<?php
// File name:	big5.inc.php
// Description:	PHP subroutine to deal with Big5 characters
// Date:	2004-04-17
// Author:	imacat <imacat@pristine.com.tw>
// Copyright:	Copyright (C) 2004-2007 Pristine Communications

// 編碼規則：
//　高位元組：  A1-FE, 8E-A0, 81-8D
//  低位元組：  40-7E, A1-FE
//  常用字：    A440-C67E
//  次常用字：  C940-F9D5
//  常用符號：  A140-A3BF
//  罕用符號：  C6A1-C8FE
//  控制碼：    A3C0-A3E0

// Set the include path
if (!defined("INCPATH_SET")) {
    require_once dirname(__FILE__) . "/incpath.inc.php";
}
// Referenced subroutines
require_once "monica/xfileio.inc.php";

// Settings
define("_BIG5_VAR", "/tmp/big5.var");
$_BIG5 = null;

// init_big5: Initialize the Big5 characters array
function init_big5()
{
    global $_BIG5;
    // Already initialized
    if (!is_null($_BIG5)) {
        return;
    }
    // Initialize from the existing table
    if (file_exists(_BIG5_VAR)) {
        eval(xfread(_BIG5_VAR));
        return;
    }
    // Initialize the characters
    $_BIG5 = array(
        "常用字"    => _big5_gen_chars(0xA440, 0xC67E),
        "次常用字"  => _big5_gen_chars(0xC940, 0xF9D5),
        "常用符號"  => _big5_gen_chars(0xA140, 0xA3BF),
        //"罕用符號"  => _big5_gen_chars(0xC6A1, 0xC8FE),
        //"控制碼"    => _big5_gen_chars(0xA3C0, 0xA3E0),
    );
    
    // Save the result
    $string = "\$_BIG5 = " . var_export($_BIG5, true) . ";\n";
    xfupdate(_BIG5_VAR, $string);
    return;
}

// _big5_gen_chars: Generate a range of Big5 characters
function _big5_gen_chars($start, $end)
{
    $chars = array();
    for ($code = $start; $code <= $end; ) {
        $chars[] = pack("n", $code);
        # Next character
        $hi = $code & 0xFF00;
        $lo = $code & 0x00FF;
        if ($lo == 0xFE) {
            $lo = 0x40;
            $hi += 0x0100;
        } elseif ($lo == 0x7E) {
            $lo = 0xA1;
        } else {
            $lo++;
        }
        $code = $hi | $lo;
    }
    return $chars;
}

// big5_feq_chars: Return the Big5 frequently-used characters
function &big5_feq_chars()
{
    init_big5();
    return $GLOBALS["_BIG5"]["常用字"];
}

// big5_nonfeq_chars: Return the Big5 non-frequently-used characters
function &big5_nonfeq_chars()
{
    init_big5();
    return $GLOBALS["_BIG5"]["次常用字"];
}

// big5_punc_chars: Return the Big5 punctuation characters
function &big5_punc_chars()
{
    init_big5();
    return $GLOBALS["_BIG5"]["常用符號"];
}

?>