includes/diff/DifferenceEngine.php

   1 <?php
   2 /**
   3  * @defgroup DifferenceEngine DifferenceEngine
   4  */
   5
   6 /**
   7  * Constant to indicate diff cache compatibility.
   8  * Bump this when changing the diff formatting in a way that
   9  * fixes important bugs or such to force cached diff views to
  10  * clear.
  11  */
  12 define( 'MW_DIFF_VERSION', '1.11a' );
  13
  14 /**
  15  * @todo document
  16  * @ingroup DifferenceEngine
  17  */
  18 class DifferenceEngine {
  19         /**#@+
  20          * @private
  21          */
  22         var $mOldid, $mNewid, $mTitle;
  23         var $mOldtitle, $mNewtitle, $mPagetitle;
  24         var $mOldtext, $mNewtext;
  25         var $mOldPage, $mNewPage;
  26         var $mRcidMarkPatrolled;
  27         var $mOldRev, $mNewRev;
  28         var $mRevisionsLoaded = false; // Have the revisions been loaded
  29         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  30         var $htmldiff;
  31         /**#@-*/
  32
  33         /**
  34          * Constructor
  35          * @param $titleObj Title object that the diff is associated with
  36          * @param $old Integer: old ID we want to show and diff with.
  37          * @param $new String: either 'prev' or 'next'.
  38          * @param $rcid Integer: ??? FIXME (default 0)
  39          * @param $refreshCache boolean If set, refreshes the diff cache
  40          * @param $htmldiff boolean If set, output using HTMLDiff instead of raw wikicode diff
  41          */
  42         function __construct( $titleObj = null, $old = 0, $new = 0, $rcid = 0, $refreshCache = false , $htmldiff = false) {
  43                 $this->mTitle = $titleObj;
  44                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  45
  46                 if ( 'prev' === $new ) {
  47                         # Show diff between revision $old and the previous one.
  48                         # Get previous one from DB.
  49                         $this->mNewid = intval($old);
  50                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  51                 } elseif ( 'next' === $new ) {
  52                         # Show diff between revision $old and the previous one.
  53                         # Get previous one from DB.
  54                         #
  55                         $this->mOldid = intval($old);
  56                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  57                         if ( false === $this->mNewid ) {
  58                                 # if no result, NewId points to the newest old revision. The only newer
  59                                 # revision is cur, which is "0".
  60                                 $this->mNewid = 0;
  61                         }
  62                 } else {
  63                         $this->mOldid = intval($old);
  64                         $this->mNewid = intval($new);
  65                         wfRunHooks( 'NewDifferenceEngine', array(&$titleObj, &$this->mOldid, &$this->mNewid, $old, $new) );
  66                 }
  67                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  68                 $this->mRefreshCache = $refreshCache;
  69                 $this->htmldiff = $htmldiff;
  70         }
  71
  72         function getTitle() {
  73                 return $this->mTitle;
  74         }
  75
  76         function showDiffPage( $diffOnly = false ) {
  77                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol, $wgEnableHtmlDiff;
  78                 wfProfileIn( __METHOD__ );
  79
  80
  81                 # If external diffs are enabled both globally and for the user,
  82                 # we'll use the application/x-external-editor interface to call
  83                 # an external diff tool like kompare, kdiff3, etc.
  84                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  85                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  86                         $wgOut->disable();
  87                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  88                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  89                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  90                         $special=$wgLang->getNsText(NS_SPECIAL);
  91                         $control=<<<CONTROL
  92                         [Process]
  93                         Type=Diff text
  94                         Engine=MediaWiki
  95                         Script={$wgServer}{$wgScript}
  96                         Special namespace={$special}
  97
  98                         [File]
  99                         Extension=wiki
 100                         URL=$url1
 101
 102                         [File 2]
 103                         Extension=wiki
 104                         URL=$url2
 105 CONTROL;
 106                         echo($control);
 107                         return;
 108                 }
 109
 110                 $wgOut->setArticleFlag( false );
 111                 if ( !$this->loadRevisionData() ) {
 112                         $t = $this->mTitle->getPrefixedText();
 113                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 114                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 115                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 116                         wfProfileOut( __METHOD__ );
 117                         return;
 118                 }
 119
 120                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 121
 122                 if ( $this->mNewRev->isCurrent() ) {
 123                         $wgOut->setArticleFlag( true );
 124                 }
 125
 126                 # mOldid is false if the difference engine is called with a "vague" query for
 127                 # a diff between a version V and its previous version V' AND the version V
 128                 # is the first version of that article. In that case, V' does not exist.
 129                 if ( $this->mOldid === false ) {
 130                         $this->showFirstRevision();
 131                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 132                         wfProfileOut( __METHOD__ );
 133                         return;
 134                 }
 135
 136                 $wgOut->suppressQuickbar();
 137
 138                 $oldTitle = $this->mOldPage->getPrefixedText();
 139                 $newTitle = $this->mNewPage->getPrefixedText();
 140                 if( $oldTitle == $newTitle ) {
 141                         $wgOut->setPageTitle( $newTitle );
 142                 } else {
 143                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 144                 }
 145                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 146                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 147
 148                 if ( !$this->mOldPage->userCanRead() || !$this->mNewPage->userCanRead() ) {
 149                         $wgOut->loginToUse();
 150                         $wgOut->output();
 151                         $wgOut->disable();
 152                         wfProfileOut( __METHOD__ );
 153                         return;
 154                 }
 155
 156                 $sk = $wgUser->getSkin();
 157
 158                 // Check if page is editable
 159                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 160                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed( 'rollback' ) ) {
 161                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 162                 } else {
 163                         $rollback = '';
 164                 }
 165
 166                 // Prepare a change patrol link, if applicable
 167                 if( $wgUseRCPatrol && $this->mTitle->userCan('patrol') ) {
 168                         // If we've been given an explicit change identifier, use it; saves time
 169                         if( $this->mRcidMarkPatrolled ) {
 170                                 $rcid = $this->mRcidMarkPatrolled;
 171                                 $rc = RecentChange::newFromId( $rcid );
 172                                 // Already patrolled?
 173                                 $rcid = is_object($rc) && !$rc->getAttribute('rc_patrolled') ? $rcid : 0;
 174                         } else {
 175                                 // Look for an unpatrolled change corresponding to this diff
 176                                 $db = wfGetDB( DB_SLAVE );
 177                                 $change = RecentChange::newFromConds(
 178                                         array(
 179                                         // Add redundant user,timestamp condition so we can use the existing index
 180                                                         'rc_user_text'  => $this->mNewRev->getRawUserText(),
 181                                                         'rc_timestamp'  => $db->timestamp( $this->mNewRev->getTimestamp() ),
 182                                                         'rc_this_oldid' => $this->mNewid,
 183                                                         'rc_last_oldid' => $this->mOldid,
 184                                                         'rc_patrolled'  => 0
 185                                         ),
 186                                         __METHOD__
 187                                 );
 188                                 if( $change instanceof RecentChange ) {
 189                                         $rcid = $change->mAttribs['rc_id'];
 190                                 } else {
 191                                         // None found
 192                                         $rcid = 0;
 193                                 }
 194                         }
 195                         // Build the link
 196                         if( $rcid ) {
 197                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj(
 198                                         $this->mTitle,
 199                                         wfMsgHtml( 'markaspatrolleddiff' ),
 200                                                 "action=markpatrolled&rcid={$rcid}"
 201                                         ) . ']</span>';
 202                         } else {
 203                                 $patrol = '';
 204                         }
 205                 } else {
 206                         $patrol = '';
 207                 }
 208
 209                 $htmldiffarg = $this->htmlDiffArgument();
 210                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 211                         'diff=prev&oldid='.$this->mOldid.$htmldiffarg, '', '', 'id="differences-prevlink"' );
 212                 if ( $this->mNewRev->isCurrent() ) {
 213                         $nextlink = '&nbsp;';
 214                 } else {
 215                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 216                                 'diff=next&oldid='.$this->mNewid.$htmldiffarg, '', '', 'id="differences-nextlink"' );
 217                 }
 218
 219                 $oldminor = '';
 220                 $newminor = '';
 221
 222                 if( $this->mOldRev->isMinor() ) {
 223                         $oldminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 224                 }
 225                 if( $this->mNewRev->isMinor() ) {
 226                         $newminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 227                 }
 228
 229                 $rdel = ''; $ldel = '';
 230                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 231                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 232                                 // If revision was hidden from sysops
 233                                 $ldel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 234                         } else {
 235                                 $query = array( 'target' => $this->mOldRev->mTitle->getPrefixedDbkey(),
 236                                         'oldid' => $this->mOldRev->getId()
 237                                 );
 238                                 $ldel = $sk->revDeleteLink( $query, $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 239                         }
 240                         $ldel = "&nbsp;&nbsp;&nbsp;$ldel ";
 241                         // We don't currently handle well changing the top revision's settings
 242                         if( $this->mNewRev->isCurrent() ) {
 243                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 244                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 245                                 // If revision was hidden from sysops
 246                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 247                         } else {
 248                                 $query = array( 'target' =>  $this->mNewRev->mTitle->getPrefixedDbkey(),
 249                                         'oldid' => $this->mNewRev->getId()
 250                                 );
 251                                 $rdel = $sk->revDeleteLink( $query, $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 252                         }
 253                         $rdel = "&nbsp;&nbsp;&nbsp;$rdel ";
 254                 }
 255
 256                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 257                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, true ) . "</div>" .
 258                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, true ) . $ldel . "</div>" .
 259                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 260                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 261                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, true ) . " $rollback</div>" .
 262                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, true ) . $rdel . "</div>" .
 263                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 264
 265                 # Output the diff
 266                 if( !$this->mOldRev->userCan(Revision::DELETED_TEXT) || !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 267                         $this->showDiffStyle();
 268                         $multi = $this->getMultiNotice();
 269                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 270                         $wgOut->addWikiMsg( 'rev-deleted-no-diff' );
 271                 } else if( $wgEnableHtmlDiff && $this->htmldiff ) {
 272                         $multi = $this->getMultiNotice();
 273                         $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'wikicodecomparison' ),
 274                                 'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=0', '', '', 'id="differences-switchtype"' ).'</div>');
 275                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 276                         $this->renderHtmlDiff();
 277                 } else {
 278                         if( $wgEnableHtmlDiff ) {
 279                                 $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'visualcomparison' ),
 280                                         'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=1', '', '', 'id="differences-switchtype"' ).'</div>');
 281                         }
 282                         $this->showDiff( $oldHeader, $newHeader );
 283                         if( !$diffOnly ) {
 284                                 $this->renderNewRevision();
 285                         }
 286                 }
 287                 wfProfileOut( __METHOD__ );
 288         }
 289
 290         /**
 291          * Show the new revision of the page.
 292          */
 293         function renderNewRevision() {
 294                 global $wgOut;
 295                 wfProfileIn( __METHOD__ );
 296
 297                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 298                 # Add deleted rev tag if needed
 299                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 300                         $wgOut->addWikiMsg( 'rev-deleted-text-permission' );
 301                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 302                         $wgOut->addWikiMsg( 'rev-deleted-text-view' );
 303                 }
 304
 305                 if( !$this->mNewRev->isCurrent() ) {
 306                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 307                 }
 308
 309                 $this->loadNewText();
 310                 if( is_object( $this->mNewRev ) ) {
 311                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 312                 }
 313
 314                 if( $this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage() ) {
 315                         // Stolen from Article::view --AG 2007-10-11
 316                         // Give hooks a chance to customise the output
 317                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 318                                 // Wrap the whole lot in a <pre> and don't parse
 319                                 $m = array();
 320                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 321                                 $wgOut->addHTML( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 322                                 $wgOut->addHTML( htmlspecialchars( $this->mNewtext ) );
 323                                 $wgOut->addHTML( "\n</pre>\n" );
 324                         }
 325                 } else {
 326                         $wgOut->addWikiTextTidy( $this->mNewtext );
 327                 }
 328
 329                 if( !$this->mNewRev->isCurrent() ) {
 330                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 331                 }
 332
 333                 wfProfileOut( __METHOD__ );
 334         }
 335
 336
 337         function renderHtmlDiff() {
 338                 global $wgOut, $wgTitle, $wgParser, $wgDebugComments;
 339                 wfProfileIn( __METHOD__ );
 340
 341                 $this->showDiffStyle();
 342
 343                 $wgOut->addHTML( '<h2>'.wfMsgHtml( 'visual-comparison' )."</h2>\n" );
 344                 #add deleted rev tag if needed
 345                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 346                         $wgOut->addWikiMsg( 'rev-deleted-text-permission' );
 347                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 348                         $wgOut->addWikiMsg( 'rev-deleted-text-view' );
 349                 }
 350
 351                 if( !$this->mNewRev->isCurrent() ) {
 352                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 353                 }
 354
 355                 $this->loadText();
 356
 357                 // Old revision
 358                 if( is_object( $this->mOldRev ) ) {
 359                         $wgOut->setRevisionId( $this->mOldRev->getId() );
 360                 }
 361
 362                 $popts = $wgOut->parserOptions();
 363                 $oldTidy = $popts->setTidy( true );
 364                 $popts->setEditSection( false );
 365
 366                 $parserOutput = $wgParser->parse( $this->mOldtext, $wgTitle, $popts, true, true, $wgOut->getRevisionId() );
 367                 $popts->setTidy( $oldTidy );
 368
 369                 //only for new?
 370                 //$wgOut->addParserOutputNoText( $parserOutput );
 371                 $oldHtml = $parserOutput->getText();
 372                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$oldHtml ) );
 373
 374                 // New revision
 375                 if( is_object( $this->mNewRev ) ) {
 376                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 377                 }
 378
 379                 $popts = $wgOut->parserOptions();
 380                 $oldTidy = $popts->setTidy( true );
 381
 382                 $parserOutput = $wgParser->parse( $this->mNewtext, $wgTitle, $popts, true, true, $wgOut->getRevisionId() );
 383                 $popts->setTidy( $oldTidy );
 384
 385                 $wgOut->addParserOutputNoText( $parserOutput );
 386                 $newHtml = $parserOutput->getText();
 387                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$newHtml ) );
 388
 389                 unset($parserOutput, $popts);
 390
 391                 $differ = new HTMLDiffer(new DelegatingContentHandler($wgOut));
 392                 $differ->htmlDiff($oldHtml, $newHtml);
 393                 if ( $wgDebugComments ) {
 394                         $wgOut->addHTML( "\n<!-- HtmlDiff Debug Output:\n" . HTMLDiffer::getDebugOutput() . " End Debug -->" );
 395                 }
 396
 397                 wfProfileOut( __METHOD__ );
 398         }
 399
 400         /**
 401          * Show the first revision of an article. Uses normal diff headers in
 402          * contrast to normal "old revision" display style.
 403          */
 404         function showFirstRevision() {
 405                 global $wgOut, $wgUser;
 406                 wfProfileIn( __METHOD__ );
 407
 408                 # Get article text from the DB
 409                 #
 410                 if ( ! $this->loadNewText() ) {
 411                         $t = $this->mTitle->getPrefixedText();
 412                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 413                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 414                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 415                         wfProfileOut( __METHOD__ );
 416                         return;
 417                 }
 418                 if ( $this->mNewRev->isCurrent() ) {
 419                         $wgOut->setArticleFlag( true );
 420                 }
 421
 422                 # Check if user is allowed to look at this page. If not, bail out.
 423                 #
 424                 if ( !$this->mTitle->userCanRead() ) {
 425                         $wgOut->loginToUse();
 426                         $wgOut->output();
 427                         wfProfileOut( __METHOD__ );
 428                         throw new MWException("Permission Error: you do not have access to view this page");
 429                 }
 430
 431                 # Prepare the header box
 432                 #
 433                 $sk = $wgUser->getSkin();
 434
 435                 $next = $this->mTitle->getNextRevisionID( $this->mNewid );
 436                 if( !$next ) {
 437                         $nextlink = '';
 438                 } else {
 439                         $nextlink = '<br/>' . $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 440                                 'diff=next&oldid=' . $this->mNewid.$this->htmlDiffArgument(), '', '', 'id="differences-nextlink"' );
 441                 }
 442                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\">" .
 443                         $sk->revUserTools( $this->mNewRev ) . "<br/>" . $sk->revComment( $this->mNewRev ) . $nextlink . "</div>\n";
 444
 445                 $wgOut->addHTML( $header );
 446
 447                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 448                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 449
 450                 wfProfileOut( __METHOD__ );
 451         }
 452
 453         function htmlDiffArgument(){
 454                 global $wgEnableHtmlDiff;
 455                 if($wgEnableHtmlDiff){
 456                         if($this->htmldiff){
 457                                 return '&htmldiff=1';
 458                         }else{
 459                                 return '&htmldiff=0';
 460                         }
 461                 }else{
 462                         return '';
 463                 }
 464         }
 465
 466         /**
 467          * Get the diff text, send it to $wgOut
 468          * Returns false if the diff could not be generated, otherwise returns true
 469          */
 470         function showDiff( $otitle, $ntitle ) {
 471                 global $wgOut;
 472                 $diff = $this->getDiff( $otitle, $ntitle );
 473                 if ( $diff === false ) {
 474                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 475                         return false;
 476                 } else {
 477                         $this->showDiffStyle();
 478                         $wgOut->addHTML( $diff );
 479                         return true;
 480                 }
 481         }
 482
 483         /**
 484          * Add style sheets and supporting JS for diff display.
 485          */
 486         function showDiffStyle() {
 487                 global $wgStylePath, $wgStyleVersion, $wgOut;
 488                 $wgOut->addStyle( 'common/diff.css' );
 489
 490                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 491                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 492         }
 493
 494         /**
 495          * Get complete diff table, including header
 496          *
 497          * @param Title $otitle Old title
 498          * @param Title $ntitle New title
 499          * @return mixed
 500          */
 501         function getDiff( $otitle, $ntitle ) {
 502                 $body = $this->getDiffBody();
 503                 if ( $body === false ) {
 504                         return false;
 505                 } else {
 506                         $multi = $this->getMultiNotice();
 507                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 508                 }
 509         }
 510
 511         /**
 512          * Get the diff table body, without header
 513          *
 514          * @return mixed
 515          */
 516         function getDiffBody() {
 517                 global $wgMemc;
 518                 wfProfileIn( __METHOD__ );
 519                 // Check if the diff should be hidden from this user
 520                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 521                         return '';
 522                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 523                         return '';
 524                 }
 525                 // Cacheable?
 526                 $key = false;
 527                 if ( $this->mOldid && $this->mNewid ) {
 528                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 529                         // Try cache
 530                         if ( !$this->mRefreshCache ) {
 531                                 $difftext = $wgMemc->get( $key );
 532                                 if ( $difftext ) {
 533                                         wfIncrStats( 'diff_cache_hit' );
 534                                         $difftext = $this->localiseLineNumbers( $difftext );
 535                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 536                                         wfProfileOut( __METHOD__ );
 537                                         return $difftext;
 538                                 }
 539                         } // don't try to load but save the result
 540                 }
 541
 542                 // Loadtext is permission safe, this just clears out the diff
 543                 if ( !$this->loadText() ) {
 544                         wfProfileOut( __METHOD__ );
 545                         return false;
 546                 }
 547
 548                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 549
 550                 // Save to cache for 7 days
 551                 if ( !wfRunHooks( 'AbortDiffCache', array( &$this ) ) ) {
 552                         wfIncrStats( 'diff_uncacheable' );
 553                 } else if ( $key !== false && $difftext !== false ) {
 554                         wfIncrStats( 'diff_cache_miss' );
 555                         $wgMemc->set( $key, $difftext, 7*86400 );
 556                 } else {
 557                         wfIncrStats( 'diff_uncacheable' );
 558                 }
 559                 // Replace line numbers with the text in the user's language
 560                 if ( $difftext !== false ) {
 561                         $difftext = $this->localiseLineNumbers( $difftext );
 562                 }
 563                 wfProfileOut( __METHOD__ );
 564                 return $difftext;
 565         }
 566
 567         /**
 568          * Generate a diff, no caching
 569          * $otext and $ntext must be already segmented
 570          */
 571         function generateDiffBody( $otext, $ntext ) {
 572                 global $wgExternalDiffEngine, $wgContLang;
 573
 574                 $otext = str_replace( "\r\n", "\n", $otext );
 575                 $ntext = str_replace( "\r\n", "\n", $ntext );
 576
 577                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 578                         # For historical reasons, external diff engine expects
 579                         # input text to be HTML-escaped already
 580                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 581                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 582                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 583                                 dl('php_wikidiff.so');
 584                         }
 585                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 586                         $this->debug( 'wikidiff1' );
 587                 }
 588
 589                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 590                         # Better external diff engine, the 2 may some day be dropped
 591                         # This one does the escaping and segmenting itself
 592                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 593                                 wfProfileIn( __METHOD__ . "-dl" );
 594                                 @dl('php_wikidiff2.so');
 595                                 wfProfileOut( __METHOD__ . "-dl" );
 596                         }
 597                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 598                                 wfProfileIn( 'wikidiff2_do_diff' );
 599                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 600                                 $text .= $this->debug( 'wikidiff2' );
 601                                 wfProfileOut( 'wikidiff2_do_diff' );
 602                                 return $text;
 603                         }
 604                 }
 605                 if ( $wgExternalDiffEngine != 'wikidiff3' && $wgExternalDiffEngine !== false ) {
 606                         # Diff via the shell
 607                         global $wgTmpDirectory;
 608                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 609                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 610
 611                         $tempFile1 = fopen( $tempName1, "w" );
 612                         if ( !$tempFile1 ) {
 613                                 wfProfileOut( __METHOD__ );
 614                                 return false;
 615                         }
 616                         $tempFile2 = fopen( $tempName2, "w" );
 617                         if ( !$tempFile2 ) {
 618                                 wfProfileOut( __METHOD__ );
 619                                 return false;
 620                         }
 621                         fwrite( $tempFile1, $otext );
 622                         fwrite( $tempFile2, $ntext );
 623                         fclose( $tempFile1 );
 624                         fclose( $tempFile2 );
 625                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 626                         wfProfileIn( __METHOD__ . "-shellexec" );
 627                         $difftext = wfShellExec( $cmd );
 628                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 629                         wfProfileOut( __METHOD__ . "-shellexec" );
 630                         unlink( $tempName1 );
 631                         unlink( $tempName2 );
 632                         return $difftext;
 633                 }
 634
 635                 # Native PHP diff
 636                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 637                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 638                 $diffs = new Diff( $ota, $nta );
 639                 $formatter = new TableDiffFormatter();
 640                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 641                 $this->debug();
 642         }
 643
 644         /**
 645          * Generate a debug comment indicating diff generating time,
 646          * server node, and generator backend.
 647          */
 648         protected function debug( $generator="internal" ) {
 649                 global $wgShowHostnames;
 650                 $data = array( $generator );
 651                 if( $wgShowHostnames ) {
 652                         $data[] = wfHostname();
 653                 }
 654                 $data[] = wfTimestamp( TS_DB );
 655                 return "<!-- diff generator: " .
 656                 implode( " ",
 657                 array_map(
 658                                         "htmlspecialchars",
 659                 $data ) ) .
 660                         " -->\n";
 661         }
 662
 663         /**
 664          * Replace line numbers with the text in the user's language
 665          */
 666         function localiseLineNumbers( $text ) {
 667                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 668                 array( &$this, 'localiseLineNumbersCb' ), $text );
 669         }
 670
 671         function localiseLineNumbersCb( $matches ) {
 672                 global $wgLang;
 673                 return wfMsgExt( 'lineno', array( 'parseinline' ), $wgLang->formatNum( $matches[1] ) );
 674         }
 675
 676
 677         /**
 678          * If there are revisions between the ones being compared, return a note saying so.
 679          */
 680         function getMultiNotice() {
 681                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 682                 return '';
 683
 684                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 685                         // Comparing two different pages? Count would be meaningless.
 686                         return '';
 687                 }
 688
 689                 $oldid = $this->mOldRev->getId();
 690                 $newid = $this->mNewRev->getId();
 691                 if ( $oldid > $newid ) {
 692                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 693                 }
 694
 695                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 696                 if ( !$n )
 697                 return '';
 698
 699                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 700         }
 701
 702
 703         /**
 704          * Add the header to a diff body
 705          */
 706         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 707                 $header = "
 708                 <table class='diff'>
 709                 <col class='diff-marker' />
 710                 <col class='diff-content' />
 711                 <col class='diff-marker' />
 712                 <col class='diff-content' />
 713                 <tr valign='top'>
 714                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 715                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 716                 </tr>
 717                 ";
 718
 719                 if ( $multi != '' )
 720                 $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 721
 722                 return $header . $diff . "</table>";
 723         }
 724
 725         /**
 726          * Use specified text instead of loading from the database
 727          */
 728         function setText( $oldText, $newText ) {
 729                 $this->mOldtext = $oldText;
 730                 $this->mNewtext = $newText;
 731                 $this->mTextLoaded = 2;
 732         }
 733
 734         /**
 735          * Load revision metadata for the specified articles. If newid is 0, then compare
 736          * the old article in oldid to the current article; if oldid is 0, then
 737          * compare the current article to the immediately previous one (ignoring the
 738          * value of newid).
 739          *
 740          * If oldid is false, leave the corresponding revision object set
 741          * to false. This is impossible via ordinary user input, and is provided for
 742          * API convenience.
 743          */
 744         function loadRevisionData() {
 745                 global $wgLang, $wgUser;
 746                 if ( $this->mRevisionsLoaded ) {
 747                         return true;
 748                 } else {
 749                         // Whether it succeeds or fails, we don't want to try again
 750                         $this->mRevisionsLoaded = true;
 751                 }
 752
 753                 // Load the new revision object
 754                 $this->mNewRev = $this->mNewid
 755                 ? Revision::newFromId( $this->mNewid )
 756                 : Revision::newFromTitle( $this->mTitle );
 757                 if( !$this->mNewRev instanceof Revision )
 758                 return false;
 759
 760                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 761                 $this->mNewid = $this->mNewRev->getId();
 762
 763                 // Check if page is editable
 764                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 765
 766                 // Set assorted variables
 767                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 768                 $this->mNewPage = $this->mNewRev->getTitle();
 769                 if( $this->mNewRev->isCurrent() ) {
 770                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 771                         $this->mPagetitle = wfMsgHTML( 'currentrev-asof', $timestamp );
 772                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 773
 774                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 775                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 776
 777                 } else {
 778                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 779                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 780                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 781
 782                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 783                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 784                 }
 785                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 786                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 787                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 788                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 789                 }
 790
 791                 // Load the old revision object
 792                 $this->mOldRev = false;
 793                 if( $this->mOldid ) {
 794                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 795                 } elseif ( $this->mOldid === 0 ) {
 796                         $rev = $this->mNewRev->getPrevious();
 797                         if( $rev ) {
 798                                 $this->mOldid = $rev->getId();
 799                                 $this->mOldRev = $rev;
 800                         } else {
 801                                 // No previous revision; mark to show as first-version only.
 802                                 $this->mOldid = false;
 803                                 $this->mOldRev = false;
 804                         }
 805                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 806
 807                 if( is_null( $this->mOldRev ) ) {
 808                         return false;
 809                 }
 810
 811                 if ( $this->mOldRev ) {
 812                         $this->mOldPage = $this->mOldRev->getTitle();
 813
 814                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 815                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 816                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 817                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 818
 819                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 820                         . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 821                         // Add an "undo" link
 822                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 823                         $htmlLink = htmlspecialchars( wfMsg( 'editundo' ) );
 824                         $htmlTitle = $wgUser->getSkin()->tooltip( 'undo' );
 825                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 826                                 $this->mNewtitle .= " (<a href='$newUndo' $htmlTitle>" . $htmlLink . "</a>)";
 827                         }
 828
 829                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 830                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 831                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 832                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 833                         }
 834                 }
 835
 836                 return true;
 837         }
 838
 839         /**
 840          * Load the text of the revisions, as well as revision data.
 841          */
 842         function loadText() {
 843                 if ( $this->mTextLoaded == 2 ) {
 844                         return true;
 845                 } else {
 846                         // Whether it succeeds or fails, we don't want to try again
 847                         $this->mTextLoaded = 2;
 848                 }
 849
 850                 if ( !$this->loadRevisionData() ) {
 851                         return false;
 852                 }
 853                 if ( $this->mOldRev ) {
 854                         $this->mOldtext = $this->mOldRev->getText( Revision::FOR_THIS_USER );
 855                         if ( $this->mOldtext === false ) {
 856                                 return false;
 857                         }
 858                 }
 859                 if ( $this->mNewRev ) {
 860                         $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 861                         if ( $this->mNewtext === false ) {
 862                                 return false;
 863                         }
 864                 }
 865                 return true;
 866         }
 867
 868         /**
 869          * Load the text of the new revision, not the old one
 870          */
 871         function loadNewText() {
 872                 if ( $this->mTextLoaded >= 1 ) {
 873                         return true;
 874                 } else {
 875                         $this->mTextLoaded = 1;
 876                 }
 877                 if ( !$this->loadRevisionData() ) {
 878                         return false;
 879                 }
 880                 $this->mNewtext = $this->mNewRev->getText();
 881                 return true;
 882         }
 883
 884
 885 }
 886
 887 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 888 //
 889 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 890 // You may copy this code freely under the conditions of the GPL.
 891 //
 892
 893 define('USE_ASSERTS', function_exists('assert'));
 894
 895 /**
 896  * @todo document
 897  * @private
 898  * @ingroup DifferenceEngine
 899  */
 900 class _DiffOp {
 901         var $type;
 902         var $orig;
 903         var $closing;
 904
 905         function reverse() {
 906                 trigger_error('pure virtual', E_USER_ERROR);
 907         }
 908
 909         function norig() {
 910                 return $this->orig ? sizeof($this->orig) : 0;
 911         }
 912
 913         function nclosing() {
 914                 return $this->closing ? sizeof($this->closing) : 0;
 915         }
 916 }
 917
 918 /**
 919  * @todo document
 920  * @private
 921  * @ingroup DifferenceEngine
 922  */
 923 class _DiffOp_Copy extends _DiffOp {
 924         var $type = 'copy';
 925
 926         function _DiffOp_Copy ($orig, $closing = false) {
 927                 if (!is_array($closing))
 928                 $closing = $orig;
 929                 $this->orig = $orig;
 930                 $this->closing = $closing;
 931         }
 932
 933         function reverse() {
 934                 return new _DiffOp_Copy($this->closing, $this->orig);
 935         }
 936 }
 937
 938 /**
 939  * @todo document
 940  * @private
 941  * @ingroup DifferenceEngine
 942  */
 943 class _DiffOp_Delete extends _DiffOp {
 944         var $type = 'delete';
 945
 946         function _DiffOp_Delete ($lines) {
 947                 $this->orig = $lines;
 948                 $this->closing = false;
 949         }
 950
 951         function reverse() {
 952                 return new _DiffOp_Add($this->orig);
 953         }
 954 }
 955
 956 /**
 957  * @todo document
 958  * @private
 959  * @ingroup DifferenceEngine
 960  */
 961 class _DiffOp_Add extends _DiffOp {
 962         var $type = 'add';
 963
 964         function _DiffOp_Add ($lines) {
 965                 $this->closing = $lines;
 966                 $this->orig = false;
 967         }
 968
 969         function reverse() {
 970                 return new _DiffOp_Delete($this->closing);
 971         }
 972 }
 973
 974 /**
 975  * @todo document
 976  * @private
 977  * @ingroup DifferenceEngine
 978  */
 979 class _DiffOp_Change extends _DiffOp {
 980         var $type = 'change';
 981
 982         function _DiffOp_Change ($orig, $closing) {
 983                 $this->orig = $orig;
 984                 $this->closing = $closing;
 985         }
 986
 987         function reverse() {
 988                 return new _DiffOp_Change($this->closing, $this->orig);
 989         }
 990 }
 991
 992 /**
 993  * Class used internally by Diff to actually compute the diffs.
 994  *
 995  * The algorithm used here is mostly lifted from the perl module
 996  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 997  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 998  *
 999  * More ideas are taken from:
1000  *       http://www.ics.uci.edu/~eppstein/161/960229.html
1001  *
1002  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
1003  * diffutils-2.7, which can be found at:
1004  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
1005  *
1006  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
1007  * are my own.
1008  *
1009  * Line length limits for robustness added by Tim Starling, 2005-08-31
1010  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
1011  *
1012  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
1013  * @private
1014  * @ingroup DifferenceEngine
1015  */
1016 class _DiffEngine {
1017
1018         const MAX_XREF_LENGTH =  10000;
1019
1020         function diff ($from_lines, $to_lines){
1021                 wfProfileIn( __METHOD__ );
1022
1023                 // Diff and store locally
1024                 $this->diff_local($from_lines, $to_lines);
1025
1026                 // Merge edits when possible
1027                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
1028                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
1029
1030                 // Compute the edit operations.
1031                 $n_from = sizeof($from_lines);
1032                 $n_to = sizeof($to_lines);
1033
1034                 $edits = array();
1035                 $xi = $yi = 0;
1036                 while ($xi < $n_from || $yi < $n_to) {
1037                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
1038                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
1039
1040                         // Skip matching "snake".
1041                         $copy = array();
1042                         while ( $xi < $n_from && $yi < $n_to
1043                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
1044                                 $copy[] = $from_lines[$xi++];
1045                                 ++$yi;
1046                         }
1047                         if ($copy)
1048                         $edits[] = new _DiffOp_Copy($copy);
1049
1050                         // Find deletes & adds.
1051                         $delete = array();
1052                         while ($xi < $n_from && $this->xchanged[$xi])
1053                         $delete[] = $from_lines[$xi++];
1054
1055                         $add = array();
1056                         while ($yi < $n_to && $this->ychanged[$yi])
1057                         $add[] = $to_lines[$yi++];
1058
1059                         if ($delete && $add)
1060                         $edits[] = new _DiffOp_Change($delete, $add);
1061                         elseif ($delete)
1062                         $edits[] = new _DiffOp_Delete($delete);
1063                         elseif ($add)
1064                         $edits[] = new _DiffOp_Add($add);
1065                 }
1066                 wfProfileOut( __METHOD__ );
1067                 return $edits;
1068         }
1069
1070         function diff_local ($from_lines, $to_lines) {
1071                 global $wgExternalDiffEngine;
1072                 wfProfileIn( __METHOD__);
1073
1074                 if($wgExternalDiffEngine == 'wikidiff3'){
1075                         // wikidiff3
1076                         $wikidiff3 = new WikiDiff3();
1077                         $wikidiff3->diff($from_lines, $to_lines);
1078                         $this->xchanged = $wikidiff3->removed;
1079                         $this->ychanged = $wikidiff3->added;
1080                         unset($wikidiff3);
1081                 }else{
1082                         // old diff
1083                         $n_from = sizeof($from_lines);
1084                         $n_to = sizeof($to_lines);
1085                         $this->xchanged = $this->ychanged = array();
1086                         $this->xv = $this->yv = array();
1087                         $this->xind = $this->yind = array();
1088                         unset($this->seq);
1089                         unset($this->in_seq);
1090                         unset($this->lcs);
1091
1092                         // Skip leading common lines.
1093                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
1094                                 if ($from_lines[$skip] !== $to_lines[$skip])
1095                                 break;
1096                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
1097                         }
1098                         // Skip trailing common lines.
1099                         $xi = $n_from; $yi = $n_to;
1100                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
1101                                 if ($from_lines[$xi] !== $to_lines[$yi])
1102                                 break;
1103                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
1104                         }
1105
1106                         // Ignore lines which do not exist in both files.
1107                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1108                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
1109                         }
1110
1111                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
1112                                 $line = $to_lines[$yi];
1113                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
1114                                 continue;
1115                                 $yhash[$this->_line_hash($line)] = 1;
1116                                 $this->yv[] = $line;
1117                                 $this->yind[] = $yi;
1118                         }
1119                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1120                                 $line = $from_lines[$xi];
1121                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
1122                                 continue;
1123                                 $this->xv[] = $line;
1124                                 $this->xind[] = $xi;
1125                         }
1126
1127                         // Find the LCS.
1128                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
1129                 }
1130                 wfProfileOut( __METHOD__ );
1131         }
1132
1133         /**
1134          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1135          */
1136         function _line_hash( $line ) {
1137                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1138                         return md5( $line );
1139                 } else {
1140                         return $line;
1141                 }
1142         }
1143
1144         /* Divide the Largest Common Subsequence (LCS) of the sequences
1145          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1146          * sized segments.
1147          *
1148          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1149          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1150          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1151          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1152          * that (X0, Y0) == (XOFF, YOFF) and
1153          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1154          *
1155          * This function assumes that the first lines of the specified portions
1156          * of the two files do not match, and likewise that the last lines do not
1157          * match.  The caller must trim matching lines from the beginning and end
1158          * of the portions it is going to specify.
1159          */
1160         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1161                 $flip = false;
1162
1163                 if ($xlim - $xoff > $ylim - $yoff) {
1164                         // Things seems faster (I'm not sure I understand why)
1165                         // when the shortest sequence in X.
1166                         $flip = true;
1167                         list ($xoff, $xlim, $yoff, $ylim)
1168                         = array( $yoff, $ylim, $xoff, $xlim);
1169                 }
1170
1171                 if ($flip)
1172                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1173                 $ymatches[$this->xv[$i]][] = $i;
1174                 else
1175                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1176                 $ymatches[$this->yv[$i]][] = $i;
1177
1178                 $this->lcs = 0;
1179                 $this->seq[0]= $yoff - 1;
1180                 $this->in_seq = array();
1181                 $ymids[0] = array();
1182
1183                 $numer = $xlim - $xoff + $nchunks - 1;
1184                 $x = $xoff;
1185                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1186                         if ($chunk > 0)
1187                         for ($i = 0; $i <= $this->lcs; $i++)
1188                         $ymids[$i][$chunk-1] = $this->seq[$i];
1189
1190                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1191                         for ( ; $x < $x1; $x++) {
1192                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1193                                 if (empty($ymatches[$line]))
1194                                 continue;
1195                                 $matches = $ymatches[$line];
1196                                 reset($matches);
1197                                 while (list ($junk, $y) = each($matches))
1198                                 if (empty($this->in_seq[$y])) {
1199                                         $k = $this->_lcs_pos($y);
1200                                         USE_ASSERTS && assert($k > 0);
1201                                         $ymids[$k] = $ymids[$k-1];
1202                                         break;
1203                                 }
1204                                 while (list ( /* $junk */, $y) = each($matches)) {
1205                                         if ($y > $this->seq[$k-1]) {
1206                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1207                                                 // Optimization: this is a common case:
1208                                                 //      next match is just replacing previous match.
1209                                                 $this->in_seq[$this->seq[$k]] = false;
1210                                                 $this->seq[$k] = $y;
1211                                                 $this->in_seq[$y] = 1;
1212                                         } else if (empty($this->in_seq[$y])) {
1213                                                 $k = $this->_lcs_pos($y);
1214                                                 USE_ASSERTS && assert($k > 0);
1215                                                 $ymids[$k] = $ymids[$k-1];
1216                                         }
1217                                 }
1218                         }
1219                 }
1220
1221                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1222                 $ymid = $ymids[$this->lcs];
1223                 for ($n = 0; $n < $nchunks - 1; $n++) {
1224                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1225                         $y1 = $ymid[$n] + 1;
1226                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1227                 }
1228                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1229
1230                 return array($this->lcs, $seps);
1231         }
1232
1233         function _lcs_pos ($ypos) {
1234                 $end = $this->lcs;
1235                 if ($end == 0 || $ypos > $this->seq[$end]) {
1236                         $this->seq[++$this->lcs] = $ypos;
1237                         $this->in_seq[$ypos] = 1;
1238                         return $this->lcs;
1239                 }
1240
1241                 $beg = 1;
1242                 while ($beg < $end) {
1243                         $mid = (int)(($beg + $end) / 2);
1244                         if ( $ypos > $this->seq[$mid] )
1245                         $beg = $mid + 1;
1246                         else
1247                         $end = $mid;
1248                 }
1249
1250                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1251
1252                 $this->in_seq[$this->seq[$end]] = false;
1253                 $this->seq[$end] = $ypos;
1254                 $this->in_seq[$ypos] = 1;
1255                 return $end;
1256         }
1257
1258         /* Find LCS of two sequences.
1259          *
1260          * The results are recorded in the vectors $this->{x,y}changed[], by
1261          * storing a 1 in the element for each line that is an insertion
1262          * or deletion (ie. is not in the LCS).
1263          *
1264          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1265          *
1266          * Note that XLIM, YLIM are exclusive bounds.
1267          * All line numbers are origin-0 and discarded lines are not counted.
1268          */
1269         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1270                 // Slide down the bottom initial diagonal.
1271                 while ($xoff < $xlim && $yoff < $ylim
1272                 && $this->xv[$xoff] == $this->yv[$yoff]) {
1273                         ++$xoff;
1274                         ++$yoff;
1275                 }
1276
1277                 // Slide up the top initial diagonal.
1278                 while ($xlim > $xoff && $ylim > $yoff
1279                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1280                         --$xlim;
1281                         --$ylim;
1282                 }
1283
1284                 if ($xoff == $xlim || $yoff == $ylim)
1285                 $lcs = 0;
1286                 else {
1287                         // This is ad hoc but seems to work well.
1288                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1289                         //$nchunks = max(2,min(8,(int)$nchunks));
1290                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1291                         list ($lcs, $seps)
1292                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1293                 }
1294
1295                 if ($lcs == 0) {
1296                         // X and Y sequences have no common subsequence:
1297                         // mark all changed.
1298                         while ($yoff < $ylim)
1299                         $this->ychanged[$this->yind[$yoff++]] = 1;
1300                         while ($xoff < $xlim)
1301                         $this->xchanged[$this->xind[$xoff++]] = 1;
1302                 } else {
1303                         // Use the partitions to split this problem into subproblems.
1304                         reset($seps);
1305                         $pt1 = $seps[0];
1306                         while ($pt2 = next($seps)) {
1307                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1308                                 $pt1 = $pt2;
1309                         }
1310                 }
1311         }
1312
1313         /* Adjust inserts/deletes of identical lines to join changes
1314          * as much as possible.
1315          *
1316          * We do something when a run of changed lines include a
1317          * line at one end and has an excluded, identical line at the other.
1318          * We are free to choose which identical line is included.
1319          * `compareseq' usually chooses the one at the beginning,
1320          * but usually it is cleaner to consider the following identical line
1321          * to be the "change".
1322          *
1323          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1324          */
1325         function _shift_boundaries ($lines, &$changed, $other_changed) {
1326                 wfProfileIn( __METHOD__ );
1327                 $i = 0;
1328                 $j = 0;
1329
1330                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1331                 $len = sizeof($lines);
1332                 $other_len = sizeof($other_changed);
1333
1334                 while (1) {
1335                         /*
1336                          * Scan forwards to find beginning of another run of changes.
1337                          * Also keep track of the corresponding point in the other file.
1338                          *
1339                          * Throughout this code, $i and $j are adjusted together so that
1340                          * the first $i elements of $changed and the first $j elements
1341                          * of $other_changed both contain the same number of zeros
1342                          * (unchanged lines).
1343                          * Furthermore, $j is always kept so that $j == $other_len or
1344                          * $other_changed[$j] == false.
1345                          */
1346                         while ($j < $other_len && $other_changed[$j])
1347                         $j++;
1348
1349                         while ($i < $len && ! $changed[$i]) {
1350                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1351                                 $i++; $j++;
1352                                 while ($j < $other_len && $other_changed[$j])
1353                                 $j++;
1354                         }
1355
1356                         if ($i == $len)
1357                         break;
1358
1359                         $start = $i;
1360
1361                         // Find the end of this run of changes.
1362                         while (++$i < $len && $changed[$i])
1363                         continue;
1364
1365                         do {
1366                                 /*
1367                                  * Record the length of this run of changes, so that
1368                                  * we can later determine whether the run has grown.
1369                                  */
1370                                 $runlength = $i - $start;
1371
1372                                 /*
1373                                  * Move the changed region back, so long as the
1374                                  * previous unchanged line matches the last changed one.
1375                                  * This merges with previous changed regions.
1376                                  */
1377                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1378                                         $changed[--$start] = 1;
1379                                         $changed[--$i] = false;
1380                                         while ($start > 0 && $changed[$start - 1])
1381                                         $start--;
1382                                         USE_ASSERTS && assert('$j > 0');
1383                                         while ($other_changed[--$j])
1384                                         continue;
1385                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1386                                 }
1387
1388                                 /*
1389                                  * Set CORRESPONDING to the end of the changed run, at the last
1390                                  * point where it corresponds to a changed run in the other file.
1391                                  * CORRESPONDING == LEN means no such point has been found.
1392                                  */
1393                                 $corresponding = $j < $other_len ? $i : $len;
1394
1395                                 /*
1396                                  * Move the changed region forward, so long as the
1397                                  * first changed line matches the following unchanged one.
1398                                  * This merges with following changed regions.
1399                                  * Do this second, so that if there are no merges,
1400                                  * the changed region is moved forward as far as possible.
1401                                  */
1402                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1403                                         $changed[$start++] = false;
1404                                         $changed[$i++] = 1;
1405                                         while ($i < $len && $changed[$i])
1406                                         $i++;
1407
1408                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1409                                         $j++;
1410                                         if ($j < $other_len && $other_changed[$j]) {
1411                                                 $corresponding = $i;
1412                                                 while ($j < $other_len && $other_changed[$j])
1413                                                 $j++;
1414                                         }
1415                                 }
1416                         } while ($runlength != $i - $start);
1417
1418                         /*
1419                          * If possible, move the fully-merged run of changes
1420                          * back to a corresponding run in the other file.
1421                          */
1422                         while ($corresponding < $i) {
1423                                 $changed[--$start] = 1;
1424                                 $changed[--$i] = 0;
1425                                 USE_ASSERTS && assert('$j > 0');
1426                                 while ($other_changed[--$j])
1427                                 continue;
1428                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1429                         }
1430                 }
1431                 wfProfileOut( __METHOD__ );
1432         }
1433 }
1434
1435 /**
1436  * Class representing a 'diff' between two sequences of strings.
1437  * @todo document
1438  * @private
1439  * @ingroup DifferenceEngine
1440  */
1441 class Diff
1442 {
1443         var $edits;
1444
1445         /**
1446          * Constructor.
1447          * Computes diff between sequences of strings.
1448          *
1449          * @param $from_lines array An array of strings.
1450          *                (Typically these are lines from a file.)
1451          * @param $to_lines array An array of strings.
1452          */
1453         function Diff($from_lines, $to_lines) {
1454                 $eng = new _DiffEngine;
1455                 $this->edits = $eng->diff($from_lines, $to_lines);
1456                 //$this->_check($from_lines, $to_lines);
1457         }
1458
1459         /**
1460          * Compute reversed Diff.
1461          *
1462          * SYNOPSIS:
1463          *
1464          *      $diff = new Diff($lines1, $lines2);
1465          *      $rev = $diff->reverse();
1466          * @return object A Diff object representing the inverse of the
1467          *                                original diff.
1468          */
1469         function reverse () {
1470                 $rev = $this;
1471                 $rev->edits = array();
1472                 foreach ($this->edits as $edit) {
1473                         $rev->edits[] = $edit->reverse();
1474                 }
1475                 return $rev;
1476         }
1477
1478         /**
1479          * Check for empty diff.
1480          *
1481          * @return bool True iff two sequences were identical.
1482          */
1483         function isEmpty () {
1484                 foreach ($this->edits as $edit) {
1485                         if ($edit->type != 'copy')
1486                         return false;
1487                 }
1488                 return true;
1489         }
1490
1491         /**
1492          * Compute the length of the Longest Common Subsequence (LCS).
1493          *
1494          * This is mostly for diagnostic purposed.
1495          *
1496          * @return int The length of the LCS.
1497          */
1498         function lcs () {
1499                 $lcs = 0;
1500                 foreach ($this->edits as $edit) {
1501                         if ($edit->type == 'copy')
1502                         $lcs += sizeof($edit->orig);
1503                 }
1504                 return $lcs;
1505         }
1506
1507         /**
1508          * Get the original set of lines.
1509          *
1510          * This reconstructs the $from_lines parameter passed to the
1511          * constructor.
1512          *
1513          * @return array The original sequence of strings.
1514          */
1515         function orig() {
1516                 $lines = array();
1517
1518                 foreach ($this->edits as $edit) {
1519                         if ($edit->orig)
1520                         array_splice($lines, sizeof($lines), 0, $edit->orig);
1521                 }
1522                 return $lines;
1523         }
1524
1525         /**
1526          * Get the closing set of lines.
1527          *
1528          * This reconstructs the $to_lines parameter passed to the
1529          * constructor.
1530          *
1531          * @return array The sequence of strings.
1532          */
1533         function closing() {
1534                 $lines = array();
1535
1536                 foreach ($this->edits as $edit) {
1537                         if ($edit->closing)
1538                         array_splice($lines, sizeof($lines), 0, $edit->closing);
1539                 }
1540                 return $lines;
1541         }
1542
1543         /**
1544          * Check a Diff for validity.
1545          *
1546          * This is here only for debugging purposes.
1547          */
1548         function _check ($from_lines, $to_lines) {
1549                 wfProfileIn( __METHOD__ );
1550                 if (serialize($from_lines) != serialize($this->orig()))
1551                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1552                 if (serialize($to_lines) != serialize($this->closing()))
1553                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1554
1555                 $rev = $this->reverse();
1556                 if (serialize($to_lines) != serialize($rev->orig()))
1557                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
1558                 if (serialize($from_lines) != serialize($rev->closing()))
1559                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1560
1561
1562                 $prevtype = 'none';
1563                 foreach ($this->edits as $edit) {
1564                         if ( $prevtype == $edit->type )
1565                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1566                         $prevtype = $edit->type;
1567                 }
1568
1569                 $lcs = $this->lcs();
1570                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1571                 wfProfileOut( __METHOD__ );
1572         }
1573 }
1574
1575 /**
1576  * @todo document, bad name.
1577  * @private
1578  * @ingroup DifferenceEngine
1579  */
1580 class MappedDiff extends Diff
1581 {
1582         /**
1583          * Constructor.
1584          *
1585          * Computes diff between sequences of strings.
1586          *
1587          * This can be used to compute things like
1588          * case-insensitve diffs, or diffs which ignore
1589          * changes in white-space.
1590          *
1591          * @param $from_lines array An array of strings.
1592          *      (Typically these are lines from a file.)
1593          *
1594          * @param $to_lines array An array of strings.
1595          *
1596          * @param $mapped_from_lines array This array should
1597          *      have the same size number of elements as $from_lines.
1598          *      The elements in $mapped_from_lines and
1599          *      $mapped_to_lines are what is actually compared
1600          *      when computing the diff.
1601          *
1602          * @param $mapped_to_lines array This array should
1603          *      have the same number of elements as $to_lines.
1604          */
1605         function MappedDiff($from_lines, $to_lines,
1606         $mapped_from_lines, $mapped_to_lines) {
1607                 wfProfileIn( __METHOD__ );
1608
1609                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1610                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1611
1612                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1613
1614                 $xi = $yi = 0;
1615                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1616                         $orig = &$this->edits[$i]->orig;
1617                         if (is_array($orig)) {
1618                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1619                                 $xi += sizeof($orig);
1620                         }
1621
1622                         $closing = &$this->edits[$i]->closing;
1623                         if (is_array($closing)) {
1624                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1625                                 $yi += sizeof($closing);
1626                         }
1627                 }
1628                 wfProfileOut( __METHOD__ );
1629         }
1630 }
1631
1632 /**
1633  * A class to format Diffs
1634  *
1635  * This class formats the diff in classic diff format.
1636  * It is intended that this class be customized via inheritance,
1637  * to obtain fancier outputs.
1638  * @todo document
1639  * @private
1640  * @ingroup DifferenceEngine
1641  */
1642 class DiffFormatter {
1643         /**
1644          * Number of leading context "lines" to preserve.
1645          *
1646          * This should be left at zero for this class, but subclasses
1647          * may want to set this to other values.
1648          */
1649         var $leading_context_lines = 0;
1650
1651         /**
1652          * Number of trailing context "lines" to preserve.
1653          *
1654          * This should be left at zero for this class, but subclasses
1655          * may want to set this to other values.
1656          */
1657         var $trailing_context_lines = 0;
1658
1659         /**
1660          * Format a diff.
1661          *
1662          * @param $diff object A Diff object.
1663          * @return string The formatted output.
1664          */
1665         function format($diff) {
1666                 wfProfileIn( __METHOD__ );
1667
1668                 $xi = $yi = 1;
1669                 $block = false;
1670                 $context = array();
1671
1672                 $nlead = $this->leading_context_lines;
1673                 $ntrail = $this->trailing_context_lines;
1674
1675                 $this->_start_diff();
1676
1677                 foreach ($diff->edits as $edit) {
1678                         if ($edit->type == 'copy') {
1679                                 if (is_array($block)) {
1680                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1681                                                 $block[] = $edit;
1682                                         }
1683                                         else{
1684                                                 if ($ntrail) {
1685                                                         $context = array_slice($edit->orig, 0, $ntrail);
1686                                                         $block[] = new _DiffOp_Copy($context);
1687                                                 }
1688                                                 $this->_block($x0, $ntrail + $xi - $x0,
1689                                                 $y0, $ntrail + $yi - $y0,
1690                                                 $block);
1691                                                 $block = false;
1692                                         }
1693                                 }
1694                                 $context = $edit->orig;
1695                         }
1696                         else {
1697                                 if (! is_array($block)) {
1698                                         $context = array_slice($context, sizeof($context) - $nlead);
1699                                         $x0 = $xi - sizeof($context);
1700                                         $y0 = $yi - sizeof($context);
1701                                         $block = array();
1702                                         if ($context)
1703                                         $block[] = new _DiffOp_Copy($context);
1704                                 }
1705                                 $block[] = $edit;
1706                         }
1707
1708                         if ($edit->orig)
1709                         $xi += sizeof($edit->orig);
1710                         if ($edit->closing)
1711                         $yi += sizeof($edit->closing);
1712                 }
1713
1714                 if (is_array($block))
1715                 $this->_block($x0, $xi - $x0,
1716                 $y0, $yi - $y0,
1717                 $block);
1718
1719                 $end = $this->_end_diff();
1720                 wfProfileOut( __METHOD__ );
1721                 return $end;
1722         }
1723
1724         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1725                 wfProfileIn( __METHOD__ );
1726                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1727                 foreach ($edits as $edit) {
1728                         if ($edit->type == 'copy')
1729                         $this->_context($edit->orig);
1730                         elseif ($edit->type == 'add')
1731                         $this->_added($edit->closing);
1732                         elseif ($edit->type == 'delete')
1733                         $this->_deleted($edit->orig);
1734                         elseif ($edit->type == 'change')
1735                         $this->_changed($edit->orig, $edit->closing);
1736                         else
1737                         trigger_error('Unknown edit type', E_USER_ERROR);
1738                 }
1739                 $this->_end_block();
1740                 wfProfileOut( __METHOD__ );
1741         }
1742
1743         function _start_diff() {
1744                 ob_start();
1745         }
1746
1747         function _end_diff() {
1748                 $val = ob_get_contents();
1749                 ob_end_clean();
1750                 return $val;
1751         }
1752
1753         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1754                 if ($xlen > 1)
1755                 $xbeg .= "," . ($xbeg + $xlen - 1);
1756                 if ($ylen > 1)
1757                 $ybeg .= "," . ($ybeg + $ylen - 1);
1758
1759                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1760         }
1761
1762         function _start_block($header) {
1763                 echo $header . "\n";
1764         }
1765
1766         function _end_block() {
1767         }
1768
1769         function _lines($lines, $prefix = ' ') {
1770                 foreach ($lines as $line)
1771                 echo "$prefix $line\n";
1772         }
1773
1774         function _context($lines) {
1775                 $this->_lines($lines);
1776         }
1777
1778         function _added($lines) {
1779                 $this->_lines($lines, '>');
1780         }
1781         function _deleted($lines) {
1782                 $this->_lines($lines, '<');
1783         }
1784
1785         function _changed($orig, $closing) {
1786                 $this->_deleted($orig);
1787                 echo "---\n";
1788                 $this->_added($closing);
1789         }
1790 }
1791
1792 /**
1793  * A formatter that outputs unified diffs
1794  * @ingroup DifferenceEngine
1795  */
1796
1797 class UnifiedDiffFormatter extends DiffFormatter {
1798         var $leading_context_lines = 2;
1799         var $trailing_context_lines = 2;
1800
1801         function _added($lines) {
1802                 $this->_lines($lines, '+');
1803         }
1804         function _deleted($lines) {
1805                 $this->_lines($lines, '-');
1806         }
1807         function _changed($orig, $closing) {
1808                 $this->_deleted($orig);
1809                 $this->_added($closing);
1810         }
1811         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1812                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1813         }
1814 }
1815
1816 /**
1817  * A pseudo-formatter that just passes along the Diff::$edits array
1818  * @ingroup DifferenceEngine
1819  */
1820 class ArrayDiffFormatter extends DiffFormatter {
1821         function format($diff) {
1822                 $oldline = 1;
1823                 $newline = 1;
1824                 $retval = array();
1825                 foreach($diff->edits as $edit)
1826                 switch($edit->type) {
1827                         case 'add':
1828                                 foreach($edit->closing as $l) {
1829                                         $retval[] = array(
1830                                                         'action' => 'add',
1831                                                         'new'=> $l,
1832                                                         'newline' => $newline++
1833                                         );
1834                                 }
1835                                 break;
1836                         case 'delete':
1837                                 foreach($edit->orig as $l) {
1838                                         $retval[] = array(
1839                                                         'action' => 'delete',
1840                                                         'old' => $l,
1841                                                         'oldline' => $oldline++,
1842                                         );
1843                                 }
1844                                 break;
1845                         case 'change':
1846                                 foreach($edit->orig as $i => $l) {
1847                                         $retval[] = array(
1848                                                         'action' => 'change',
1849                                                         'old' => $l,
1850                                                         'new' => @$edit->closing[$i],
1851                                                         'oldline' => $oldline++,
1852                                                         'newline' => $newline++,
1853                                         );
1854                                 }
1855                                 break;
1856                         case 'copy':
1857                                 $oldline += count($edit->orig);
1858                                 $newline += count($edit->orig);
1859                 }
1860                 return $retval;
1861         }
1862 }
1863
1864 /**
1865  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1866  *
1867  */
1868
1869 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1870
1871 /**
1872  * @todo document
1873  * @private
1874  * @ingroup DifferenceEngine
1875  */
1876 class _HWLDF_WordAccumulator {
1877         function _HWLDF_WordAccumulator () {
1878                 $this->_lines = array();
1879                 $this->_line = '';
1880                 $this->_group = '';
1881                 $this->_tag = '';
1882         }
1883
1884         function _flushGroup ($new_tag) {
1885                 if ($this->_group !== '') {
1886                         if ($this->_tag == 'ins')
1887                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1888                         htmlspecialchars ( $this->_group ) . '</ins>';
1889                         elseif ($this->_tag == 'del')
1890                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1891                         htmlspecialchars ( $this->_group ) . '</del>';
1892                         else
1893                         $this->_line .= htmlspecialchars ( $this->_group );
1894                 }
1895                 $this->_group = '';
1896                 $this->_tag = $new_tag;
1897         }
1898
1899         function _flushLine ($new_tag) {
1900                 $this->_flushGroup($new_tag);
1901                 if ($this->_line != '')
1902                 array_push ( $this->_lines, $this->_line );
1903                 else
1904                 # make empty lines visible by inserting an NBSP
1905                 array_push ( $this->_lines, NBSP );
1906                 $this->_line = '';
1907         }
1908
1909         function addWords ($words, $tag = '') {
1910                 if ($tag != $this->_tag)
1911                 $this->_flushGroup($tag);
1912
1913                 foreach ($words as $word) {
1914                         // new-line should only come as first char of word.
1915                         if ($word == '')
1916                         continue;
1917                         if ($word[0] == "\n") {
1918                                 $this->_flushLine($tag);
1919                                 $word = substr($word, 1);
1920                         }
1921                         assert(!strstr($word, "\n"));
1922                         $this->_group .= $word;
1923                 }
1924         }
1925
1926         function getLines() {
1927                 $this->_flushLine('~done');
1928                 return $this->_lines;
1929         }
1930 }
1931
1932 /**
1933  * @todo document
1934  * @private
1935  * @ingroup DifferenceEngine
1936  */
1937 class WordLevelDiff extends MappedDiff {
1938         const MAX_LINE_LENGTH = 10000;
1939
1940         function WordLevelDiff ($orig_lines, $closing_lines) {
1941                 wfProfileIn( __METHOD__ );
1942
1943                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1944                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1945
1946                 $this->MappedDiff($orig_words, $closing_words,
1947                 $orig_stripped, $closing_stripped);
1948                 wfProfileOut( __METHOD__ );
1949         }
1950
1951         function _split($lines) {
1952                 wfProfileIn( __METHOD__ );
1953
1954                 $words = array();
1955                 $stripped = array();
1956                 $first = true;
1957                 foreach ( $lines as $line ) {
1958                         # If the line is too long, just pretend the entire line is one big word
1959                         # This prevents resource exhaustion problems
1960                         if ( $first ) {
1961                                 $first = false;
1962                         } else {
1963                                 $words[] = "\n";
1964                                 $stripped[] = "\n";
1965                         }
1966                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1967                                 $words[] = $line;
1968                                 $stripped[] = $line;
1969                         } else {
1970                                 $m = array();
1971                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1972                                 $line, $m))
1973                                 {
1974                                         $words = array_merge( $words, $m[0] );
1975                                         $stripped = array_merge( $stripped, $m[1] );
1976                                 }
1977                         }
1978                 }
1979                 wfProfileOut( __METHOD__ );
1980                 return array($words, $stripped);
1981         }
1982
1983         function orig () {
1984                 wfProfileIn( __METHOD__ );
1985                 $orig = new _HWLDF_WordAccumulator;
1986
1987                 foreach ($this->edits as $edit) {
1988                         if ($edit->type == 'copy')
1989                         $orig->addWords($edit->orig);
1990                         elseif ($edit->orig)
1991                         $orig->addWords($edit->orig, 'del');
1992                 }
1993                 $lines = $orig->getLines();
1994                 wfProfileOut( __METHOD__ );
1995                 return $lines;
1996         }
1997
1998         function closing () {
1999                 wfProfileIn( __METHOD__ );
2000                 $closing = new _HWLDF_WordAccumulator;
2001
2002                 foreach ($this->edits as $edit) {
2003                         if ($edit->type == 'copy')
2004                         $closing->addWords($edit->closing);
2005                         elseif ($edit->closing)
2006                         $closing->addWords($edit->closing, 'ins');
2007                 }
2008                 $lines = $closing->getLines();
2009                 wfProfileOut( __METHOD__ );
2010                 return $lines;
2011         }
2012 }
2013
2014 /**
2015  * Wikipedia Table style diff formatter.
2016  * @todo document
2017  * @private
2018  * @ingroup DifferenceEngine
2019  */
2020 class TableDiffFormatter extends DiffFormatter {
2021         function TableDiffFormatter() {
2022                 $this->leading_context_lines = 2;
2023                 $this->trailing_context_lines = 2;
2024         }
2025
2026         public static function escapeWhiteSpace( $msg ) {
2027                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
2028                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
2029                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
2030                 return $msg;
2031         }
2032
2033         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
2034                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
2035                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
2036                 return $r;
2037         }
2038
2039         function _start_block( $header ) {
2040                 echo $header;
2041         }
2042
2043         function _end_block() {
2044         }
2045
2046         function _lines( $lines, $prefix=' ', $color='white' ) {
2047         }
2048
2049         # HTML-escape parameter before calling this
2050         function addedLine( $line ) {
2051                 return $this->wrapLine( '+', 'diff-addedline', $line );
2052         }
2053
2054         # HTML-escape parameter before calling this
2055         function deletedLine( $line ) {
2056                 return $this->wrapLine( '-', 'diff-deletedline', $line );
2057         }
2058
2059         # HTML-escape parameter before calling this
2060         function contextLine( $line ) {
2061                 return $this->wrapLine( ' ', 'diff-context', $line );
2062         }
2063
2064         private function wrapLine( $marker, $class, $line ) {
2065                 if( $line !== '' ) {
2066                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
2067                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
2068                 }
2069                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
2070         }
2071
2072         function emptyLine() {
2073                 return '<td colspan="2">&nbsp;</td>';
2074         }
2075
2076         function _added( $lines ) {
2077                 foreach ($lines as $line) {
2078                         echo '<tr>' . $this->emptyLine() .
2079                         $this->addedLine( '<ins class="diffchange">' .
2080                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
2081                 }
2082         }
2083
2084         function _deleted($lines) {
2085                 foreach ($lines as $line) {
2086                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
2087                         htmlspecialchars ( $line ) . '</del>' ) .
2088                         $this->emptyLine() . "</tr>\n";
2089                 }
2090         }
2091
2092         function _context( $lines ) {
2093                 foreach ($lines as $line) {
2094                         echo '<tr>' .
2095                         $this->contextLine( htmlspecialchars ( $line ) ) .
2096                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
2097                 }
2098         }
2099
2100         function _changed( $orig, $closing ) {
2101                 wfProfileIn( __METHOD__ );
2102
2103                 $diff = new WordLevelDiff( $orig, $closing );
2104                 $del = $diff->orig();
2105                 $add = $diff->closing();
2106
2107                 # Notice that WordLevelDiff returns HTML-escaped output.
2108                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2109
2110                 while ( $line = array_shift( $del ) ) {
2111                         $aline = array_shift( $add );
2112                         echo '<tr>' . $this->deletedLine( $line ) .
2113                         $this->addedLine( $aline ) . "</tr>\n";
2114                 }
2115                 foreach ($add as $line) {       # If any leftovers
2116                         echo '<tr>' . $this->emptyLine() .
2117                         $this->addedLine( $line ) . "</tr>\n";
2118                 }
2119                 wfProfileOut( __METHOD__ );
2120         }
2121 }