$ filename =
Http://smsrc.sina.com.cn/newsinarc/include/gb2utf8.txt
// Use of the comparison table
$ filename = "GB2UTF8.TXT";
$ fp = fopen ($ filename, "r");
While (! Feof ($ fp)) {
List ($ GB, $ UTF8) = FGETCSV ($ FP, 10);
$ Charset [$ GB] = $ UTF8;
}
Fclose ($ fp);
// Read the comparison table to the array
/ ** GB2312 to UTF-8 ** /
Function GB2UTF8 ($ TEXT, & $ Charset) {
/ / Extract the ingredients in the text, the Chinese characters are an element, and the continuous non-Chinese characters are an element.
PREG_MATCH_ALL ("/ (?: [/ x7f] /", $ text, $ TMP);
$ TMP = $ TMP [0];
// Separate Chinese characters
$ ar = array_intersect ($ TMP, Array_Keys ($ Charset);
// Replace Chinese character encoding
Foreach ($ AS $ K => $ V)
$ TMP [$ K] = $ Charset [$ V];
// Return to the changed string
Return Join ('', $ TMP);
}
/ ** UTF-8 to 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] = $ UTF8 [$ V];
Return Join ('', $ R [0]);
}
//test
$ S = GB2UTF8 ('This is the test of the control ", $ Charset);
Echo UTF82GB ($ S, $ CHARSET);
?>
