includes/ParserXML.php

   1 <?php
   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 $ret ;
  38         }
  39
  40     function createInternalLink ( &$parser , $target , $display_title , $options )
  41         {
  42         $tp = explode ( ":" , $target ) ; # tp = target parts
  43         $title = "" ;     # The plain title
  44         $language = "" ;  # The language/meta/etc. part
  45         $namespace = "" ; # The namespace, if any
  46         $subtarget = "" ; # The '#' thingy
  47         if ( count ( $tp ) == 1 ) $title = $target ; # Plain and simple case
  48         else
  49                 {
  50                 # To be implemented
  51                 }
  52
  53         if ( $language != "" ) # External link within the WikiMedia project
  54                 {
  55                 return "{language link}" ;
  56                 }
  57         else if ( $namespace != "" ) # Link to another namespace, check for image/media stuff
  58                 {
  59                 return "{namespace link}" ;
  60                 }
  61         else
  62                 {
  63                 return "{internal link}" ;
  64                 }
  65         }
  66
  67     function makeInternalLink ( &$parser )
  68         {
  69         $target = "" ;
  70         $option = array () ;
  71       foreach ($this->children as $child) {
  72             if ( is_string($child) ) {
  73                 # This shouldn't be the case!
  74             } else {
  75                 if ( $child->name == "LINKTARGET" )
  76                         $target = trim ( $child->makeXHTML ( $parser ) ) ;
  77                     else
  78                         $option[] = trim ( $child->makeXHTML ( $parser ) ) ;
  79             }
  80            }
  81
  82       if ( count ( $option ) == 0 ) $option[] = $target ; # Create dummy display title
  83       $display_title = array_pop ( $option ) ;
  84       return $this->createInternalLink ( $parser , $target , $display_title , $option ) ;
  85         }
  86
  87     function makeXHTML ( &$parser )
  88         {
  89         $ret = "" ;
  90         $n = $this->name ; # Shortcut
  91         if ( $n == "ARTICLE" )
  92                 $ret .= $this->sub_makeXHTML ( $parser ) ;
  93         else if ( $n == "HEADING" )
  94                 $ret .= $this->sub_makeXHTML ( $parser , "h" . $this->attrs["LEVEL"] ) ;
  95         else if ( $n == "PARAGRAPH" )
  96                 $ret .= $this->sub_makeXHTML ( $parser , "p" ) ;
  97         else if ( $n == "BOLD" )
  98                 $ret .= $this->sub_makeXHTML ( $parser , "strong" ) ;
  99         else if ( $n == "ITALICS" )
 100                 $ret .= $this->sub_makeXHTML ( $parser , "em" ) ;
 101
 102         else if ( $n == "LINK" )
 103                 $ret .= $this->makeInternalLink ( $parser ) ;
 104         else if ( $n == "LINKTARGET" )
 105                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 106         else if ( $n == "LINKOPTION" )
 107                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 108
 109         else if ( $n == "EXTENSION" ) # This is currently a dummy!!!
 110                 {
 111                 $ext = $this->attrs["NAME"] ;
 112
 113                 $ret .= "&lt;" . $ext . "&gt;" ;
 114                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 115                 $ret .= "&lt;/" . $ext . "&gt; " ;
 116                 }
 117
 118         else if ( $n == "TABLE" )
 119                 {
 120                 $ret .= $this->sub_makeXHTML ( $parser , "table" ) ;
 121                 }
 122         else if ( $n == "TABLEROW" )
 123                 {
 124                 $ret .= $this->sub_makeXHTML ( $parser , "tr" ) ;
 125                 }
 126         else if ( $n == "TABLECELL" )
 127                 {
 128                 $ret .= $this->sub_makeXHTML ( $parser , "td" ) ;
 129                 }
 130
 131
 132         else if ( $n == "LISTITEM" )
 133                 $ret .= $this->sub_makeXHTML ( $parser , "li" ) ;
 134         else if ( $n == "LIST" )
 135                 {
 136                 $type = "ol" ; # Default
 137                 if ( $this->attrs["TYPE"] == "bullet" ) $type = "ul" ;
 138                 $ret .= $this->sub_makeXHTML ( $parser , $type ) ;
 139                 }
 140
 141         else
 142                 {
 143                 $ret .= "&lt;" . $n . "&gt;" ;
 144                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 145                 $ret .= "&lt;/" . $n . "&gt; " ;
 146                 }
 147         $ret = "\n{$ret}\n" ;
 148         $ret = str_replace ( "\n\n" , "\n" , $ret ) ;
 149         return $ret ;
 150         }
 151
 152     function myPrint() {
 153         $ret = "<ul>\n";
 154         $ret .= "<li> <b> Name: </b> $this->name </li>\n";
 155         // print attributes
 156         $ret .= '<li> <b> Attributes: </b>';
 157         foreach ($this->attrs as $name => $value) {
 158             $ret .= "$name => $value; " ;
 159         }
 160         $ret .= " </li>\n";
 161         // print children
 162         foreach ($this->children as $child) {
 163             if ( is_string($child) ) {
 164                 $ret .= "<li> $child </li>\n";
 165             } else {
 166                 $ret .= $child->myPrint();
 167             }
 168         }
 169         $ret .= "</ul>\n";
 170         return $ret;
 171     }
 172 }
 173
 174 $ancStack = array();    // the stack with ancestral elements
 175
 176 // Three global functions needed for parsing, sorry guys
 177 function wgXMLstartElement($parser, $name, $attrs) {
 178     global $ancStack;
 179
 180     $newElem = new element;
 181     $newElem->name = $name;
 182     $newElem->attrs = $attrs;
 183
 184     array_push($ancStack, $newElem);
 185 }
 186
 187 function wgXMLendElement($parser, $name) {
 188     global $ancStack, $rootElem;
 189     // pop element off stack
 190     $elem = array_pop ($ancStack);
 191     if (count ($ancStack) == 0)
 192         $rootElem = $elem;
 193     else
 194         // add it to its parent
 195         array_push ($ancStack[count($ancStack)-1]->children, $elem);
 196 }
 197
 198 function wgXMLcharacterData($parser, $data) {
 199     global $ancStack;
 200     $data = trim ($data); // Don't add blank lines, they're no use...
 201     // add to parent if parent exists
 202     if ( $ancStack && $data != "" ) {
 203         array_push ($ancStack[count($ancStack)-1]->children, $data);
 204     }
 205 }
 206
 207
 208 /**
 209  * Here's the class that generates a nice tree
 210  */
 211 class xml2php {
 212
 213     function &scanFile( $filename ) {
 214         global $ancStack, $rootElem;
 215         $ancStack = array();
 216
 217         $xml_parser = xml_parser_create();
 218         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 219         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 220         if (!($fp = fopen($filename, 'r'))) {
 221             die('could not open XML input');
 222         }
 223         while ($data = fread($fp, 4096)) {
 224             if (!xml_parse($xml_parser, $data, feof($fp))) {
 225                 die(sprintf("XML error: %s at line %d",
 226                     xml_error_string(xml_get_error_code($xml_parser)),
 227                     xml_get_current_line_number($xml_parser)));
 228             }
 229         }
 230         xml_parser_free($xml_parser);
 231
 232         // return the remaining root element we copied in the beginning
 233         return $rootElem;
 234     }
 235
 236     function scanString ( $input ) {
 237         global $ancStack, $rootElem;
 238         $ancStack = array();
 239
 240         $xml_parser = xml_parser_create();
 241         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 242         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 243
 244         if (!xml_parse ($xml_parser, $input, true)) {
 245             die (sprintf ("XML error: %s at line %d",
 246                 xml_error_string(xml_get_error_code($xml_parser)),
 247                 xml_get_current_line_number($xml_parser)));
 248         }
 249         xml_parser_free ($xml_parser);
 250
 251         // return the remaining root element we copied in the beginning
 252         return $rootElem;
 253     }
 254
 255 }
 256
 257 /* Example code:
 258
 259     $w = new xml2php;
 260     $filename = 'sample.xml';
 261     $result = $w->scanFile( $filename );
 262     print $result->myPrint();
 263 */
 264
 265 $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>" ;
 266
 267 class ParserXML EXTENDS Parser
 268         {
 269         /**#@+
 270          * @access private
 271          */
 272         # Persistent:
 273         var $mTagHooks;
 274
 275         # Cleared with clearState():
 276         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 277         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 278
 279         # Temporary:
 280         var $mOptions, $mTitle, $mOutputType,
 281             $mTemplates,        // cache of already loaded templates, avoids
 282                                 // multiple SQL queries for the same string
 283             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 284                                 // in this path. Used for loop detection.
 285
 286         /**#@-*/
 287
 288         /**
 289          * Constructor
 290          *
 291          * @access public
 292          */
 293         function ParserXML() {
 294                 $this->mTemplates = array();
 295                 $this->mTemplatePath = array();
 296                 $this->mTagHooks = array();
 297                 $this->clearState();
 298         }
 299
 300         /**
 301          * Clear Parser state
 302          *
 303          * @access private
 304          */
 305         function clearState() {
 306                 $this->mOutput = new ParserOutput;
 307                 $this->mAutonumber = 0;
 308                 $this->mLastSection = "";
 309                 $this->mDTopen = false;
 310                 $this->mVariables = false;
 311                 $this->mIncludeCount = array();
 312                 $this->mStripState = array();
 313                 $this->mArgStack = array();
 314                 $this->mInPre = false;
 315         }
 316
 317         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 318                 global $dummytext ;
 319                 $text = $dummytext ;
 320
 321                 $w = new xml2php;
 322                 $result = $w->scanString( $text );
 323                 $text .= "<hr>" . $result->makeXHTML ( $this );
 324                 $text .= "<hr>" . $result->myPrint();
 325
 326                 $this->mOutput->setText ( $text ) ;
 327                 return $this->mOutput;
 328         }
 329
 330         }
 331
 332 ?>