135 lines
3.8 KiB
PHP
135 lines
3.8 KiB
PHP
<?php
|
||
// File name: gb2312.inc.php
|
||
// Description: PHP subroutine to deal with GB2312 characters
|
||
// Date: 2004-04-27
|
||
// Author: imacat <imacat@pristine.com.tw>
|
||
// Copyright: Copyright (C) 2004-2007 Pristine Communications
|
||
|
||
// 編碼規則：
|
||
// 區號值： 1-94
|
||
// 位號值： 1-94
|
||
// 字碼： 最高位元1 + (區號值+0x20(32)), 最高位元1 + (位號值+0x20(32))
|
||
// 符號： 1-9區
|
||
// 一級字： 16-55區（常用漢字）
|
||
// 二級字： 56-87區（次常用漢字）
|
||
// 空白區： 10-16區, 87-94區
|
||
|
||
// 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("_GB2312_VAR", "/tmp/gb2312.var");
|
||
$_GB2312 = null;
|
||
|
||
// init_gb2312: Initialize the GB2312 characters array
|
||
function init_gb2312()
|
||
{
|
||
global $_GB2312;
|
||
// Already initialized
|
||
if (!is_null($_GB2312)) {
|
||
return;
|
||
}
|
||
// Initialize from the existing table
|
||
if (file_exists(_GB2312_VAR)) {
|
||
eval(xfread(_GB2312_VAR));
|
||
return;
|
||
}
|
||
// Initialize the characters
|
||
$_GB2312 = array(
|
||
// 符號
|
||
"睫瘍" => _gb2312_gen_chars(1, 9),
|
||
// 一級字
|
||
"珨撰趼" => _gb2312_gen_chars(16, 55),
|
||
// 二級字
|
||
"媼撰趼" => _gb2312_gen_chars(56, 87),
|
||
);
|
||
|
||
// Save the result
|
||
$string = "\$_GB2312 = " . var_export($_GB2312, true) . ";\n";
|
||
xfupdate(_GB2312_VAR, $string);
|
||
return;
|
||
}
|
||
|
||
// _gb2312_gen_chars: Generate a range of GB2312 characters
|
||
function _gb2312_gen_chars($start, $end)
|
||
{
|
||
$chars = array();
|
||
for ($sec = $start, $pos = 1; $sec <= $end; ) {
|
||
$code = ((($sec+0x20)|0x80) << 8) | (($pos+0x20)|0x80);
|
||
$chars[] = pack("n", $code);
|
||
// Next character
|
||
$pos++;
|
||
// Carry 進位
|
||
if ($pos > 94) {
|
||
$pos = 1;
|
||
$sec++;
|
||
}
|
||
// 不用的符號
|
||
if ($sec == 2 && $pos == 1) { // 02,01-02,16
|
||
$pos = 17;
|
||
} elseif ($sec == 2 && $pos == 67) { // 02,67-02,68
|
||
$pos = 69;
|
||
} elseif ($sec == 2 && $pos == 79) { // 02,79-02,80
|
||
$pos = 81;
|
||
} elseif ($sec == 2 && $pos == 93) { // 02,93-02,94
|
||
$pos = 1;
|
||
$sec++;
|
||
} elseif ($sec == 4 && $pos == 84) { // 04,84-04,94
|
||
$pos = 1;
|
||
$sec++;
|
||
} elseif ($sec == 5 && $pos == 87) { // 05,87-05,94
|
||
$pos = 1;
|
||
$sec++;
|
||
} elseif ($sec == 6 && $pos == 25) { // 06,25-06,32
|
||
$pos = 33;
|
||
} elseif ($sec == 6 && $pos == 57) { // 06,57-06,94
|
||
$pos = 1;
|
||
$sec++;
|
||
} elseif ($sec == 7 && $pos == 34) { // 07,34-07,48
|
||
$pos = 49;
|
||
} elseif ($sec == 7 && $pos == 82) { // 07,82-07,94
|
||
$pos = 1;
|
||
$sec++;
|
||
} elseif ($sec == 8 && $pos == 27) { // 08,27-08,36
|
||
$pos = 37;
|
||
} elseif ($sec == 8 && $pos == 74) { // 08,74-09,03
|
||
$pos = 4;
|
||
$sec++;
|
||
} elseif ($sec == 9 && $pos == 80) { // 09,80-09,94
|
||
break;
|
||
}
|
||
// 不用的一級字
|
||
if ($sec == 55 && $pos == 90) {
|
||
break;
|
||
}
|
||
}
|
||
return $chars;
|
||
}
|
||
|
||
// gb2312_punc_chars: Return the GB2312 punctuation characters
|
||
function gb2312_punc_chars()
|
||
{
|
||
init_gb2312();
|
||
return $GLOBALS["_GB2312"]["符號"];
|
||
}
|
||
|
||
// gb2312_feq_chars: Return the GB2312 frequently-used characters
|
||
function gb2312_feq_chars()
|
||
{
|
||
init_gb2312();
|
||
return $GLOBALS["_GB2312"]["一級字"];
|
||
}
|
||
|
||
// gb2312_nonfeq_chars: Return the GB2312 non-frequently-used characters
|
||
function gb2312_nonfeq_chars()
|
||
{
|
||
init_gb2312();
|
||
return $GLOBALS["_GB2312"]["二級字"];
|
||
}
|
||
|
||
?>
|