打印本文 打印本文  关闭窗口 关闭窗口
提取HTML标签
作者:采集员 文章来源:来源于网络 点击数: 更新时间:2005/9/10 14:35:53
<?php
/*********************************
*
*  作者: 徐祖宁 (唠叨)
*  邮箱: czjsz_ah@stats.gov.cn
*  开发: 2002.07
*
*
*  函数: tags
*  功能: 从文件中提取HTML标签
*
*  入口:
*  $filename 文件名
*  $tag      标签名
*  返回:
*  数组,每项为:
*   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));
  $buffer = eregi_replace(" ","",$buffer);
  $tagkey = sql_regcase($tag);
  $buffer = eregi_replace("<$tagkey "," <$tag ",$buffer);
  $ar = split(" ",$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];
    $s = trim(eregi_replace("[ ]+"," ",$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("



打印本文 打印本文  关闭窗口 关闭窗口