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