宜配屋

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。

返回列表

上一篇：Python字符转换

下一篇：PHP生成静态页面详解

相关文章

PHP MemCached高级缓存配置图文教程

PHP MemCached高级缓存配置图文教程

1.Memcache相关介绍 memcache是一个高性能的分布式的内存对象缓存系统，它能够用来存储各种格式的数据，包括图像、视频、文件以及数据库检索的结果等。 使用Memcache的网...

php连接函数implode与分割explode的深入解析

目前学习php的人有很多，很多进行php培训的朋友在学习中总会问到这样一个问题:php连接函数implode是什么呢？php可以将字符串分割成数组，同时翻过了也可以将数组连接成字符串，确...

PHP实现的简单操作SQLite数据库类与用法示例

本文实例讲述了PHP实现的简单操作SQLite数据库类与用法。分享给大家供大家参考，具体如下： SQLite是一款轻型的数据库，是遵守ACID的关联式数据库管理系统，它的设计目标是嵌入式...

PHP实现手机号码中间四位用星号（*）隐藏的自定义函数分享

php屏蔽电话号码中间四位： 复制代码 代码如下: Method 1： function hidtel($phone){     $IsWhat...

PHP利用正则表达式将相对路径转成绝对路径的方法示例

前言 大家应该都有所体会，很多时候在做网络爬虫的时候特别需要将爬虫搜索到的超链接进行处理，统一都改成绝对路径的，所以本文就写了一个正则表达式来对搜索到的链接进行处理。下面话不多说，来看看...

© YiPeiWu.com 【宜配屋】 粤ICP备17031333号

Powered By Z-BlogPHP. Theme by TOYEAN.