Project

General

Profile

« Previous | Next » 

Revision 1666

Added by Aaron Marcuse-Kubitza almost 13 years ago

bin/tapir/tapir2flat.php: Fixed XML parsing to strip control chars so DOMDocument::loadXML() wouldn't complain about "PCDATA invalid Char value 8 in Entity", etc.

View differences:

bin/tapir/tapir2flat.php
34 34 
$record_num_filename = "recordnum.dat";
35 35 
$error_log_filename = "error.log";
36 36 

  		




  
  37
  
    
$ctrl_chars = array_flip(array_map("chr", range(0, 31)));


  		




  
  38
  
    
foreach (array("\t", "\n", "\r") as $whitespace)


  		




  
  39
  
    
    unset($ctrl_chars[$whitespace]);


  		




  
  40
  
    
$ctrl_chars = array_keys($ctrl_chars);


  		




  
  41
  
    

  		




  37
  42
  
    
$supportedConcepts = getAllConcepts();


  		




  38
  43
  
    
if(!is_array($supportedConcepts)){


  		




  39
  44
  
    
  $error_log = fopen($error_log_filename,"a");


  		




  ......




  154
  159
  
    
       continue;


  		




  155
  160
  
    
    }


  		




  156
  161
  
    

  		




  157
  
  
    
    #Weird encoding bug - need to remove the ^F and ^L characters so


  		




  158
  
  
    
    #that the xml parser won't choke.


  		




  
  162
  
    
    $response = filter_var(str_replace($ctrl_chars, "", $response),


  		




  
  163
  
    
        FILTER_UNSAFE_RAW, FILTER_FLAG_ENCODE_HIGH);


  		




  159
  164
  
    
    $xmlDoc = new DOMDocument();


  		




  160
  
  
    
    $response = preg_replace("//","6",$response);


  		




  161
  
  
    
    $response = preg_replace("//","12",$response);


  		




  162
  165
  
    
    $xmlDoc->loadXML($response);


  		




  163
  166
  
    

  		




  164
  167
  
    
    $errors = $xmlDoc->getElementsByTagName("error");


  		












Also available in:  Unified diff