php正则替换处理HTML页面的方法_php技巧-php教程-PHP中文網

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：
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
<&#63;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="/static/imghw/default1.png"  data-src=".$matches[2].$this- alt="php正则替换处理HTML页面的方法_php技巧" >CI->config->item("  class="lazy" (.+&#63;)src=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i'
  * 2. a href : '/<a(.+&#63;)href=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i'
  * 3. ifram.src : '/<iframe(.+&#63;)src=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i'
  * 4. frame src : '/<frame(.+&#63;)src=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i'
  * 5. js  : '/window.open([( ]+&#63;)([\'" ]+&#63;)(.+&#63;)([ )+&#63;])/i'
  * 6. css  : '/background(.+&#63;)url([( ])([\'" ]+&#63;)(.+&#63;)([ )+&#63;])/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;
   }
   $pattern = '/<img  src="/static/imghw/default1.png"  data-src=".$matches[2].$this- alt="php正则替换处理HTML页面的方法_php技巧" >CI->config->item("  class="lazy" (.+&#63;)src=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i';
   $content = preg_replace_callback( $pattern, array($this, 'image_replace') , $content );
   $pattern = '/<a(.+&#63;)href=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i';
   $content = preg_replace_callback( $pattern, array($this, 'html_replace') , $content );
   $pattern = '/<iframe(.+&#63;)src=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i';
   $content = preg_replace_callback( $pattern, array($this, 'iframe_replace') , $content );
   $pattern = '/<frame(.+&#63;)src=([\'\" ])&#63;(.+&#63;)([ >]+&#63;)/i'; 
   $content = preg_replace_callback( $pattern, array($this, 'frame_replace'), $content );
   $pattern = '/window.open([( ]+&#63;)([\'" ]+&#63;)(.+&#63;)([ )]+&#63;)/i';
   $content = preg_replace_callback( $pattern, array($this, 'js_replace'), $content );
   $pattern = '/background(.+&#63;)url([( ])([\'" ]+&#63;)(.+&#63;)([ )+&#63;])/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($this->moudle == 'dresult') {
     return "<img  src="/static/imghw/default1.png"  data-src=".$matches[2].$this- alt="php正则替换处理HTML页面的方法_php技巧" >CI->config->item("  class="lazy" ".$matches[1]."base_url")."cdms/".$this->moudle."/readpic/$image_id&#63;pid=".$this->projectid .$matches[2]. $matches[4];
    } else {
     return "<img  src="/static/imghw/default1.png"  data-src=".$matches[2].$this- alt="php正则替换处理HTML页面的方法_php技巧" >CI->config->item("  class="lazy" ".$matches[1]."base_url")."cdms/".$this->moudle."/picfile/$image_id&#63;pid=".$this->projectid .$matches[2]. $matches[4];
    }
   } else {
    return "<img  src="/static/imghw/default1.png"  data-src=".$matches[2].$this- alt="php正则替换处理HTML页面的方法_php技巧" >CI->config->item("  class="lazy" ".$matches[1]."';
   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](.+&#63;)/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&#63;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&#63;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]."base_url")."cdms/".$this->moudle."/readfile/$txtfile_id&#63;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&#63;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&#63;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&#63;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&#63;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/$txtfile_id&#63;pid=".$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/$image_id&#63;pid=".$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/$image_id&#63;pid=".$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 */
登入後複製
希望本文所述对大家的php程序设计有所帮助。