汉字转Unicode编码，Unicode编码转汉字

           

<?php

/**
 * 汉字转Unicode编码
 * @param string $str 原始汉字的字符串
 * @param string $encoding 原始汉字的编码
 * @param boot $ishex 是否为十六进制表示（支持十六进制和十进制）
 * @param string $prefix 编码后的前缀
 * @param string $postfix 编码后的后缀
 */
function unicode_encode($str, $encoding = &#39;UTF-8&#39;, $ishex = false, $prefix = &#39;&#&#39;, $postfix = &#39;;&#39;) {
	$str = iconv($encoding, &#39;UCS-2&#39;, $str);
	$arrstr = str_split($str, 2);
	$unistr = &#39;&#39;;
	for($i = 0, $len = count($arrstr); $i < $len; $i++) {
		$dec = $ishex ? bin2hex($arrstr[$i]) : hexdec(bin2hex($arrstr[$i]));
		$unistr .= $prefix . $dec . $postfix;
	}
	return $unistr;
}

/**
 * Unicode编码转汉字
 * @param string $str Unicode编码的字符串
 * @param string $decoding 原始汉字的编码
 * @param boot $ishex 是否为十六进制表示（支持十六进制和十进制）
 * @param string $prefix 编码后的前缀
 * @param string $postfix 编码后的后缀
 */
function unicode_decode($unistr, $encoding = &#39;UTF-8&#39;, $ishex = false, $prefix = &#39;&#&#39;, $postfix = &#39;;&#39;) {
	$arruni = explode($prefix, $unistr);
	$unistr = &#39;&#39;;
	for($i = 1, $len = count($arruni); $i < $len; $i++) {
		if (strlen($postfix) > 0) {
			$arruni[$i] = substr($arruni[$i], 0, strlen($arruni[$i]) - strlen($postfix));
		}
		$temp = $ishex ? hexdec($arruni[$i]) : intval($arruni[$i]);
		$unistr .= ($temp < 256) ? chr(0) . chr($temp) : chr($temp / 256) . chr($temp % 256);
	}
	return iconv(&#39;UCS-2&#39;, $encoding, $unistr);
}

header(&#39;Content-Type: text/html; charset=UTF-8&#39;);

// UTF-8字符串测试
$str = &#39;龕龖龗龘龙龚龛龜龝龞龟龠龡龢龣龤龥&#39;;
var_dump($str);

// 简单的
$uni_str = mb_convert_encoding($str, &#39;HTML-ENTITIES&#39;, &#39;UTF-8&#39;);
var_dump($uni_str);

$str3 = mb_convert_encoding($uni_str, &#39;UTF-8&#39;, &#39;HTML-ENTITIES&#39;);
var_dump($str3);

$uni_str = unicode_encode($str);
var_dump($uni_str); // 龕龖龗龘龙龚龛龜龝龞龟龠龡龢龣龤龥

$str2 = unicode_decode($uni_str);
var_dump($str2); // 龕龖龗龘龙龚龛龜龝龞龟龠龡龢龣龤龥

$uni_str = unicode_encode($str, &#39;UTF-8&#39;, true, &#39;\u&#39;, &#39;&#39;);
var_dump($uni_str); // \u9f95\u9f96\u9f97\u9f98\u9f99\u9f9a\u9f9b\u9f9c\u9f9d\u9f9e\u9f9f\u9fa0\u9fa1\u9fa2\u9fa3\u9fa4\u9fa5

$str2 = unicode_decode($uni_str, &#39;UTF-8&#39;, true, &#39;\u&#39;, &#39;&#39;);
var_dump($str2); // 龕龖龗龘龙龚龛龜龝龞龟龠龡龢龣龤龥


// GBK字符串测试
$str = &#39;PHP汉字转UNICODE&#39;;

$str = iconv(&#39;UTF-8&#39;, &#39;GBK//IGNORE&#39;, $str);
$uni_str = unicode_encode($str, &#39;GBK&#39;);
var_dump($uni_str); // PHP汉字转UNICODE

$str2 = unicode_decode($uni_str, &#39;GBK&#39;);
$str2 = iconv(&#39;GBK&#39;, &#39;UTF-8&#39;, $str2);
var_dump($str2); // PHP汉字转UNICODE

$uni_str = unicode_encode($str, &#39;GBK&#39;, true, &#39;\u&#39;, &#39;&#39;);
var_dump($uni_str); // \u0050\u0048\u0050\u6c49\u5b57\u8f6c\u0055\u004e\u0049\u0043\u004f\u0044\u0045

$str2 = unicode_decode($uni_str, &#39;GBK&#39;, true, &#39;\u&#39;, &#39;&#39;);
$str2 = iconv(&#39;GBK&#39;, &#39;UTF-8&#39;, $str2);
var_dump($str2); // PHP汉字转UNICODE