includes/ParserXML.php

   1 <?
   2 require_once ( "Parser.php" ) ;
   3
   4 /**
   5  * This should one day become the XML->(X)HTML parser
   6  * Based on work by Jan Hidders and Magnus Manske
   7  * @package MediaWiki
   8  * @subpackage Experimental
   9  */
  10
  11 /**
  12  * the base class for an element
  13  */
  14 class element {
  15     var $name = '';
  16     var $attrs = array();
  17     var $children = array();
  18
  19     function sub_makeXHTML ( &$parser , $tag = "" , $attr = "" )
  20         {
  21         $ret = "" ;
  22         if ( $tag != "" )
  23                 {
  24                 $ret .= "<" . $tag ;
  25                 if ( $attr != "" ) $ret .= " " . $attr ;
  26                 $ret .= ">" ;
  27                 }
  28       foreach ($this->children as $child) {
  29             if ( is_string($child) ) {
  30                 $ret .= $child ;
  31             } else {
  32                 $ret .= $child->makeXHTML ( $parser );
  33             }
  34            }
  35         if ( $tag != "" )
  36                 $ret .= "</" . $tag . ">\n" ;
  37       return $ret ;
  38         }
  39
  40     function makeXHTML ( &$parser )
  41         {
  42         $ret = "" ;
  43         $n = $this->name ; # Shortcut
  44         if ( $n == "ARTICLE" )
  45                 $ret .= $this->sub_makeXHTML ( $parser ) ;
  46         else if ( $n == "HEADING" )
  47                 $ret .= $this->sub_makeXHTML ( $parser , "h" . $this->attrs["LEVEL"] ) ;
  48         else if ( $n == "PARAGRAPH" )
  49                 $ret .= $this->sub_makeXHTML ( $parser , "p" ) ;
  50         else if ( $n == "BOLD" )
  51                 $ret .= $this->sub_makeXHTML ( $parser , "strong" ) ;
  52         else if ( $n == "ITALICS" )
  53                 $ret .= $this->sub_makeXHTML ( $parser , "em" ) ;
  54
  55         else if ( $n == "EXTENSION" )
  56                 {
  57                 $ext = $this->attrs["NAME"] ;
  58
  59 #               $ret .= $this->sub_makeXHTML ( $parser , "em" ) ;
  60                 }
  61
  62         else if ( $n == "TABLE" )
  63                 {
  64                 $ret .= $this->sub_makeXHTML ( $parser , "table" ) ;
  65                 }
  66         else if ( $n == "TABLEROW" )
  67                 {
  68                 $ret .= $this->sub_makeXHTML ( $parser , "tr" ) ;
  69                 }
  70         else if ( $n == "TABLECELL" )
  71                 {
  72                 $ret .= $this->sub_makeXHTML ( $parser , "td" ) ;
  73                 }
  74
  75
  76         else if ( $n == "LISTITEM" )
  77                 $ret .= $this->sub_makeXHTML ( $parser , "li" ) ;
  78         else if ( $n == "LIST" )
  79                 {
  80                 $type = "ol" ; # Default
  81                 if ( $this->attrs["TYPE"] == "bullet" ) $type = "ul" ;
  82                 $ret .= $this->sub_makeXHTML ( $parser , $type ) ;
  83                 }
  84
  85         else
  86                 {
  87                 $ret .= "&lt;" . $n . "&gt;" ;
  88                 $ret .= $this->sub_makeXHTML ( $parser ) ;
  89                 $ret .= "&lt;/" . $n . "&gt;" ;
  90                 }
  91         return $ret ;
  92         }
  93
  94     function myPrint() {
  95         $ret = "<ul>\n";
  96         $ret .= "<li> <b> Name: </b> $this->name </li>\n";
  97         // print attributes
  98         $ret .= '<li> <b> Attributes: </b>';
  99         foreach ($this->attrs as $name => $value) {
 100             $ret .= "$name => $value; " ;
 101         }
 102         $ret .= " </li>\n";
 103         // print children
 104         foreach ($this->children as $child) {
 105             if ( is_string($child) ) {
 106                 $ret .= "<li> $child </li>\n";
 107             } else {
 108                 $ret .= $child->myPrint();
 109             }
 110         }
 111         $ret .= "</ul>\n";
 112         return $ret;
 113     }
 114 }
 115
 116 $ancStack = array();    // the stack with ancestral elements
 117
 118 // Three global functions needed for parsing, sorry guys
 119 function wgXMLstartElement($parser, $name, $attrs) {
 120     global $ancStack;
 121
 122     $newElem = new element;
 123     $newElem->name = $name;
 124     $newElem->attrs = $attrs;
 125
 126     array_push($ancStack, $newElem);
 127 }
 128
 129 function wgXMLendElement($parser, $name) {
 130     global $ancStack, $rootElem;
 131     // pop element off stack
 132     $elem = array_pop ($ancStack);
 133     if (count ($ancStack) == 0)
 134         $rootElem = $elem;
 135     else
 136         // add it to its parent
 137         array_push ($ancStack[count($ancStack)-1]->children, $elem);
 138 }
 139
 140 function wgXMLcharacterData($parser, $data) {
 141     global $ancStack;
 142     $data = trim ($data); // Don't add blank lines, they're no use...
 143     // add to parent if parent exists
 144     if ( $ancStack && $data != "" ) {
 145         array_push ($ancStack[count($ancStack)-1]->children, $data);
 146     }
 147 }
 148
 149
 150 /**
 151  * Here's the class that generates a nice tree
 152  */
 153 class xml2php {
 154
 155     function &scanFile( $filename ) {
 156         global $ancStack, $rootElem;
 157         $ancStack = array();
 158
 159         $xml_parser = xml_parser_create();
 160         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 161         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 162         if (!($fp = fopen($filename, 'r'))) {
 163             die('could not open XML input');
 164         }
 165         while ($data = fread($fp, 4096)) {
 166             if (!xml_parse($xml_parser, $data, feof($fp))) {
 167                 die(sprintf("XML error: %s at line %d",
 168                     xml_error_string(xml_get_error_code($xml_parser)),
 169                     xml_get_current_line_number($xml_parser)));
 170             }
 171         }
 172         xml_parser_free($xml_parser);
 173
 174         // return the remaining root element we copied in the beginning
 175         return $rootElem;
 176     }
 177
 178     function scanString ( $input ) {
 179         global $ancStack, $rootElem;
 180         $ancStack = array();
 181
 182         $xml_parser = xml_parser_create();
 183         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 184         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 185
 186         if (!xml_parse ($xml_parser, $input, true)) {
 187             die (sprintf ("XML error: %s at line %d",
 188                 xml_error_string(xml_get_error_code($xml_parser)),
 189                 xml_get_current_line_number($xml_parser)));
 190         }
 191         xml_parser_free ($xml_parser);
 192
 193         // return the remaining root element we copied in the beginning
 194         return $rootElem;
 195     }
 196
 197 }
 198
 199 /* Example code:
 200
 201     $w = new xml2php;
 202     $filename = 'sample.xml';
 203     $result = $w->scanFile( $filename );
 204     print $result->myPrint();
 205 */
 206
 207 $dummytext = "<article><heading level='2'> R-type </heading><paragraph><link><linktarget>image:a.jpg</linktarget><linkoption>1</linkoption><linkoption>2</linkoption><linkoption>3</linkoption><linkoption>text</linkoption></link></paragraph><paragraph>The <link><linktarget>video game</linktarget><linkoption>computer game</linkoption></link> <bold>R-type</bold> is <extension name='nowiki'>cool &amp; stuff</extension> because:</paragraph><list type='bullet'><listitem>it's nice</listitem><listitem>it's fast</listitem><listitem>it has:<list type='bullet'><listitem>graphics</listitem><listitem>sound</listitem></list></listitem></list><table><tablerow><tablecell>Version 1     </tablecell><tablecell>not bad</tablecell></tablerow><tablerow><tablecell>Version 2     </tablecell><tablecell>much better </tablecell></tablerow></table><paragraph>This is a || token in the middle of text.</paragraph></article>" ;
 208
 209 class ParserXML EXTENDS Parser
 210         {
 211         /**#@+
 212          * @access private
 213          */
 214         # Persistent:
 215         var $mTagHooks;
 216
 217         # Cleared with clearState():
 218         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 219         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 220
 221         # Temporary:
 222         var $mOptions, $mTitle, $mOutputType,
 223             $mTemplates,        // cache of already loaded templates, avoids
 224                                 // multiple SQL queries for the same string
 225             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 226                                 // in this path. Used for loop detection.
 227
 228         /**#@-*/
 229
 230         /**
 231          * Constructor
 232          *
 233          * @access public
 234          */
 235         function ParserXML() {
 236                 $this->mTemplates = array();
 237                 $this->mTemplatePath = array();
 238                 $this->mTagHooks = array();
 239                 $this->clearState();
 240         }
 241
 242         /**
 243          * Clear Parser state
 244          *
 245          * @access private
 246          */
 247         function clearState() {
 248                 $this->mOutput = new ParserOutput;
 249                 $this->mAutonumber = 0;
 250                 $this->mLastSection = "";
 251                 $this->mDTopen = false;
 252                 $this->mVariables = false;
 253                 $this->mIncludeCount = array();
 254                 $this->mStripState = array();
 255                 $this->mArgStack = array();
 256                 $this->mInPre = false;
 257         }
 258
 259         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 260                 global $dummytext ;
 261                 $text = $dummytext ;
 262
 263                 $w = new xml2php;
 264                 $result = $w->scanString( $text );
 265                 $text .= "<hr>" . $result->makeXHTML ( $this );
 266                 $text .= "<hr>" . $result->myPrint();
 267
 268                 $this->mOutput->setText ( $text ) ;
 269                 return $this->mOutput;
 270         }
 271
 272         }
 273
 274 ?>