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