提取HTML標簽

2019-11-18 20:18:39

字體：大中小

來源：轉載

供稿：網友

<?php
/*********************************
*
*  作者: 徐祖寧 (嘮叨)
*  郵箱: czjsz_ah@stats.gov.cn
*  開發: 2002.07
*
*
*  函數: tags
*  功能: 從文件中提取HTML標簽
*
*  入口:
*  $filename 文件名
*  $tag      標簽名
*  返回:
*  數組,每項為:
*   tagName    String
*   Text       String
*   Attrs      Array
*
*  示例:
*  PRint_r(tags("test1.htm","a"));
*  print_r("http://localhost/index.htm","img");
*
*/

function tags($filename,$tag) {
  $buffer = join("",file($filename));
  $buffer = eregi_replace("/r/n","",$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,"="));
      $value = strtok("/0");
      if(eregi("^[/"']",$value))
        $value = substr($value,1,-1);
      if($name)
        $p[Attrs][$name] = $value;
    }
    $pp[] = $p;
  }
  return $pp;
}

?>

上一篇：輸出控制類

下一篇：提供一種“間接防止另存為”的方法