includes/Content.php

   1 <?php
   2
   3 /**
   4  * A content object represents page content, e.g. the text to show on a page.
   5  * Content objects have no knowledge about how they relate to Wiki pages.
   6  * Content objects are imutable.
   7  *
   8  */
   9 abstract class Content {
  10
  11     public function __construct( $modelName = null ) { #FIXME: really need revId? annoying! #FIXME: really $title? or just when parsing, every time?
  12         $this->mModelName = $modelName;
  13     }
  14
  15     public function getModelName() {
  16         return $this->mModelName;
  17     }
  18
  19     public abstract function getTextForSearchIndex( );
  20
  21     public abstract function getWikitextForTransclusion( );
  22
  23     public abstract function getTextForSummary( $maxlength = 250 );
  24
  25     /**
  26      * Returns native represenation of the data. Interpretation depends on the data model used,
  27      * as given by getDataModel().
  28      *
  29      * @return mixed the native representation of the content. Could be a string, a nested array
  30      *         structure, an object, a binary blob... anything, really.
  31      */
  32     public abstract function getNativeData( ); #FIXME: review all calls carefully, caller must be aware of content model!
  33
  34     /**
  35      * returns the content's nominal size in bogo-bytes.
  36      */
  37     public abstract function getSize( ); #XXX: do we really need/want this here? we could just use the byte syse of the serialized form...
  38
  39     public function isEmpty() {
  40         return $this->getSize() == 0;
  41     }
  42
  43     public function equals( Content $that ) {
  44         if ( empty( $that ) ) return false;
  45         if ( $that === $this ) return true;
  46         if ( $that->getModelName() !== $this->getModelName() ) return false;
  47
  48         return $this->getNativeData() == $that->getNativeData();
  49     }
  50
  51     /**
  52      * Returns true if this content is countable as a "real" wiki page, provided
  53      * that it's also in a countable location (e.g. a current revision in the main namespace).
  54      *
  55      * @param $hasLinks Bool: if it is known whether this content contains links, provide this information here,
  56      *                        to avoid redundant parsing to find out.
  57      */
  58     public abstract function isCountable( $hasLinks = null ) ;
  59
  60     public abstract function getParserOutput( Title $title = null, $revId = null, ParserOptions $options = NULL );
  61
  62     public function getRedirectChain() { #TODO: document!
  63         return null;
  64     }
  65
  66     public function getRedirectTarget() {
  67         return null;
  68     }
  69
  70     public function isRedirect() {
  71         return $this->getRedirectTarget() != null;
  72     }
  73
  74     /**
  75      * Returns the section with the given id.
  76      *
  77      * The default implementation returns null.
  78      *
  79      * @param String $sectionId the section's id
  80      * @return Content|Boolean|null the section, or false if no such section exist, or null if sections are not supported
  81      */
  82     public function getSection( $sectionId ) {
  83         return null;
  84     }
  85
  86     /**
  87      * Replaces a section of the content and returns a Content object with the section replaced.
  88      *
  89      * @param $section empty/null/false or a section number (0, 1, 2, T1, T2...), or "new"
  90      * @param $with Content: new content of the section
  91      * @param $sectionTitle String: new section's subject, only if $section is 'new'
  92      * @return string Complete article text, or null if error
  93      */
  94     public function replaceSection( $section, Content $with, $sectionTitle = ''  ) {
  95         return $this;
  96     }
  97
  98     #TODO: implement specialized ParserOutput for Wikidata model
  99     #TODO: provide "combined" ParserOutput for Multipart... somehow.
 100
 101     # XXX: isCacheable( ) # can/should we do this here?
 102
 103     # TODO: WikiPage::getUndoText( Revision $undo, Revision $undoafter = null )
 104
 105     # TODO: EditPage::getPreloadedText( $preload ) // $wgParser->getPreloadText
 106     # TODO: tie into EditPage, make it use Content-objects throughout, make edit form aware of content model and format
 107     # TODO: tie into WikiPage, make it use Content-objects throughout, especially in doEdit(), doDelete(), updateRevisionOn(), etc
 108     # TODO: make model-aware diff view!
 109     # TODO: handle ImagePage and CategoryPage
 110
 111     # TODO: Title::newFromRedirectRecurse( $this->getRawText() );
 112
 113     # TODO: tie into API to provide contentModel for Revisions
 114     # TODO: tie into API to provide serialized version and contentFormat for Revisions
 115     # TODO: tie into API edit interface
 116
 117 }
 118
 119 /**
 120  * Content object implementation for representing flat text. The
 121  */
 122 abstract class TextContent extends Content {
 123     public function __construct( $text, $modelName = null ) {
 124         parent::__construct($modelName);
 125
 126         $this->mText = $text;
 127     }
 128
 129     public function getTextForSummary( $maxlength = 250 ) {
 130         global $wgContLang;
 131
 132         $text = $this->getNativeData();
 133
 134         $truncatedtext = $wgContLang->truncate(
 135             preg_replace( "/[\n\r]/", ' ', $text ),
 136             max( 0, $maxlength ) );
 137
 138         return $truncatedtext;
 139     }
 140
 141     /**
 142      * returns the content's nominal size in bogo-bytes.
 143      */
 144     public function getSize( ) { #FIXME: use! replace strlen in WikiPage.
 145         $text = $this->getNativeData( );
 146         return strlen( $text );
 147     }
 148
 149     /**
 150      * Returns true if this content is not a redirect, and $wgArticleCountMethod is "any".
 151      *
 152      * @param $hasLinks Bool: if it is known whether this content contains links, provide this information here,
 153      *                        to avoid redundant parsing to find out.
 154      */
 155     public function isCountable( $hasLinks = null ) {
 156         global $wgArticleCountMethod;
 157
 158         if ( $this->isRedirect( ) ) {
 159             return false;
 160         }
 161
 162         if (  $wgArticleCountMethod === 'any' ) {
 163             return true;
 164         }
 165
 166         return false;
 167     }
 168
 169     /**
 170      * Returns the text represented by this Content object, as a string.
 171      *
 172      * @return String the raw text
 173      */
 174     public function getNativeData( ) {
 175         $text = $this->mText;
 176         return $text;
 177     }
 178
 179     /**
 180      * Returns the text represented by this Content object, as a string.
 181      *
 182      * @return String the raw text
 183      */
 184     public function getTextForSearchIndex( ) { #FIXME: use!
 185         return $this->getNativeData();
 186     }
 187
 188     /**
 189      * Returns the text represented by this Content object, as a string.
 190      *
 191      * @return String the raw text
 192      */
 193     public function getWikitextForTransclusion( ) { #FIXME: use!
 194         return $this->getNativeData();
 195     }
 196
 197     /**
 198      * Returns a generic ParserOutput object, wrapping the HTML returned by getHtml().
 199      *
 200      * @return ParserOutput representing the HTML form of the text
 201      */
 202     public function getParserOutput( Title $title = null, $revId = null, ParserOptions $options = null ) {
 203         # generic implementation, relying on $this->getHtml()
 204
 205         $html = $this->getHtml( $options );
 206         $po = new ParserOutput( $html );
 207
 208         if ( $this->mTitle ) $po->setTitleText( $this->mTitle->getText() );
 209
 210         #TODO: cache settings, etc?
 211
 212         return $po;
 213     }
 214
 215     protected abstract function getHtml( );
 216
 217 }
 218
 219 class WikitextContent extends TextContent {
 220     public function __construct( $text ) {
 221         parent::__construct($text, CONTENT_MODEL_WIKITEXT);
 222
 223         $this->mDefaultParserOptions = null; #TODO: use per-class static member?!
 224     }
 225
 226     protected function getHtml( ) {
 227         throw new MWException( "getHtml() not implemented for wikitext. Use getParserOutput()->getText()." );
 228     }
 229
 230     public function getDefaultParserOptions() {
 231         global $wgUser, $wgContLang;
 232
 233         if ( !$this->mDefaultParserOptions ) { #TODO: use per-class static member?!
 234             $this->mDefaultParserOptions = ParserOptions::newFromUserAndLang( $wgUser, $wgContLang );
 235         }
 236
 237         return $this->mDefaultParserOptions;
 238     }
 239
 240     /**
 241      * Returns a ParserOutput object reesulting from parsing the content's text using $wgParser
 242      *
 243      * @return ParserOutput representing the HTML form of the text
 244      */
 245     public function getParserOutput( Title $title = null, $revId = null, ParserOptions $options = null ) {
 246         global $wgParser;
 247
 248         if ( !$options ) {
 249             $options = $this->getDefaultParserOptions();
 250         }
 251
 252         $po = $wgParser->parse( $this->mText, $this->getTitle(), $options, true, true, $this->mRevId );
 253
 254         return $po;
 255     }
 256
 257     /**
 258      * Returns the section with the given id.
 259      *
 260      * @param String $sectionId the section's id
 261      * @return Content|false|null the section, or false if no such section exist, or null if sections are not supported
 262      */
 263     public function getSection( $section ) {
 264         global $wgParser;
 265
 266         $text = $this->getNativeData();
 267         $sect = $wgParser->getSection( $text, $section, false );
 268
 269         return  new WikitextContent( $sect );
 270     }
 271
 272     /**
 273      * Replaces a section in the wikitext
 274      *
 275      * @param $section empty/null/false or a section number (0, 1, 2, T1, T2...), or "new"
 276      * @param $with Content: new content of the section
 277      * @param $sectionTitle String: new section's subject, only if $section is 'new'
 278      * @return string Complete article text, or null if error
 279      */
 280     public function replaceSection( $section, Content $with, $sectionTitle = '' ) {
 281         global $wgParser;
 282
 283         wfProfileIn( __METHOD__ );
 284
 285         $myModelName = $this->getModelName();
 286         $sectionModelName = $with->getModelName();
 287
 288         if ( $sectionModelName != $myModelName  ) {
 289             throw new MWException( "Incompatible content model for section: document uses $myModelName, section uses $sectionModelName." );
 290         }
 291
 292         $oldtext = $this->getNativeData();
 293         $text = $with->getNativeData();
 294
 295         if ( $section == 'new' ) {
 296             # Inserting a new section
 297             $subject = $sectionTitle ? wfMsgForContent( 'newsectionheaderdefaultlevel', $sectionTitle ) . "\n\n" : '';
 298             if ( wfRunHooks( 'PlaceNewSection', array( $this, $oldtext, $subject, &$text ) ) ) {
 299                 $text = strlen( trim( $oldtext ) ) > 0
 300                     ? "{$oldtext}\n\n{$subject}{$text}"
 301                     : "{$subject}{$text}";
 302             }
 303         } else {
 304             # Replacing an existing section; roll out the big guns
 305             global $wgParser;
 306
 307             $text = $wgParser->replaceSection( $oldtext, $section, $text );
 308         }
 309
 310         $newContent = new WikitextContent( $text );
 311
 312         wfProfileOut( __METHOD__ );
 313         return $newContent;
 314     }
 315
 316     public function getRedirectChain() {
 317         $text = $this->getNativeData();
 318         return Title::newFromRedirectArray( $text );
 319     }
 320
 321     public function getRedirectTarget() {
 322         $text = $this->getNativeData();
 323         return Title::newFromRedirect( $text );
 324     }
 325
 326     /**
 327      * Returns true if this content is not a redirect, and this content's text is countable according to
 328      * the criteria defiend by $wgArticleCountMethod.
 329      *
 330      * @param $hasLinks Bool: if it is known whether this content contains links, provide this information here,
 331      *                        to avoid redundant parsing to find out.
 332      */
 333     public function isCountable( $hasLinks = null ) {
 334         global $wgArticleCountMethod;
 335
 336         if ( $this->isRedirect( ) ) {
 337             return false;
 338         }
 339
 340         $text = $this->getNativeData();
 341
 342         switch ( $wgArticleCountMethod ) {
 343             case 'any':
 344                 return true;
 345             case 'comma':
 346                 if ( $text === false ) {
 347                     $text = $this->getRawText();
 348                 }
 349                 return strpos( $text,  ',' ) !== false;
 350             case 'link':
 351                 if ( $hasLinks === null ) { # not know, find out
 352                     $po = $this->getParserOutput();
 353                     $links = $po->getLinks();
 354                     $hasLinks = !empty( $links );
 355                 }
 356
 357                 return $hasLinks;
 358         }
 359     }
 360
 361     public function getTextForSummary( $maxlength = 250 ) {
 362         $truncatedtext = parent::getTextForSummary( $maxlength );
 363
 364         #clean up unfinished links
 365         #XXX: make this optional? wasn't there in autosummary, but required for deletion summary.
 366         $truncatedtext = preg_replace( '/\[\[([^\]]*)\]?$/', '$1', $truncatedtext );
 367
 368         return $truncatedtext;
 369     }
 370
 371 }
 372
 373 class MessageContent extends TextContent {
 374     public function __construct( $msg_key, $params = null, $options = null ) {
 375         parent::__construct(null, CONTENT_MODEL_WIKITEXT);
 376
 377         $this->mMessageKey = $msg_key;
 378
 379         $this->mParameters = $params;
 380
 381         if ( !$options ) $options = array();
 382         $this->mOptions = $options;
 383
 384         $this->mHtmlOptions = null;
 385     }
 386
 387     /**
 388      * Returns the message as rendered HTML, using the options supplied to the constructor plus "parse".
 389      */
 390     protected function getHtml(  ) {
 391         $opt = array_merge( $this->mOptions, array('parse') );
 392
 393         return wfMsgExt( $this->mMessageKey, $this->mParameters, $opt );
 394     }
 395
 396
 397     /**
 398      * Returns the message as raw text, using the options supplied to the constructor minus "parse" and "parseinline".
 399      */
 400     public function getNativeData( ) {
 401         $opt = array_diff( $this->mOptions, array('parse', 'parseinline') );
 402
 403         return wfMsgExt( $this->mMessageKey, $this->mParameters, $opt );
 404     }
 405
 406 }
 407
 408
 409 class JavaScriptContent extends TextContent {
 410     public function __construct( $text ) {
 411         parent::__construct($text, CONTENT_MODEL_JAVASCRIPT);
 412     }
 413
 414     protected function getHtml( ) {
 415         $html = "";
 416         $html .= "<pre class=\"mw-code mw-js\" dir=\"ltr\">\n";
 417         $html .= htmlspecialchars( $this->getNativeData() );
 418         $html .= "\n</pre>\n";
 419
 420         return $html;
 421     }
 422
 423 }
 424
 425 class CssContent extends TextContent {
 426     public function __construct( $text ) {
 427         parent::__construct($text, CONTENT_MODEL_CSS);
 428     }
 429
 430     protected function getHtml( ) {
 431         $html = "";
 432         $html .= "<pre class=\"mw-code mw-css\" dir=\"ltr\">\n";
 433         $html .= htmlspecialchars( $this->getNativeData() );
 434         $html .= "\n</pre>\n";
 435
 436         return $html;
 437     }
 438 }
 439
 440 #FUTURE: special type for redirects?!
 441 #FUTURE: MultipartMultipart < WikipageContent (Main + Links + X)
 442 #FUTURE: LinksContent < LanguageLinksContent, CategoriesContent
 443 #EXAMPLE: CoordinatesContent
 444 #EXAMPLE: WikidataContent