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