当前位置：Gxlcms > PHP教程 > php正则替换处理HTML页面的方法

php正则替换处理HTML页面的方法

时间：2021-07-01 10:21:17 帮助过：4人阅读

本文实例讲述了php正则替换处理HTML页面的方法。分享给大家供大家参考。具体如下：

  1. <?php
  2. if(!defined('BASEPATH')) exit('No direct script access allowed');
  3.  /**
  4.  * HTML替换处理类，考虑如下几种替换
  5.  * 1. img src : '/<img(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'
  6.  * 2. a href : '/<a(.+?)href=([\'\" ])?(.+?)([ >]+?)/i'
  7.  * 3. ifram.src : '/<iframe(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'
  8.  * 4. frame src : '/<frame(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'
  9.  * 5. js : '/window.open([( ]+?)([\'" ]+?)(.+?)([ )+?])/i'
  10.  * 6. css : '/background(.+?)url([( ])([\'" ]+?)(.+?)([ )+?])/i'
  11.  */
  12.  class Myreplace {
  13.  private $moudle_array = array('udata','tdata','tresult','dresult');
  14.  private $content;
  15.  private $relative_dirname;
  16.  private $projectid;
  17.  private $moudle;
  18.  function __construct() {
  19.   $this->CI = &get_instance ();
  20.  }
  21.  /**
  22.   * 替换
  23.   * @param string $content HTML内容
  24.   * @param string $relative 相对路径
  25.   * @param int $projectid 项目id
  26.   * @moudle string $moudle 模板标识: udata,tdata,tresult,dresult
  27.   */
  28.  public function my_replace($content,$relative,$projectid,$moudle) {
  29.   $this->content = $content;
  30.   $this->relative_dirname = $relative;
  31.   $this->projectid = $projectid;
  32.   if(in_array(strtolower($moudle),$this->moudle_array))
  33.   $this->moudle = $moudle;
  34.   else exit;
  35.   switch($this->moudle) {
  36.   case 'udata':
  37.    $this->CI->load->model('mupload_data','model');
  38.    break;
  39.   case 'tdata':
  40.    $this->CI->load->model('taskdata','model');
  41.    break;
  42.   case 'tresult':
  43.    $this->CI->load->model('taskresult','model');
  44.    break;
  45.   case 'dresult':
  46.    $this->CI->load->model('dmsresult','model');
  47.    break;
  48.   default:
  49.    break;
  50.   }
  51.   $pattern = '/<img(.+?)src=([\'\" ])?(.+?)([ >]+?)/i';
  52.   $content = preg_replace_callback( $pattern, array($this, 'image_replace') , $content );
  53.   $pattern = '/<a(.+?)href=([\'\" ])?(.+?)([ >]+?)/i';
  54.   $content = preg_replace_callback( $pattern, array($this, 'html_replace') , $content );
  55.   $pattern = '/<iframe(.+?)src=([\'\" ])?(.+?)([ >]+?)/i';
  56.   $content = preg_replace_callback( $pattern, array($this, 'iframe_replace') , $content );
  57.   $pattern = '/<frame(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'; 
  58.   $content = preg_replace_callback( $pattern, array($this, 'frame_replace'), $content );
  59.   $pattern = '/window.open([( ]+?)([\'" ]+?)(.+?)([ )]+?)/i';
  60.   $content = preg_replace_callback( $pattern, array($this, 'js_replace'), $content );
  61.   $pattern = '/background(.+?)url([( ])([\'" ]+?)(.+?)([ )+?])/i';
  62.   $content = preg_replace_callback( $pattern, array($this, 'css_replace'), $content);
  63.   return $content;
  64.  }
  65.  private function image_replace($matches) {
  66.   if(count($matches) < 4) return '';
  67.   if( empty($matches[3]) ) return '';
  68.   $matches[3] = rtrim($matches[3],'\'"/');
  69.   //获取图片的id
  70.   $parent_dir_num = substr_count( $matches[3], '../');
  71.   $relative_dirname = $this->relative_dirname;
  72.   for($i=0; $i<$parent_dir_num; $i++) {
  73.   $relative_dirname = substr( $relative_dirname, 0, strrpos($relative_dirname,"/") );
  74.   }
  75.   $relativepath = rtrim($relative_dirname,'/') . '/'.ltrim($matches[3],'./');
  76.   $image_id = $this->CI->model->get_id_by_path_and_project($relativepath,$this->projectid);
  77.   //
输出 if( !empty($image_id) ) { if($this->moudle == 'dresult') { return "<img".$matches[1]."src=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/readpic/$image_id?pid=".$this->projectid .$matches[2]. $matches[4]; } else { return "<img".$matches[1]."src=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/picfile/$image_id?pid=".$this->projectid .$matches[2]. $matches[4]; } } else { return "<img".$matches[1]."src=".$matches[2].$matches[3].$matches[2].$matches[4]; } } private function html_replace( $matches ) { if(count($matches) < 4) return ''; if( empty($matches[3]) ) return ''; //如果href的链接($matches[3])以http或www或mailto开始，则不进行处理 //if(preg_match('/^[http|www|mailto](.+?)/i',$matches[3])) // return "<a".$matches[1]."href=".$matches[2].$matches[3].$matches[4]; $matches[3] = rtrim($matches[3],'\'"/'); //处理锚点 if(substr_count($matches[3],'#')>0) $matches[3] = substr($matches[3],0,strrpos($matches[3],'#')); //获取html的id $parent_dir_num = substr_count( $matches[3], '../'); $relative_dirname = $this->relative_dirname; for($i=0; $i<$parent_dir_num; $i++) { $relative_dirname = substr( $relative_dirname, 0, strrpos($relative_dirname,"/") ); } $relativepath = rtrim($relative_dirname,'/') . '/'.ltrim($matches[3],'./'); $txtfile_id = $this->CI->model->get_id_by_path_and_project($relativepath,$this->projectid); //输出 if( !empty($txtfile_id ) ) { if($this->moudle == 'dresult') { return "<a".$matches[1]."href=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/readfile/$txtfile_id?pid=".$this->projectid .$matches[2].$matches[4]; } else { return "<a".$matches[1]."href=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/txtfile/$txtfile_id?pid=".$this->projectid .$matches[2].$matches[4]; } } else { return "<a".$matches[1]."href=".$matches[2].$matches[3].$matches[2].$matches[4]; } } private function iframe_replace( $matches ) { if(count($matches) < 4) return ''; if( empty($matches[3]) ) return ''; $matches[3] = rtrim($matches[3],'\'"/'); //处理锚点 if(substr_count($matches[3],'#')>0) $matches[3] = substr($matches[3],0,strrpos($matches[3],'#')); //获取html的id $parent_dir_num = substr_count( $matches[3], '../'); $relative_dirname = $this->relative_dirname; for($i=0; $i<$parent_dir_num; $i++) { $relative_dirname = substr( $relative_dirname, 0, strrpos($relative_dirname,"/") ); } $relativepath = rtrim($relative_dirname,'/') . '/'.ltrim($matches[3],'./'); $txtfile_id = $this->CI->model->get_id_by_path_and_project($relativepath,$this->projectid); //输出 if( !empty($txtfile_id ) ) { if($this->moudle == 'dresult') { return "<iframe".$matches[1]."src=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/readfile/$txtfile_id?pid=".$this->projectid .$matches[2].$matches[4]; } else { return "<iframe".$matches[1]."src=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/txtfile/$txtfile_id?pid=".$this->projectid .$matches[2].$matches[4]; } } else { return "<iframe".$matches[1]."src=".$matches[2].$matches[3].$matches[2].$matches[4]; } } private function frame_replace( $matches ) { if(count($matches) < 4) return ''; if( empty($matches[3]) ) return ''; $matches[3] = rtrim($matches[3],'\'"/'); //处理锚点 if(substr_count($matches[3],'#')>0) $matches[3] = substr($matches[3],0,strrpos($matches[3],'#')); //获取html的id $parent_dir_num = substr_count( $matches[3], '../'); $relative_dirname = $this->relative_dirname; for($i=0; $i<$parent_dir_num; $i++) { $relative_dirname = substr( $relative_dirname, 0, strrpos($relative_dirname,"/") ); } $relativepath = rtrim($relative_dirname,'/') . '/'.ltrim($matches[3],'./'); $txtfile_id = $this->CI->model->get_id_by_path_and_project($relativepath,$this->projectid); //输出 if( !empty($txtfile_id ) ) { if($this->moudle == 'dresult') { return "<frame".$matches[1]."src=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/readfile/$txtfile_id?pid=".$this->projectid.$matches[2].$matches[4]; } else { return "<frame".$matches[1]."src=".$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/txtfile/$txtfile_id?pid=".$this->projectid.$matches[2].$matches[4]; } } else { return "<frame".$matches[1]."src=".$matches[2].$matches[3].$matches[2].$matches[4]; } } private function js_replace( $matches ){ if(count($matches) < 4) return ''; if( empty($matches[3]) ) return ''; //处理链接 $arr_html = split(',',$matches[3]); $href = $arr_html[0]; $other = ''; for($i=0; $i<count($arr_html); $i++) $other = $arr_html[$i].", "; $other = rtrim($other,"\, "); $href =rtrim($href,'\'\"'); //处理锚点 if(substr_count($href,'#')>0) return "window.open".$matches[1].$matches[2].$matches[3].$matches[4];; //获取html的id $parent_dir_num = substr_count( $href, '../'); $relative_dirname = $this->relative_dirname; for($i=0; $i<$parent_dir_num; $i++) { $relative_dirname = substr( $relative_dirname, 0, strrpos($relative_dirname,"/") ); } $relativepath = rtrim($relative_dirname,'/') . '/'.ltrim($href,'./'); $txtfile_id = $this->CI->model->get_id_by_path_and_project($relativepath,$this->projectid); //输出 if( !empty($txtfile_id ) ) { if($this->moudle == 'dresult') { return "window.open".$matches[1].$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/readfile/$txtfile_id?pid=".$this->projectid.$matches[2].','.$other.$matches[4]; } else { return "window.open".$matches[1].$matches[2].$this->CI->config->item("base_url")."cdms/".$this->moudle."/txtfile/$txtfile_id?pid=".$this->projectid.$matches[2].','.$other.$matches[4]; } } else { return "window.open".$matches[1].$matches[2].$matches[3].$matches[4]; } } private function css_replace( $matches ) { if(count($matches) < 5) return ''; if( empty($matches[4]) ) return ''; $matches[4] = rtrim($matches[4],'\'"/'); //获取图片的id $parent_dir_num = substr_count( $matches[4], '../'); $relative_dirname = $this->relative_dirname; for($i=0; $i<$parent_dir_num; $i++) { $relative_dirname = substr( $relative_dirname, 0, strrpos($relative_dirname,"/") ); } $relativepath = rtrim($relative_dirname,'/') . '/'.ltrim($matches[4],'./'); $image_id = $this->CI->model->get_id_by_path_and_project($relativepath,$this->projectid); //输出 if( !empty($image_id) ) { if($this->moudle == 'dresult') { return "background".$matches[1]."url".$matches[2].$matches[3].$this->CI->config->item("base_url")."cdms/".$this->moudle."/readpic/$image_id?pid=".$this->projectid .$matches[3]. $matches[5]; } else { return "background".$matches[1]."url".$matches[2].$matches[3].$this->CI->config->item("base_url")."cdms/".$this->moudle."/picfile/$image_id?pid=".$this->projectid .$matches[3]. $matches[5]; } } else { return "background".$matches[1]."url".$matches[2].$matches[3].$matches[4].$matches[3].$matches[5]; } } } /* End of Myreplace.php */ /* Location: /application/libraries/Myreplace.php */

PS：这里再为大家提供2款非常方便的正则表达式工具供大家参考使用：

JavaScript正则表达式在线测试工具：
http://tools.jb51.net/regex/javascript

正则表达式在线生成工具：
http://tools.jb51.net/regex/create_reg

希望本文所述对大家的php程序设计有所帮助。

人气教程排行

本站所有资源全部来源于网络，若本站发布的内容侵害到您的隐私或者利益，请联系我们删除！