EC7}'
|
|
. '\x{10F30}-\x{10F45}\x{10F51}-\x{10F59}\x{1EC71}-\x{1ECB4}\x{1ED01}-\x{1ED3D}'
|
|
. '\x{1EE00}-\x{1EE03}\x{1EE05}-\x{1EE1F}\x{1EE21}\x{1EE22}\x{1EE24}\x{1EE27}'
|
|
. '\x{1EE29}-\x{1EE32}\x{1EE34}-\x{1EE37}\x{1EE39}\x{1EE3B}\x{1EE42}\x{1EE47}\x{1EE49}\x{1EE4B}'
|
|
. '\x{1EE4D}-\x{1EE4F}\x{1EE51}\x{1EE52}\x{1EE54}\x{1EE57}\x{1EE59}\x{1EE5B}\x{1EE5D}\x{1EE5F}'
|
|
. '\x{1EE61}\x{1EE62}\x{1EE64}\x{1EE67}-\x{1EE6A}\x{1EE6C}-\x{1EE72}\x{1EE74}-\x{1EE77}'
|
|
. '\x{1EE79}-\x{1EE7C}\x{1EE7E}\x{1EE80}-\x{1EE89}\x{1EE8B}-\x{1EE9B}\x{1EEA1}-\x{1EEA3}'
|
|
. '\x{1EEA5}-\x{1EEA9}\x{1EEAB}-\x{1EEBB}'
|
|
. ']/u';
|
|
}
|