-
-
//对照表的使用 - $filename = "gb2utf8.txt";
- $fp = fopen($filename,"r");
- while(! feof($fp)) {
- list($gb,$utf8) = fgetcsv($fp,10);
- $charset[$gb] = $utf8;
- }
- fclose($fp);
- //以上读取对照表到数组备用
/** gb2312到utf-8 **/ - function gb2utf8($text, &$charset) {
- //提取文本中的成分，汉字为一个元素，连续的非汉字为一个元素
- preg_match_all("/(?:[\x80-\xff].)|[\x01-\x7f]+/",$text,$tmp);
- $tmp = $tmp[0];
- //分离出汉字
- $ar = array_intersect($tmp, array_keys($charset));
- //替换汉字编码
- foreach($ar as $k=>$v)
- $tmp[$k] = $charset[$v];
- //返回换码后的串
- return join('',$tmp);
- }
/** utf-8到gb2312 **/ - function utf82gb($text, &$charset) {
- $p = "/[xf0-xf7][x80-xbf]{3}|[xe0-xef][x80-xbf]{2}|[xc2-xdf][x80-xbf]|[x01-x7f]+/";
- preg_match_all($p,$text,$r);
- $utf8 = array_flip($charset);
- foreach($r[0] as $k=>$v)
- if(isset($utf8[$v]))
- $r[0][$k] = $utf8[$v];
- return join('',$r[0]);
- }
//测试 - $s = gb2utf8('这是对照表的测试', $charset);
- echo utf82gb($s, $charset);
- ?>
-
|
