具体实现代码如下:
<?php
/**
* 提取页面中的 html标签的代码
* 示例:
* print_r(tags("test1.htm","a"));
* print_r("http://localhost/index.htm","img");
*
*/
function tags($filename,$tag) {
$buffer = join("",file($filename));
$buffer = eregi_replace("rn","",$buffer);
$tagkey = sql_regcase($tag);
$buffer = eregi_replace("<$tagkey ","n<$tag ",$buffer);
$ar = split("n",$buffer);
foreach($ar as $v) {
if(! eregi("<$tagkey ",$v)) continue;
eregi("<$tagkey ([^>]*)((.*)</$tagkey)?",$v,$regs);
$p[tagName] = strtoupper($tag);
if($regs[3])
$p[Text] = $regs[3];
$s = trim(eregi_replace("[ t]+"," ",$regs[1]))." ";
$s = eregi_replace(" *= *","=",$s);
$a = split(" ",$s);
for($i=0;$i<count($a);$i++) {
$ch = array();
if(eregi("=["']",$a[$i])) {
$j = $i+1;
while(!eregi("["']$",$a[$i])) {
$a[$i] .= " ".$a[$j];
unset($a[$j]);
}
}
}
foreach($a as $k) {
$name = strtoupper(strtok($k,"="));
$value = strtok("