php正则替换处理HTML页面的方法

yipeiwu_com5年前PHP代码库

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:

<?php
if(!defined('BASEPATH')) exit('No direct script access allowed');
 /**
 * HTML替换处理类,考虑如下几种替换
 * 1. img src : '/<img(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'
 * 2. a href : '/<a(.+?)href=([\'\" ])?(.+?)([ >]+?)/i'
 * 3. ifram.src : '/<iframe(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'
 * 4. frame src : '/<frame(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'
 * 5. js : '/window.open([( ]+?)([\'" ]+?)(.+?)([ )+?])/i'
 * 6. css : '/background(.+?)url([( ])([\'" ]+?)(.+?)([ )+?])/i'
 */
 class Myreplace {
 private $moudle_array = array('udata','tdata','tresult','dresult');
 private $content;
 private $relative_dirname;
 private $projectid;
 private $moudle;
 function __construct() {
  $this->CI = &get_instance ();
 }
 /**
  * 替换
  * @param string $content HTML内容
  * @param string $relative 相对路径
  * @param int $projectid 项目id
  * @moudle string $moudle 模板标识: udata,tdata,tresult,dresult
  */
 public function my_replace($content,$relative,$projectid,$moudle) {
  $this->content = $content;
  $this->relative_dirname = $relative;
  $this->projectid = $projectid;
  if(in_array(strtolower($moudle),$this->moudle_array))
  $this->moudle = $moudle;
  else exit;
  switch($this->moudle) {
  case 'udata':
   $this->CI->load->model('mupload_data','model');
   break;
  case 'tdata':
   $this->CI->load->model('taskdata','model');
   break;
  case 'tresult':
   $this->CI->load->model('taskresult','model');
   break;
  case 'dresult':
   $this->CI->load->model('dmsresult','model');
   break;
  default:
   break;
  }
  $pattern = '/<img(.+?)src=([\'\" ])?(.+?)([ >]+?)/i';
  $content = preg_replace_callback( $pattern, array($this, 'image_replace') , $content );
  $pattern = '/<a(.+?)href=([\'\" ])?(.+?)([ >]+?)/i';
  $content = preg_replace_callback( $pattern, array($this, 'html_replace') , $content );
  $pattern = '/<iframe(.+?)src=([\'\" ])?(.+?)([ >]+?)/i';
  $content = preg_replace_callback( $pattern, array($this, 'iframe_replace') , $content );
  $pattern = '/<frame(.+?)src=([\'\" ])?(.+?)([ >]+?)/i'; 
  $content = preg_replace_callback( $pattern, array($this, 'frame_replace'), $content );
  $pattern = '/window.open([( ]+?)([\'" ]+?)(.+?)([ )]+?)/i';
  $content = preg_replace_callback( $pattern, array($this, 'js_replace'), $content );
  $pattern = '/background(.+?)url([( ])([\'" ]+?)(.+?)([ )+?])/i';
  $content = preg_replace_callback( $pattern, array($this, 'css_replace'), $content);
  return $content;
 }
 private function image_replace($matches) {
  if(count($matches) < 4) return '';
  if( empty($matches[3]) ) return '';
  $matches[3] = rtrim($matches[3],'\'"/');
  //获取图片的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],'./');
  $image_id = $this->CI->model->get_id_by_path_and_project($relativepath,$this->projectid);
  //输出
  if( !empty($image_id) ) {
  if($this->moudle == 'dresult') {
   return "<img".$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。

相关文章

php下使用curl模拟用户登陆的代码

bool curl_setopt (int ch, string option, mixed value) curl_setopt()函数将为一个CURL会话设置选项。option参数是...

PHP Swoole异步读取、写入文件操作示例

PHP Swoole异步读取、写入文件操作示例

本文实例讲述了PHP Swoole异步读取、写入文件操作。分享给大家供大家参考,具体如下: 异步读取文件:swoole_async_readfile 异步写入文件:swoole_asyn...

php简单生成一组与多组随机字符串的方法

本文实例讲述了php简单生成随机字符串的方法。分享给大家供大家参考,具体如下: 生成一组: <?php $str = "0123456789abcdefghijklmn...

浏览器预览PHP文件时顶部出现空白影响布局分析原因及解决办法

在编写PHP文件过程中,发现在浏览器预览PHP文件时,顶部会出现一行空白,影响了页面的布局。 关于BOM header的解释如下: 通常情况下,使用Windows系统自带的记事本程序编写...

深入分析PHP优化及注意事项

1.尽量静态化: 如果一个方法能被静态,那就声明它为静态的,速度可提高1/4,甚至我测试的时候,这个提高了近三倍。 当然了,这个测试方法需要在十万级以上次执行,效果才明显。 其实静态...