首页
所有分类
TAGS
文字工具 EMOJI BIBLE
编程 PHP

php正则替换处理HTML页面的方法

Posted in PHP onJune 17, 2015

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 */

php正则替换处理HTML页面的方法

- Author -

邪恶的小Y

声明：登载此文出于传递更多信息之目的，并不意味着赞同其观点或证实其描述。

PHP 相关文章推荐
手把手教你使用DedeCms V3的在线采集图文教程
Apr 03 PHP
PHP 批量更新网页内容实现代码
Jan 05 PHP
php提示Call-time pass-by-reference has been deprecated in的解决方法[已测]
May 06 PHP
遭遇php的in_array低性能问题
Sep 17 PHP
7个鲜为人知却非常实用的PHP函数
Jul 01 PHP
在Debian系统下配置LNMP的教程
Jul 09 PHP
PHP去掉json字符串中的反斜杠\及去掉双引号前的反斜杠
Sep 30 PHP
PHP培训要多少钱
Jun 06 PHP
php封装的mongodb操作类代码
Aug 06 PHP
在Laravel的Model层做数据缓存的实现
Sep 26 PHP
php连接mysql之mysql_connect()与mysqli_connect()的区别
Jul 19 PHP
PHP基于phpqrcode类生成二维码的方法示例详解
Aug 07 PHP
PHP排序算法类实例
Jun 17 #PHP
php准确获取文件MIME类型的方法
Jun 17 #PHP
php操作MongoDB类实例
Jun 17 #PHP
PHP实现的购物车类实例
Jun 17 #PHP
CodeIgniter实现从网站抓取图片并自动下载到文件夹里的方法
Jun 17 #PHP
PHP基于MySQL数据库实现对象持久层的方法
Jun 17 #PHP
php使用curl打开https网站的方法
Jun 17 #PHP
进度条(1) 鲲鹏(1) Flask(2) Python(3) 死锁(3) reduce()(1) 生命周期(1) 裁剪图片(1) 信号同步(1) PIL(1)
You might like
php截取字符串之截取utf8或gbk编码的中英文字符串示例
2014/03/12 PHP
php中unserialize返回false的解决方法
2014/09/22 PHP
WordPress的文章自动添加关键词及关键词的SEO优化
2016/03/01 PHP
php实现图片以base64显示的方法
2016/10/13 PHP
php简单构造json多维数组的方法示例
2017/06/08 PHP
PHP后台实现微信小程序登录
2018/08/03 PHP
php统计数组不同元素的个数的实例方法
2019/09/26 PHP
jQuery 入门讲解1
2009/04/15 Javascript
JQuery跨Iframe选择实现代码
2010/08/19 Javascript
浅谈javascript中字符串String与数组Array
2014/12/31 Javascript
jquery插件Jplayer使用方法简析
2016/04/22 Javascript
JavaScript中style.left与offsetLeft的使用及区别详解
2016/06/08 Javascript
浅谈json取值(对象和数组)
2016/06/24 Javascript
JS实现添加,替换,删除节点元素的方法
2016/06/30 Javascript
AngularJS入门教程之双向绑定详解
2016/08/18 Javascript
深入浅析JavaScript中的3DES
2016/08/24 Javascript
AngularJS轻松实现双击排序的功能
2016/08/30 Javascript
jQuery实现的放大镜效果示例
2016/09/13 Javascript
vue.js中Vue-router 2.0基础实践教程
2017/05/08 Javascript
es7学习教程之Decorators（修饰器）详解
2017/07/21 Javascript
JS中touchstart事件与click事件冲突的解决方法
2018/03/12 Javascript
element-ui 表格数据时间格式化的方法
2018/08/24 Javascript
QRCode.js二维码生成并能长按识别
2018/10/16 Javascript
详解angularjs跨页面传参遇到的一些问题
2018/11/01 Javascript
uni-app 组件里面获取元素宽高的实现
2019/12/27 Javascript
vue-cli4使用全局less文件中的变量配置操作
2020/10/21 Javascript
在实例中重学JavaScript事件循环
2020/12/03 Javascript
python基础之包的导入和__init__.py的介绍
2018/01/08 Python
已安装tensorflow-gpu,但keras无法使用GPU加速的解决
2020/02/07 Python
django列表筛选功能的实现代码
2020/03/27 Python
python文件排序的方法总结
2020/09/13 Python
css3的transition效果和transfor效果示例介绍
2013/10/30 HTML / CSS
CSS3实现点击放大的动画实例代码
2017/02/27 HTML / CSS
趣味运动会活动方案
2014/02/12 职场文书
工作态度怎么写
2015/06/25 职场文书
python中pd.cut()与pd.qcut()的对比及示例
2022/06/16 Python