1 <?php
  2 /**
  3  * Base class for XML parsers
  4  * @package sapphire
  5  * @subpackage misc
  6  */
  7 class XML extends Object {
  8     protected $parser;
  9     protected $context, $attributeStore;    
 10     protected $parsed;  
 11     protected $collatedText;
 12     
 13     
 14     function tidyXHTML($content) {
 15         $cleanFile = TEMP_FOLDER . "/cleaner.tmp";
 16         $fh = fopen($cleanFile,"wb");
 17         fwrite($fh, $content);
 18         fclose($fh);
 19         
 20         if(file_exists($cleanFile)) {
 21             $result = `tidy -asxhtml $cleanFile`;
 22             unlink($cleanFile);
 23             return $result;
 24         }
 25     }
 26     
 27     function parse($content, $recursive = false) {
 28         $this->parser = xml_parser_create('UTF-8');
 29         
 30         // Andrew keeps giving me the wrong FSKING encoding! :-P
 31         $content = ereg_replace('encoding="[^"]+"','encoding="utf-8"', $content);
 32         
 33         xml_parser_set_option($this->parser, XML_OPTION_CASE_FOLDING, false);
 34         //xml_parser_set_option($this->parser, XML_OPTION_TARGET_ENCODING, 'UTF-8');
 35         xml_set_object($this->parser, $this);
 36         xml_set_element_handler($this->parser, "tag_open", "tag_close");
 37         xml_set_character_data_handler($this->parser, "cdata");
 38         
 39         $this->parsed = null;
 40         $this->context = array();
 41         $this->attributeStore = array();
 42         
 43         xml_parse($this->parser, $content);
 44         
 45         // Handle a bad encoding type by forcing ISO-8859-15
 46         if(xml_get_error_code($this->parser) == 32 && !$recursive) {
 47             $content = ereg_replace('encoding="[^"]+"','encoding="utf-8"', $content);
 48             return $this->parse($content, true);
 49         }
 50         
 51         if($err = xml_get_error_code($this->parser)) {
 52             user_error("XML parser broke with error $err:" . xml_error_string($err), E_USER_ERROR);
 53         }
 54 
 55         return $this->parsed;
 56     }
 57 
 58     function inContext() {
 59         $items = func_get_args();
 60         $i=0;
 61         foreach($items as $item) {
 62             while($i < sizeof($this->context)) {
 63                 if($this->context[$i] == $item) break;
 64                 $i++;
 65             }
 66             if($this->context[$i] != $item) return false;
 67         }
 68         return true;
 69     }   
 70     
 71     function stackActionFor($tag) {
 72         for($i=sizeof($this->contextStack)-1;$i>=0;$i--) {
 73             if($this->context[$i]['tag'] == $tag) return $this->contextStack[$i]['action'];
 74         }
 75     }
 76     
 77     function tag_open($parser, $tag, $attributes) {
 78         // Strip namespaces out of tags and attributes
 79         $tag = ereg_replace('[^:]+:','',$tag);
 80         if($attributes) foreach($attributes as $k => $v) $newAttributes[ereg_replace('[^:]+:','',$k)] = $v;
 81         $attributes = isset($newAttributes) ? $newAttributes : $attributes;
 82         
 83         
 84         if(isset($attributes['class'])) {
 85             $this->context[] = "$tag.{$attributes['class']}";
 86         } else {
 87             $this->context[] = $tag;
 88         }
 89         $this->attributeStore[] = $attributes;
 90         
 91         $this->collatedText = "";
 92 
 93         $tagProcessorFunc = "process_$tag";
 94         if($this->hasMethod($tagProcessorFunc)) {
 95             $this->$tagProcessorFunc($attributes);
 96         }elseif($this->hasMethod($tagProcessorFunc = "process_tag")){
 97             $this->$tagProcessorFunc($tag, $attributes);    
 98         }
 99         
100         
101         if($attributes) foreach($attributes as $k => $v) {
102             $attProcessorFunc = "processatt_$k";
103             if($this->hasMethod($attProcessorFunc)) {
104                 $this->$attProcessorFunc($tag, $attributes);
105             }
106         }
107     }
108 
109     function tag_close($parser, $tag) {
110         $tag = ereg_replace('[^:]+:','',$tag);
111 
112         array_pop($this->context);
113         $attributes = array_pop($this->attributeStore);
114         
115         if(method_exists($this, $funcName = "process_{$tag}_end")) {
116             $this->$funcName($this->collatedText, $attributes);
117         }elseif(method_exists($this,$funcName = "process_tag_end")){ 
118             // else run default method
119                 $this->$funcName($tag,$this->collatedText, $attributes);    
120         }
121         
122         $this->collatedText = "";
123     }
124     
125     function cdata($parser, $cdata) {
126         $this->collatedText .= $cdata;
127     }
128 }