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