includes/DifferenceEngine.php

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