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          */
  42         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0 ) {
  43                 $this->mTitle = $titleObj;
  44                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  45
  46                 if ( 'prev' === $new ) {
  47                         # Show diff between revision $old and the previous one.
  48                         # Get previous one from DB.
  49                         #
  50                         $this->mNewid = intval($old);
  51
  52                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  53
  54                 } elseif ( 'next' === $new ) {
  55                         # Show diff between revision $old and the previous one.
  56                         # Get previous one from DB.
  57                         #
  58                         $this->mOldid = intval($old);
  59                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  60                         if ( false === $this->mNewid ) {
  61                                 # if no result, NewId points to the newest old revision. The only newer
  62                                 # revision is cur, which is "0".
  63                                 $this->mNewid = 0;
  64                         }
  65
  66                 } else {
  67                         $this->mOldid = intval($old);
  68                         $this->mNewid = intval($new);
  69                 }
  70                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  71         }
  72
  73         function showDiffPage( $diffOnly = false ) {
  74                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  75                 $fname = 'DifferenceEngine::showDiffPage';
  76                 wfProfileIn( $fname );
  77
  78                 # If external diffs are enabled both globally and for the user,
  79                 # we'll use the application/x-external-editor interface to call
  80                 # an external diff tool like kompare, kdiff3, etc.
  81                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  82                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  83                         $wgOut->disable();
  84                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  85                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  86                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  87                         $special=$wgLang->getNsText(NS_SPECIAL);
  88                         $control=<<<CONTROL
  89 [Process]
  90 Type=Diff text
  91 Engine=MediaWiki
  92 Script={$wgServer}{$wgScript}
  93 Special namespace={$special}
  94
  95 [File]
  96 Extension=wiki
  97 URL=$url1
  98
  99 [File 2]
 100 Extension=wiki
 101 URL=$url2
 102 CONTROL;
 103                         echo($control);
 104                         return;
 105                 }
 106
 107                 $wgOut->setArticleFlag( false );
 108                 if ( ! $this->loadRevisionData() ) {
 109                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, {$this->mNewid})";
 110                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 111                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 112                         $wgOut->addWikitext( $mtext );
 113                         wfProfileOut( $fname );
 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( $fname );
 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( $fname );
 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                 if( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isAllowed( 'patrol' ) ) {
 160                         $patrol = ' [' . $sk->makeKnownLinkObj( $this->mTitle, wfMsg( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 161                 } else {
 162                         $patrol = '';
 163                 }
 164
 165                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 166                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 167                 if ( $this->mNewRev->isCurrent() ) {
 168                         $nextlink = '&nbsp;';
 169                 } else {
 170                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 171                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 172                 }
 173
 174                 $oldminor = '';
 175                 $newminor = '';
 176
 177                 if ($this->mOldRev->mMinorEdit == 1) {
 178                         $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
 179                                 wfMsg( 'minoreditletter') ) . ' ';
 180                 }
 181
 182                 if ($this->mNewRev->mMinorEdit == 1) {
 183                         $newminor = wfElement( 'span', array( 'class' => 'minor' ),
 184                         wfMsg( 'minoreditletter') ) . ' ';
 185                 }
 186
 187                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />" .
 188                         $sk->revUserTools( $this->mOldRev ) . "<br />" .
 189                         $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly ) . "<br />" .
 190                         $prevlink;
 191                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />" .
 192                         $sk->revUserTools( $this->mNewRev ) . " $rollback<br />" .
 193                         $newminor . $sk->revComment( $this->mNewRev, !$diffOnly ) . "<br />" .
 194                         $nextlink . $patrol;
 195
 196                 $this->showDiff( $oldHeader, $newHeader );
 197
 198                 if ( !$diffOnly )
 199                         $this->renderNewRevision();
 200
 201                 wfProfileOut( $fname );
 202         }
 203
 204         /**
 205          * Show the new revision of the page.
 206          */
 207         function renderNewRevision() {
 208                 global $wgOut;
 209                 $fname = 'DifferenceEngine::renderNewRevision';
 210                 wfProfileIn( $fname );
 211
 212                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 213                 #add deleted rev tag if needed
 214                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 215                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-permission' ) );
 216                 }
 217
 218                 if( !$this->mNewRev->isCurrent() ) {
 219                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 220                 }
 221
 222                 $this->loadNewText();
 223                 if( is_object( $this->mNewRev ) ) {
 224                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 225                 }
 226
 227                 $wgOut->addWikiTextTidy( $this->mNewtext );
 228
 229                 if( !$this->mNewRev->isCurrent() ) {
 230                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 231                 }
 232
 233                 wfProfileOut( $fname );
 234         }
 235
 236         /**
 237          * Show the first revision of an article. Uses normal diff headers in
 238          * contrast to normal "old revision" display style.
 239          */
 240         function showFirstRevision() {
 241                 global $wgOut, $wgUser;
 242
 243                 $fname = 'DifferenceEngine::showFirstRevision';
 244                 wfProfileIn( $fname );
 245
 246                 # Get article text from the DB
 247                 #
 248                 if ( ! $this->loadNewText() ) {
 249                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 250                           "{$this->mNewid})";
 251                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 252                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 253                         $wgOut->addWikitext( $mtext );
 254                         wfProfileOut( $fname );
 255                         return;
 256                 }
 257                 if ( $this->mNewRev->isCurrent() ) {
 258                         $wgOut->setArticleFlag( true );
 259                 }
 260
 261                 # Check if user is allowed to look at this page. If not, bail out.
 262                 #
 263                 if ( !( $this->mTitle->userCanRead() ) ) {
 264                         $wgOut->loginToUse();
 265                         $wgOut->output();
 266                         wfProfileOut( $fname );
 267                         exit;
 268                 }
 269
 270                 # Prepare the header box
 271                 #
 272                 $sk = $wgUser->getSkin();
 273
 274                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 275                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 276                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 277                         $sk->revComment( $this->mNewRev ) . "<br />" .
 278                         $nextlink . "</div>\n";
 279
 280                 $wgOut->addHTML( $header );
 281
 282                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 283                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 284
 285                 wfProfileOut( $fname );
 286         }
 287
 288         /**
 289          * Get the diff text, send it to $wgOut
 290          * Returns false if the diff could not be generated, otherwise returns true
 291          */
 292         function showDiff( $otitle, $ntitle ) {
 293                 global $wgOut;
 294                 $diff = $this->getDiff( $otitle, $ntitle );
 295                 if ( $diff === false ) {
 296                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
 297                         return false;
 298                 } else {
 299                         $wgOut->addStyle( 'common/diff.css' );
 300                         $wgOut->addHTML( $diff );
 301                         return true;
 302                 }
 303         }
 304
 305         /**
 306          * Get diff table, including header
 307          * Note that the interface has changed, it's no longer static.
 308          * Returns false on error
 309          */
 310         function getDiff( $otitle, $ntitle ) {
 311                 $body = $this->getDiffBody();
 312                 if ( $body === false ) {
 313                         return false;
 314                 } else {
 315                         $multi = $this->getMultiNotice();
 316                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 317                 }
 318         }
 319
 320         /**
 321          * Get the diff table body, without header
 322          * Results are cached
 323          * Returns false on error
 324          */
 325         function getDiffBody() {
 326                 global $wgMemc;
 327                 $fname = 'DifferenceEngine::getDiffBody';
 328                 wfProfileIn( $fname );
 329
 330                 // Cacheable?
 331                 $key = false;
 332                 if ( $this->mOldid && $this->mNewid ) {
 333                         // Try cache
 334                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 335                         $difftext = $wgMemc->get( $key );
 336                         if ( $difftext ) {
 337                                 wfIncrStats( 'diff_cache_hit' );
 338                                 $difftext = $this->localiseLineNumbers( $difftext );
 339                                 $difftext .= "\n<!-- diff cache key $key -->\n";
 340                                 wfProfileOut( $fname );
 341                                 return $difftext;
 342                         }
 343                 }
 344
 345                 #loadtext is permission safe, this just clears out the diff
 346                 if ( !$this->loadText() ) {
 347                         wfProfileOut( $fname );
 348                         return false;
 349                 } else if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 350                   return '';
 351                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 352                   return '';
 353                 }
 354
 355                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 356
 357                 // Save to cache for 7 days
 358                 if ( $key !== false && $difftext !== false ) {
 359                         wfIncrStats( 'diff_cache_miss' );
 360                         $wgMemc->set( $key, $difftext, 7*86400 );
 361                 } else {
 362                         wfIncrStats( 'diff_uncacheable' );
 363                 }
 364                 // Replace line numbers with the text in the user's language
 365                 if ( $difftext !== false ) {
 366                         $difftext = $this->localiseLineNumbers( $difftext );
 367                 }
 368                 wfProfileOut( $fname );
 369                 return $difftext;
 370         }
 371
 372         /**
 373          * Generate a diff, no caching
 374          * $otext and $ntext must be already segmented
 375          */
 376         function generateDiffBody( $otext, $ntext ) {
 377                 global $wgExternalDiffEngine, $wgContLang;
 378                 $fname = 'DifferenceEngine::generateDiffBody';
 379
 380                 $otext = str_replace( "\r\n", "\n", $otext );
 381                 $ntext = str_replace( "\r\n", "\n", $ntext );
 382
 383                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 384                         # For historical reasons, external diff engine expects
 385                         # input text to be HTML-escaped already
 386                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 387                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 388                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 389                                 dl('php_wikidiff.so');
 390                         }
 391                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 392                 }
 393
 394                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 395                         # Better external diff engine, the 2 may some day be dropped
 396                         # This one does the escaping and segmenting itself
 397                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 398                                 wfProfileIn( "$fname-dl" );
 399                                 @dl('php_wikidiff2.so');
 400                                 wfProfileOut( "$fname-dl" );
 401                         }
 402                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 403                                 wfProfileIn( 'wikidiff2_do_diff' );
 404                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 405                                 wfProfileOut( 'wikidiff2_do_diff' );
 406                                 return $text;
 407                         }
 408                 }
 409                 if ( $wgExternalDiffEngine !== false ) {
 410                         # Diff via the shell
 411                         global $wgTmpDirectory;
 412                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 413                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 414
 415                         $tempFile1 = fopen( $tempName1, "w" );
 416                         if ( !$tempFile1 ) {
 417                                 wfProfileOut( $fname );
 418                                 return false;
 419                         }
 420                         $tempFile2 = fopen( $tempName2, "w" );
 421                         if ( !$tempFile2 ) {
 422                                 wfProfileOut( $fname );
 423                                 return false;
 424                         }
 425                         fwrite( $tempFile1, $otext );
 426                         fwrite( $tempFile2, $ntext );
 427                         fclose( $tempFile1 );
 428                         fclose( $tempFile2 );
 429                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 430                         wfProfileIn( "$fname-shellexec" );
 431                         $difftext = wfShellExec( $cmd );
 432                         wfProfileOut( "$fname-shellexec" );
 433                         unlink( $tempName1 );
 434                         unlink( $tempName2 );
 435                         return $difftext;
 436                 }
 437
 438                 # Native PHP diff
 439                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 440                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 441                 $diffs = new Diff( $ota, $nta );
 442                 $formatter = new TableDiffFormatter();
 443                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 444         }
 445
 446
 447         /**
 448          * Replace line numbers with the text in the user's language
 449          */
 450         function localiseLineNumbers( $text ) {
 451                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 452                         array( &$this, 'localiseLineNumbersCb' ), $text );
 453         }
 454
 455         function localiseLineNumbersCb( $matches ) {
 456                 global $wgLang;
 457                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 458         }
 459
 460
 461         /**
 462          * If there are revisions between the ones being compared, return a note saying so.
 463          */
 464         function getMultiNotice() {
 465                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 466                         return '';
 467
 468                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 469                         // Comparing two different pages? Count would be meaningless.
 470                         return '';
 471                 }
 472
 473                 $oldid = $this->mOldRev->getId();
 474                 $newid = $this->mNewRev->getId();
 475                 if ( $oldid > $newid ) {
 476                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 477                 }
 478
 479                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 480                 if ( !$n )
 481                         return '';
 482
 483                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 484         }
 485
 486
 487         /**
 488          * Add the header to a diff body
 489          */
 490         function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 491                 global $wgOut;
 492
 493                 if ( $this->mOldRev && $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 494                    $otitle = '<span class="history-deleted">'.$otitle.'</span>';
 495                 }
 496                 if ( $this->mNewRev && $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 497                    $ntitle = '<span class="history-deleted">'.$ntitle.'</span>';
 498                 }
 499                 $header = "
 500                         <table class='diff'>
 501                         <col class='diff-marker' />
 502                         <col class='diff-content' />
 503                         <col class='diff-marker' />
 504                         <col class='diff-content' />
 505                         <tr>
 506                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 507                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 508                         </tr>
 509                 ";
 510
 511                 if ( $multi != '' )
 512                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 513
 514                 return $header . $diff . "</table>";
 515         }
 516
 517         /**
 518          * Use specified text instead of loading from the database
 519          */
 520         function setText( $oldText, $newText ) {
 521                 $this->mOldtext = $oldText;
 522                 $this->mNewtext = $newText;
 523                 $this->mTextLoaded = 2;
 524         }
 525
 526         /**
 527          * Load revision metadata for the specified articles. If newid is 0, then compare
 528          * the old article in oldid to the current article; if oldid is 0, then
 529          * compare the current article to the immediately previous one (ignoring the
 530          * value of newid).
 531          *
 532          * If oldid is false, leave the corresponding revision object set
 533          * to false. This is impossible via ordinary user input, and is provided for
 534          * API convenience.
 535          */
 536         function loadRevisionData() {
 537                 global $wgLang;
 538                 if ( $this->mRevisionsLoaded ) {
 539                         return true;
 540                 } else {
 541                         // Whether it succeeds or fails, we don't want to try again
 542                         $this->mRevisionsLoaded = true;
 543                 }
 544
 545                 // Load the new revision object
 546                 if( $this->mNewid ) {
 547                         $this->mNewRev = Revision::newFromId( $this->mNewid );
 548                 } else {
 549                         $this->mNewRev = Revision::newFromTitle( $this->mTitle );
 550                 }
 551
 552                 if( is_null( $this->mNewRev ) ) {
 553                         return false;
 554                 }
 555
 556                 // Set assorted variables
 557                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 558                 $this->mNewPage = $this->mNewRev->getTitle();
 559                 if( $this->mNewRev->isCurrent() ) {
 560                         $newLink = $this->mNewPage->escapeLocalUrl();
 561                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 562                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 563
 564                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
 565                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 566
 567                 } else {
 568                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 569                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 570                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 571
 572                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
 573                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 574                 }
 575
 576                 // Load the old revision object
 577                 $this->mOldRev = false;
 578                 if( $this->mOldid ) {
 579                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 580                 } elseif ( $this->mOldid === 0 ) {
 581                         $rev = $this->mNewRev->getPrevious();
 582                         if( $rev ) {
 583                                 $this->mOldid = $rev->getId();
 584                                 $this->mOldRev = $rev;
 585                         } else {
 586                                 // No previous revision; mark to show as first-version only.
 587                                 $this->mOldid = false;
 588                                 $this->mOldRev = false;
 589                         }
 590                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 591
 592                 if( is_null( $this->mOldRev ) ) {
 593                         return false;
 594                 }
 595
 596                 if ( $this->mOldRev ) {
 597                         $this->mOldPage = $this->mOldRev->getTitle();
 598
 599                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 600                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 601                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 602                         $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) )
 603                                 . "</a> (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 604                         //now that we considered old rev, we can make undo link (bug 8133, multi-edit undo)
 605                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 606                         $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 607                 }
 608
 609                 return true;
 610         }
 611
 612         /**
 613          * Load the text of the revisions, as well as revision data.
 614          */
 615         function loadText() {
 616                 if ( $this->mTextLoaded == 2 ) {
 617                         return true;
 618                 } else {
 619                         // Whether it succeeds or fails, we don't want to try again
 620                         $this->mTextLoaded = 2;
 621                 }
 622
 623                 if ( !$this->loadRevisionData() ) {
 624                         return false;
 625                 }
 626                 if ( $this->mOldRev ) {
 627                         // FIXME: permission tests
 628                         $this->mOldtext = $this->mOldRev->revText();
 629                         if ( $this->mOldtext === false ) {
 630                                 return false;
 631                         }
 632                 }
 633                 if ( $this->mNewRev ) {
 634                         $this->mNewtext = $this->mNewRev->revText();
 635                         if ( $this->mNewtext === false ) {
 636                                 return false;
 637                         }
 638                 }
 639                 return true;
 640         }
 641
 642         /**
 643          * Load the text of the new revision, not the old one
 644          */
 645         function loadNewText() {
 646                 if ( $this->mTextLoaded >= 1 ) {
 647                         return true;
 648                 } else {
 649                         $this->mTextLoaded = 1;
 650                 }
 651                 if ( !$this->loadRevisionData() ) {
 652                         return false;
 653                 }
 654                 $this->mNewtext = $this->mNewRev->getText();
 655                 return true;
 656         }
 657
 658
 659 }
 660
 661 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 662 //
 663 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 664 // You may copy this code freely under the conditions of the GPL.
 665 //
 666
 667 define('USE_ASSERTS', function_exists('assert'));
 668
 669 /**
 670  * @todo document
 671  * @private
 672  * @addtogroup DifferenceEngine
 673  */
 674 class _DiffOp {
 675         var $type;
 676         var $orig;
 677         var $closing;
 678
 679         function reverse() {
 680                 trigger_error('pure virtual', E_USER_ERROR);
 681         }
 682
 683         function norig() {
 684                 return $this->orig ? sizeof($this->orig) : 0;
 685         }
 686
 687         function nclosing() {
 688                 return $this->closing ? sizeof($this->closing) : 0;
 689         }
 690 }
 691
 692 /**
 693  * @todo document
 694  * @private
 695  * @addtogroup DifferenceEngine
 696  */
 697 class _DiffOp_Copy extends _DiffOp {
 698         var $type = 'copy';
 699
 700         function _DiffOp_Copy ($orig, $closing = false) {
 701                 if (!is_array($closing))
 702                         $closing = $orig;
 703                 $this->orig = $orig;
 704                 $this->closing = $closing;
 705         }
 706
 707         function reverse() {
 708                 return new _DiffOp_Copy($this->closing, $this->orig);
 709         }
 710 }
 711
 712 /**
 713  * @todo document
 714  * @private
 715  * @addtogroup DifferenceEngine
 716  */
 717 class _DiffOp_Delete extends _DiffOp {
 718         var $type = 'delete';
 719
 720         function _DiffOp_Delete ($lines) {
 721                 $this->orig = $lines;
 722                 $this->closing = false;
 723         }
 724
 725         function reverse() {
 726                 return new _DiffOp_Add($this->orig);
 727         }
 728 }
 729
 730 /**
 731  * @todo document
 732  * @private
 733  * @addtogroup DifferenceEngine
 734  */
 735 class _DiffOp_Add extends _DiffOp {
 736         var $type = 'add';
 737
 738         function _DiffOp_Add ($lines) {
 739                 $this->closing = $lines;
 740                 $this->orig = false;
 741         }
 742
 743         function reverse() {
 744                 return new _DiffOp_Delete($this->closing);
 745         }
 746 }
 747
 748 /**
 749  * @todo document
 750  * @private
 751  * @addtogroup DifferenceEngine
 752  */
 753 class _DiffOp_Change extends _DiffOp {
 754         var $type = 'change';
 755
 756         function _DiffOp_Change ($orig, $closing) {
 757                 $this->orig = $orig;
 758                 $this->closing = $closing;
 759         }
 760
 761         function reverse() {
 762                 return new _DiffOp_Change($this->closing, $this->orig);
 763         }
 764 }
 765
 766
 767 /**
 768  * Class used internally by Diff to actually compute the diffs.
 769  *
 770  * The algorithm used here is mostly lifted from the perl module
 771  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 772  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 773  *
 774  * More ideas are taken from:
 775  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 776  *
 777  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 778  * diffutils-2.7, which can be found at:
 779  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 780  *
 781  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 782  * are my own.
 783  *
 784  * Line length limits for robustness added by Tim Starling, 2005-08-31
 785  *
 786  * @author Geoffrey T. Dairiki, Tim Starling
 787  * @private
 788  * @addtogroup DifferenceEngine
 789  */
 790 class _DiffEngine
 791 {
 792         const MAX_XREF_LENGTH =  10000;
 793
 794         function diff ($from_lines, $to_lines) {
 795                 $fname = '_DiffEngine::diff';
 796                 wfProfileIn( $fname );
 797
 798                 $n_from = sizeof($from_lines);
 799                 $n_to = sizeof($to_lines);
 800
 801                 $this->xchanged = $this->ychanged = array();
 802                 $this->xv = $this->yv = array();
 803                 $this->xind = $this->yind = array();
 804                 unset($this->seq);
 805                 unset($this->in_seq);
 806                 unset($this->lcs);
 807
 808                 // Skip leading common lines.
 809                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 810                         if ($from_lines[$skip] !== $to_lines[$skip])
 811                                 break;
 812                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 813                 }
 814                 // Skip trailing common lines.
 815                 $xi = $n_from; $yi = $n_to;
 816                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 817                         if ($from_lines[$xi] !== $to_lines[$yi])
 818                                 break;
 819                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 820                 }
 821
 822                 // Ignore lines which do not exist in both files.
 823                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 824                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 825                 }
 826
 827                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 828                         $line = $to_lines[$yi];
 829                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 830                                 continue;
 831                         $yhash[$this->_line_hash($line)] = 1;
 832                         $this->yv[] = $line;
 833                         $this->yind[] = $yi;
 834                 }
 835                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 836                         $line = $from_lines[$xi];
 837                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 838                                 continue;
 839                         $this->xv[] = $line;
 840                         $this->xind[] = $xi;
 841                 }
 842
 843                 // Find the LCS.
 844                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 845
 846                 // Merge edits when possible
 847                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 848                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 849
 850                 // Compute the edit operations.
 851                 $edits = array();
 852                 $xi = $yi = 0;
 853                 while ($xi < $n_from || $yi < $n_to) {
 854                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 855                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 856
 857                         // Skip matching "snake".
 858                         $copy = array();
 859                         while ( $xi < $n_from && $yi < $n_to
 860                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 861                                 $copy[] = $from_lines[$xi++];
 862                                 ++$yi;
 863                         }
 864                         if ($copy)
 865                                 $edits[] = new _DiffOp_Copy($copy);
 866
 867                         // Find deletes & adds.
 868                         $delete = array();
 869                         while ($xi < $n_from && $this->xchanged[$xi])
 870                                 $delete[] = $from_lines[$xi++];
 871
 872                         $add = array();
 873                         while ($yi < $n_to && $this->ychanged[$yi])
 874                                 $add[] = $to_lines[$yi++];
 875
 876                         if ($delete && $add)
 877                                 $edits[] = new _DiffOp_Change($delete, $add);
 878                         elseif ($delete)
 879                                 $edits[] = new _DiffOp_Delete($delete);
 880                         elseif ($add)
 881                                 $edits[] = new _DiffOp_Add($add);
 882                 }
 883                 wfProfileOut( $fname );
 884                 return $edits;
 885         }
 886
 887         /**
 888          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 889          */
 890         function _line_hash( $line ) {
 891                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 892                         return md5( $line );
 893                 } else {
 894                         return $line;
 895                 }
 896         }
 897
 898
 899         /* Divide the Largest Common Subsequence (LCS) of the sequences
 900          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 901          * sized segments.
 902          *
 903          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 904          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 905          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 906          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 907          * that (X0, Y0) == (XOFF, YOFF) and
 908          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 909          *
 910          * This function assumes that the first lines of the specified portions
 911          * of the two files do not match, and likewise that the last lines do not
 912          * match.  The caller must trim matching lines from the beginning and end
 913          * of the portions it is going to specify.
 914          */
 915         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 916                 $fname = '_DiffEngine::_diag';
 917                 wfProfileIn( $fname );
 918                 $flip = false;
 919
 920                 if ($xlim - $xoff > $ylim - $yoff) {
 921                         // Things seems faster (I'm not sure I understand why)
 922                                 // when the shortest sequence in X.
 923                                 $flip = true;
 924                         list ($xoff, $xlim, $yoff, $ylim)
 925                         = array( $yoff, $ylim, $xoff, $xlim);
 926                 }
 927
 928                 if ($flip)
 929                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 930                                 $ymatches[$this->xv[$i]][] = $i;
 931                 else
 932                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 933                                 $ymatches[$this->yv[$i]][] = $i;
 934
 935                 $this->lcs = 0;
 936                 $this->seq[0]= $yoff - 1;
 937                 $this->in_seq = array();
 938                 $ymids[0] = array();
 939
 940                 $numer = $xlim - $xoff + $nchunks - 1;
 941                 $x = $xoff;
 942                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 943                         wfProfileIn( "$fname-chunk" );
 944                         if ($chunk > 0)
 945                                 for ($i = 0; $i <= $this->lcs; $i++)
 946                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 947
 948                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 949                         for ( ; $x < $x1; $x++) {
 950                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 951                                         if (empty($ymatches[$line]))
 952                                                 continue;
 953                                 $matches = $ymatches[$line];
 954                                 reset($matches);
 955                                 while (list ($junk, $y) = each($matches))
 956                                         if (empty($this->in_seq[$y])) {
 957                                                 $k = $this->_lcs_pos($y);
 958                                                 USE_ASSERTS && assert($k > 0);
 959                                                 $ymids[$k] = $ymids[$k-1];
 960                                                 break;
 961                                         }
 962                                 while (list ( /* $junk */, $y) = each($matches)) {
 963                                         if ($y > $this->seq[$k-1]) {
 964                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 965                                                 // Optimization: this is a common case:
 966                                                 //      next match is just replacing previous match.
 967                                                 $this->in_seq[$this->seq[$k]] = false;
 968                                                 $this->seq[$k] = $y;
 969                                                 $this->in_seq[$y] = 1;
 970                                         } else if (empty($this->in_seq[$y])) {
 971                                                 $k = $this->_lcs_pos($y);
 972                                                 USE_ASSERTS && assert($k > 0);
 973                                                 $ymids[$k] = $ymids[$k-1];
 974                                         }
 975                                 }
 976                         }
 977                         wfProfileOut( "$fname-chunk" );
 978                 }
 979
 980                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 981                 $ymid = $ymids[$this->lcs];
 982                 for ($n = 0; $n < $nchunks - 1; $n++) {
 983                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 984                         $y1 = $ymid[$n] + 1;
 985                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 986                 }
 987                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 988
 989                 wfProfileOut( $fname );
 990                 return array($this->lcs, $seps);
 991         }
 992
 993         function _lcs_pos ($ypos) {
 994                 $fname = '_DiffEngine::_lcs_pos';
 995                 wfProfileIn( $fname );
 996
 997                 $end = $this->lcs;
 998                 if ($end == 0 || $ypos > $this->seq[$end]) {
 999                         $this->seq[++$this->lcs] = $ypos;
1000                         $this->in_seq[$ypos] = 1;
1001                         wfProfileOut( $fname );
1002                         return $this->lcs;
1003                 }
1004
1005                 $beg = 1;
1006                 while ($beg < $end) {
1007                         $mid = (int)(($beg + $end) / 2);
1008                         if ( $ypos > $this->seq[$mid] )
1009                                 $beg = $mid + 1;
1010                         else
1011                                 $end = $mid;
1012                 }
1013
1014                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1015
1016                 $this->in_seq[$this->seq[$end]] = false;
1017                 $this->seq[$end] = $ypos;
1018                 $this->in_seq[$ypos] = 1;
1019                 wfProfileOut( $fname );
1020                 return $end;
1021         }
1022
1023         /* Find LCS of two sequences.
1024          *
1025          * The results are recorded in the vectors $this->{x,y}changed[], by
1026          * storing a 1 in the element for each line that is an insertion
1027          * or deletion (ie. is not in the LCS).
1028          *
1029          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1030          *
1031          * Note that XLIM, YLIM are exclusive bounds.
1032          * All line numbers are origin-0 and discarded lines are not counted.
1033          */
1034         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1035                 $fname = '_DiffEngine::_compareseq';
1036                 wfProfileIn( $fname );
1037
1038                 // Slide down the bottom initial diagonal.
1039                 while ($xoff < $xlim && $yoff < $ylim
1040                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1041                         ++$xoff;
1042                         ++$yoff;
1043                 }
1044
1045                 // Slide up the top initial diagonal.
1046                 while ($xlim > $xoff && $ylim > $yoff
1047                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1048                         --$xlim;
1049                         --$ylim;
1050                 }
1051
1052                 if ($xoff == $xlim || $yoff == $ylim)
1053                         $lcs = 0;
1054                 else {
1055                         // This is ad hoc but seems to work well.
1056                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1057                         //$nchunks = max(2,min(8,(int)$nchunks));
1058                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1059                         list ($lcs, $seps)
1060                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1061                 }
1062
1063                 if ($lcs == 0) {
1064                         // X and Y sequences have no common subsequence:
1065                         // mark all changed.
1066                         while ($yoff < $ylim)
1067                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1068                         while ($xoff < $xlim)
1069                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1070                 } else {
1071                         // Use the partitions to split this problem into subproblems.
1072                         reset($seps);
1073                         $pt1 = $seps[0];
1074                         while ($pt2 = next($seps)) {
1075                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1076                                 $pt1 = $pt2;
1077                         }
1078                 }
1079                 wfProfileOut( $fname );
1080         }
1081
1082         /* Adjust inserts/deletes of identical lines to join changes
1083          * as much as possible.
1084          *
1085          * We do something when a run of changed lines include a
1086          * line at one end and has an excluded, identical line at the other.
1087          * We are free to choose which identical line is included.
1088          * `compareseq' usually chooses the one at the beginning,
1089          * but usually it is cleaner to consider the following identical line
1090          * to be the "change".
1091          *
1092          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1093          */
1094         function _shift_boundaries ($lines, &$changed, $other_changed) {
1095                 $fname = '_DiffEngine::_shift_boundaries';
1096                 wfProfileIn( $fname );
1097                 $i = 0;
1098                 $j = 0;
1099
1100                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1101                 $len = sizeof($lines);
1102                 $other_len = sizeof($other_changed);
1103
1104                 while (1) {
1105                         /*
1106                          * Scan forwards to find beginning of another run of changes.
1107                          * Also keep track of the corresponding point in the other file.
1108                          *
1109                          * Throughout this code, $i and $j are adjusted together so that
1110                          * the first $i elements of $changed and the first $j elements
1111                          * of $other_changed both contain the same number of zeros
1112                          * (unchanged lines).
1113                          * Furthermore, $j is always kept so that $j == $other_len or
1114                          * $other_changed[$j] == false.
1115                          */
1116                         while ($j < $other_len && $other_changed[$j])
1117                                 $j++;
1118
1119                         while ($i < $len && ! $changed[$i]) {
1120                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1121                                 $i++; $j++;
1122                                 while ($j < $other_len && $other_changed[$j])
1123                                         $j++;
1124                         }
1125
1126                         if ($i == $len)
1127                                 break;
1128
1129                         $start = $i;
1130
1131                         // Find the end of this run of changes.
1132                         while (++$i < $len && $changed[$i])
1133                                 continue;
1134
1135                         do {
1136                                 /*
1137                                  * Record the length of this run of changes, so that
1138                                  * we can later determine whether the run has grown.
1139                                  */
1140                                 $runlength = $i - $start;
1141
1142                                 /*
1143                                  * Move the changed region back, so long as the
1144                                  * previous unchanged line matches the last changed one.
1145                                  * This merges with previous changed regions.
1146                                  */
1147                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1148                                         $changed[--$start] = 1;
1149                                         $changed[--$i] = false;
1150                                         while ($start > 0 && $changed[$start - 1])
1151                                                 $start--;
1152                                         USE_ASSERTS && assert('$j > 0');
1153                                         while ($other_changed[--$j])
1154                                                 continue;
1155                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1156                                 }
1157
1158                                 /*
1159                                  * Set CORRESPONDING to the end of the changed run, at the last
1160                                  * point where it corresponds to a changed run in the other file.
1161                                  * CORRESPONDING == LEN means no such point has been found.
1162                                  */
1163                                 $corresponding = $j < $other_len ? $i : $len;
1164
1165                                 /*
1166                                  * Move the changed region forward, so long as the
1167                                  * first changed line matches the following unchanged one.
1168                                  * This merges with following changed regions.
1169                                  * Do this second, so that if there are no merges,
1170                                  * the changed region is moved forward as far as possible.
1171                                  */
1172                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1173                                         $changed[$start++] = false;
1174                                         $changed[$i++] = 1;
1175                                         while ($i < $len && $changed[$i])
1176                                                 $i++;
1177
1178                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1179                                         $j++;
1180                                         if ($j < $other_len && $other_changed[$j]) {
1181                                                 $corresponding = $i;
1182                                                 while ($j < $other_len && $other_changed[$j])
1183                                                         $j++;
1184                                         }
1185                                 }
1186                         } while ($runlength != $i - $start);
1187
1188                         /*
1189                          * If possible, move the fully-merged run of changes
1190                          * back to a corresponding run in the other file.
1191                          */
1192                         while ($corresponding < $i) {
1193                                 $changed[--$start] = 1;
1194                                 $changed[--$i] = 0;
1195                                 USE_ASSERTS && assert('$j > 0');
1196                                 while ($other_changed[--$j])
1197                                         continue;
1198                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1199                         }
1200                 }
1201                 wfProfileOut( $fname );
1202         }
1203 }
1204
1205 /**
1206  * Class representing a 'diff' between two sequences of strings.
1207  * @todo document
1208  * @private
1209  * @addtogroup DifferenceEngine
1210  */
1211 class Diff
1212 {
1213         var $edits;
1214
1215         /**
1216          * Constructor.
1217          * Computes diff between sequences of strings.
1218          *
1219          * @param $from_lines array An array of strings.
1220          *                (Typically these are lines from a file.)
1221          * @param $to_lines array An array of strings.
1222          */
1223         function Diff($from_lines, $to_lines) {
1224                 $eng = new _DiffEngine;
1225                 $this->edits = $eng->diff($from_lines, $to_lines);
1226                 //$this->_check($from_lines, $to_lines);
1227         }
1228
1229         /**
1230          * Compute reversed Diff.
1231          *
1232          * SYNOPSIS:
1233          *
1234          *      $diff = new Diff($lines1, $lines2);
1235          *      $rev = $diff->reverse();
1236          * @return object A Diff object representing the inverse of the
1237          *                                original diff.
1238          */
1239         function reverse () {
1240                 $rev = $this;
1241                 $rev->edits = array();
1242                 foreach ($this->edits as $edit) {
1243                         $rev->edits[] = $edit->reverse();
1244                 }
1245                 return $rev;
1246         }
1247
1248         /**
1249          * Check for empty diff.
1250          *
1251          * @return bool True iff two sequences were identical.
1252          */
1253         function isEmpty () {
1254                 foreach ($this->edits as $edit) {
1255                         if ($edit->type != 'copy')
1256                                 return false;
1257                 }
1258                 return true;
1259         }
1260
1261         /**
1262          * Compute the length of the Longest Common Subsequence (LCS).
1263          *
1264          * This is mostly for diagnostic purposed.
1265          *
1266          * @return int The length of the LCS.
1267          */
1268         function lcs () {
1269                 $lcs = 0;
1270                 foreach ($this->edits as $edit) {
1271                         if ($edit->type == 'copy')
1272                                 $lcs += sizeof($edit->orig);
1273                 }
1274                 return $lcs;
1275         }
1276
1277         /**
1278          * Get the original set of lines.
1279          *
1280          * This reconstructs the $from_lines parameter passed to the
1281          * constructor.
1282          *
1283          * @return array The original sequence of strings.
1284          */
1285         function orig() {
1286                 $lines = array();
1287
1288                 foreach ($this->edits as $edit) {
1289                         if ($edit->orig)
1290                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1291                 }
1292                 return $lines;
1293         }
1294
1295         /**
1296          * Get the closing set of lines.
1297          *
1298          * This reconstructs the $to_lines parameter passed to the
1299          * constructor.
1300          *
1301          * @return array The sequence of strings.
1302          */
1303         function closing() {
1304                 $lines = array();
1305
1306                 foreach ($this->edits as $edit) {
1307                         if ($edit->closing)
1308                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1309                 }
1310                 return $lines;
1311         }
1312
1313         /**
1314          * Check a Diff for validity.
1315          *
1316          * This is here only for debugging purposes.
1317          */
1318         function _check ($from_lines, $to_lines) {
1319                 $fname = 'Diff::_check';
1320                 wfProfileIn( $fname );
1321                 if (serialize($from_lines) != serialize($this->orig()))
1322                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1323                 if (serialize($to_lines) != serialize($this->closing()))
1324                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1325
1326                 $rev = $this->reverse();
1327                 if (serialize($to_lines) != serialize($rev->orig()))
1328                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1329                 if (serialize($from_lines) != serialize($rev->closing()))
1330                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1331
1332
1333                 $prevtype = 'none';
1334                 foreach ($this->edits as $edit) {
1335                         if ( $prevtype == $edit->type )
1336                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1337                         $prevtype = $edit->type;
1338                 }
1339
1340                 $lcs = $this->lcs();
1341                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1342                 wfProfileOut( $fname );
1343         }
1344 }
1345
1346 /**
1347  * @todo document, bad name.
1348  * @private
1349  * @addtogroup DifferenceEngine
1350  */
1351 class MappedDiff extends Diff
1352 {
1353         /**
1354          * Constructor.
1355          *
1356          * Computes diff between sequences of strings.
1357          *
1358          * This can be used to compute things like
1359          * case-insensitve diffs, or diffs which ignore
1360          * changes in white-space.
1361          *
1362          * @param $from_lines array An array of strings.
1363          *      (Typically these are lines from a file.)
1364          *
1365          * @param $to_lines array An array of strings.
1366          *
1367          * @param $mapped_from_lines array This array should
1368          *      have the same size number of elements as $from_lines.
1369          *      The elements in $mapped_from_lines and
1370          *      $mapped_to_lines are what is actually compared
1371          *      when computing the diff.
1372          *
1373          * @param $mapped_to_lines array This array should
1374          *      have the same number of elements as $to_lines.
1375          */
1376         function MappedDiff($from_lines, $to_lines,
1377                                                 $mapped_from_lines, $mapped_to_lines) {
1378                 $fname = 'MappedDiff::MappedDiff';
1379                 wfProfileIn( $fname );
1380
1381                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1382                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1383
1384                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1385
1386                 $xi = $yi = 0;
1387                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1388                         $orig = &$this->edits[$i]->orig;
1389                         if (is_array($orig)) {
1390                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1391                                 $xi += sizeof($orig);
1392                         }
1393
1394                         $closing = &$this->edits[$i]->closing;
1395                         if (is_array($closing)) {
1396                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1397                                 $yi += sizeof($closing);
1398                         }
1399                 }
1400                 wfProfileOut( $fname );
1401         }
1402 }
1403
1404 /**
1405  * A class to format Diffs
1406  *
1407  * This class formats the diff in classic diff format.
1408  * It is intended that this class be customized via inheritance,
1409  * to obtain fancier outputs.
1410  * @todo document
1411  * @private
1412  * @addtogroup DifferenceEngine
1413  */
1414 class DiffFormatter
1415 {
1416         /**
1417          * Number of leading context "lines" to preserve.
1418          *
1419          * This should be left at zero for this class, but subclasses
1420          * may want to set this to other values.
1421          */
1422         var $leading_context_lines = 0;
1423
1424         /**
1425          * Number of trailing context "lines" to preserve.
1426          *
1427          * This should be left at zero for this class, but subclasses
1428          * may want to set this to other values.
1429          */
1430         var $trailing_context_lines = 0;
1431
1432         /**
1433          * Format a diff.
1434          *
1435          * @param $diff object A Diff object.
1436          * @return string The formatted output.
1437          */
1438         function format($diff) {
1439                 $fname = 'DiffFormatter::format';
1440                 wfProfileIn( $fname );
1441
1442                 $xi = $yi = 1;
1443                 $block = false;
1444                 $context = array();
1445
1446                 $nlead = $this->leading_context_lines;
1447                 $ntrail = $this->trailing_context_lines;
1448
1449                 $this->_start_diff();
1450
1451                 foreach ($diff->edits as $edit) {
1452                         if ($edit->type == 'copy') {
1453                                 if (is_array($block)) {
1454                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1455                                                 $block[] = $edit;
1456                                         }
1457                                         else{
1458                                                 if ($ntrail) {
1459                                                         $context = array_slice($edit->orig, 0, $ntrail);
1460                                                         $block[] = new _DiffOp_Copy($context);
1461                                                 }
1462                                                 $this->_block($x0, $ntrail + $xi - $x0,
1463                                                                           $y0, $ntrail + $yi - $y0,
1464                                                                           $block);
1465                                                 $block = false;
1466                                         }
1467                                 }
1468                                 $context = $edit->orig;
1469                         }
1470                         else {
1471                                 if (! is_array($block)) {
1472                                         $context = array_slice($context, sizeof($context) - $nlead);
1473                                         $x0 = $xi - sizeof($context);
1474                                         $y0 = $yi - sizeof($context);
1475                                         $block = array();
1476                                         if ($context)
1477                                                 $block[] = new _DiffOp_Copy($context);
1478                                 }
1479                                 $block[] = $edit;
1480                         }
1481
1482                         if ($edit->orig)
1483                                 $xi += sizeof($edit->orig);
1484                         if ($edit->closing)
1485                                 $yi += sizeof($edit->closing);
1486                 }
1487
1488                 if (is_array($block))
1489                         $this->_block($x0, $xi - $x0,
1490                                                   $y0, $yi - $y0,
1491                                                   $block);
1492
1493                 $end = $this->_end_diff();
1494                 wfProfileOut( $fname );
1495                 return $end;
1496         }
1497
1498         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1499                 $fname = 'DiffFormatter::_block';
1500                 wfProfileIn( $fname );
1501                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1502                 foreach ($edits as $edit) {
1503                         if ($edit->type == 'copy')
1504                                 $this->_context($edit->orig);
1505                         elseif ($edit->type == 'add')
1506                                 $this->_added($edit->closing);
1507                         elseif ($edit->type == 'delete')
1508                                 $this->_deleted($edit->orig);
1509                         elseif ($edit->type == 'change')
1510                                 $this->_changed($edit->orig, $edit->closing);
1511                         else
1512                                 trigger_error('Unknown edit type', E_USER_ERROR);
1513                 }
1514                 $this->_end_block();
1515                 wfProfileOut( $fname );
1516         }
1517
1518         function _start_diff() {
1519                 ob_start();
1520         }
1521
1522         function _end_diff() {
1523                 $val = ob_get_contents();
1524                 ob_end_clean();
1525                 return $val;
1526         }
1527
1528         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1529                 if ($xlen > 1)
1530                         $xbeg .= "," . ($xbeg + $xlen - 1);
1531                 if ($ylen > 1)
1532                         $ybeg .= "," . ($ybeg + $ylen - 1);
1533
1534                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1535         }
1536
1537         function _start_block($header) {
1538                 echo $header;
1539         }
1540
1541         function _end_block() {
1542         }
1543
1544         function _lines($lines, $prefix = ' ') {
1545                 foreach ($lines as $line)
1546                         echo "$prefix $line\n";
1547         }
1548
1549         function _context($lines) {
1550                 $this->_lines($lines);
1551         }
1552
1553         function _added($lines) {
1554                 $this->_lines($lines, '>');
1555         }
1556         function _deleted($lines) {
1557                 $this->_lines($lines, '<');
1558         }
1559
1560         function _changed($orig, $closing) {
1561                 $this->_deleted($orig);
1562                 echo "---\n";
1563                 $this->_added($closing);
1564         }
1565 }
1566
1567
1568 /**
1569  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1570  *
1571  */
1572
1573 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1574
1575 /**
1576  * @todo document
1577  * @private
1578  * @addtogroup DifferenceEngine
1579  */
1580 class _HWLDF_WordAccumulator {
1581         function _HWLDF_WordAccumulator () {
1582                 $this->_lines = array();
1583                 $this->_line = '';
1584                 $this->_group = '';
1585                 $this->_tag = '';
1586         }
1587
1588         function _flushGroup ($new_tag) {
1589                 if ($this->_group !== '') {
1590                         if ($this->_tag == 'ins')
1591                                 $this->_line .= '<ins class="diffchange">' .
1592                                         htmlspecialchars ( $this->_group ) . '</ins>';
1593                         elseif ($this->_tag == 'del')
1594                                 $this->_line .= '<del class="diffchange">' .
1595                                         htmlspecialchars ( $this->_group ) . '</del>';
1596                         else
1597                                 $this->_line .= htmlspecialchars ( $this->_group );
1598                 }
1599                 $this->_group = '';
1600                 $this->_tag = $new_tag;
1601         }
1602
1603         function _flushLine ($new_tag) {
1604                 $this->_flushGroup($new_tag);
1605                 if ($this->_line != '')
1606                         array_push ( $this->_lines, $this->_line );
1607                 else
1608                         # make empty lines visible by inserting an NBSP
1609                         array_push ( $this->_lines, NBSP );
1610                 $this->_line = '';
1611         }
1612
1613         function addWords ($words, $tag = '') {
1614                 if ($tag != $this->_tag)
1615                         $this->_flushGroup($tag);
1616
1617                 foreach ($words as $word) {
1618                         // new-line should only come as first char of word.
1619                         if ($word == '')
1620                                 continue;
1621                         if ($word[0] == "\n") {
1622                                 $this->_flushLine($tag);
1623                                 $word = substr($word, 1);
1624                         }
1625                         assert(!strstr($word, "\n"));
1626                         $this->_group .= $word;
1627                 }
1628         }
1629
1630         function getLines() {
1631                 $this->_flushLine('~done');
1632                 return $this->_lines;
1633         }
1634 }
1635
1636 /**
1637  * @todo document
1638  * @private
1639  * @addtogroup DifferenceEngine
1640  */
1641 class WordLevelDiff extends MappedDiff
1642 {
1643         const MAX_LINE_LENGTH = 10000;
1644
1645         function WordLevelDiff ($orig_lines, $closing_lines) {
1646                 $fname = 'WordLevelDiff::WordLevelDiff';
1647                 wfProfileIn( $fname );
1648
1649                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1650                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1651
1652                 $this->MappedDiff($orig_words, $closing_words,
1653                                                   $orig_stripped, $closing_stripped);
1654                 wfProfileOut( $fname );
1655         }
1656
1657         function _split($lines) {
1658                 $fname = 'WordLevelDiff::_split';
1659                 wfProfileIn( $fname );
1660
1661                 $words = array();
1662                 $stripped = array();
1663                 $first = true;
1664                 foreach ( $lines as $line ) {
1665                         # If the line is too long, just pretend the entire line is one big word
1666                         # This prevents resource exhaustion problems
1667                         if ( $first ) {
1668                                 $first = false;
1669                         } else {
1670                                 $words[] = "\n";
1671                                 $stripped[] = "\n";
1672                         }
1673                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1674                                 $words[] = $line;
1675                                 $stripped[] = $line;
1676                         } else {
1677                                 $m = array();
1678                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1679                                         $line, $m))
1680                                 {
1681                                         $words = array_merge( $words, $m[0] );
1682                                         $stripped = array_merge( $stripped, $m[1] );
1683                                 }
1684                         }
1685                 }
1686                 wfProfileOut( $fname );
1687                 return array($words, $stripped);
1688         }
1689
1690         function orig () {
1691                 $fname = 'WordLevelDiff::orig';
1692                 wfProfileIn( $fname );
1693                 $orig = new _HWLDF_WordAccumulator;
1694
1695                 foreach ($this->edits as $edit) {
1696                         if ($edit->type == 'copy')
1697                                 $orig->addWords($edit->orig);
1698                         elseif ($edit->orig)
1699                                 $orig->addWords($edit->orig, 'del');
1700                 }
1701                 $lines = $orig->getLines();
1702                 wfProfileOut( $fname );
1703                 return $lines;
1704         }
1705
1706         function closing () {
1707                 $fname = 'WordLevelDiff::closing';
1708                 wfProfileIn( $fname );
1709                 $closing = new _HWLDF_WordAccumulator;
1710
1711                 foreach ($this->edits as $edit) {
1712                         if ($edit->type == 'copy')
1713                                 $closing->addWords($edit->closing);
1714                         elseif ($edit->closing)
1715                                 $closing->addWords($edit->closing, 'ins');
1716                 }
1717                 $lines = $closing->getLines();
1718                 wfProfileOut( $fname );
1719                 return $lines;
1720         }
1721 }
1722
1723 /**
1724  *      Wikipedia Table style diff formatter.
1725  * @todo document
1726  * @private
1727  * @addtogroup DifferenceEngine
1728  */
1729 class TableDiffFormatter extends DiffFormatter
1730 {
1731         function TableDiffFormatter() {
1732                 $this->leading_context_lines = 2;
1733                 $this->trailing_context_lines = 2;
1734         }
1735
1736         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1737                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1738                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1739                 return $r;
1740         }
1741
1742         function _start_block( $header ) {
1743                 echo $header;
1744         }
1745
1746         function _end_block() {
1747         }
1748
1749         function _lines( $lines, $prefix=' ', $color='white' ) {
1750         }
1751
1752         # HTML-escape parameter before calling this
1753         function addedLine( $line ) {
1754                 return $this->wrapLine( '+', 'diff-addedline', $line );
1755         }
1756
1757         # HTML-escape parameter before calling this
1758         function deletedLine( $line ) {
1759                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1760         }
1761
1762         # HTML-escape parameter before calling this
1763         function contextLine( $line ) {
1764                 return $this->wrapLine( ' ', 'diff-context', $line );
1765         }
1766
1767         private function wrapLine( $marker, $class, $line ) {
1768                 if( $line !== '' ) {
1769                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1770                         $line = "<div>$line</div>";
1771                 }
1772                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1773         }
1774
1775         function emptyLine() {
1776                 return '<td colspan="2">&nbsp;</td>';
1777         }
1778
1779         function _added( $lines ) {
1780                 foreach ($lines as $line) {
1781                         echo '<tr>' . $this->emptyLine() .
1782                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1783                 }
1784         }
1785
1786         function _deleted($lines) {
1787                 foreach ($lines as $line) {
1788                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1789                           $this->emptyLine() . "</tr>\n";
1790                 }
1791         }
1792
1793         function _context( $lines ) {
1794                 foreach ($lines as $line) {
1795                         echo '<tr>' .
1796                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1797                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1798                 }
1799         }
1800
1801         function _changed( $orig, $closing ) {
1802                 $fname = 'TableDiffFormatter::_changed';
1803                 wfProfileIn( $fname );
1804
1805                 $diff = new WordLevelDiff( $orig, $closing );
1806                 $del = $diff->orig();
1807                 $add = $diff->closing();
1808
1809                 # Notice that WordLevelDiff returns HTML-escaped output.
1810                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1811
1812                 while ( $line = array_shift( $del ) ) {
1813                         $aline = array_shift( $add );
1814                         echo '<tr>' . $this->deletedLine( $line ) .
1815                                 $this->addedLine( $aline ) . "</tr>\n";
1816                 }
1817                 foreach ($add as $line) {       # If any leftovers
1818                         echo '<tr>' . $this->emptyLine() .
1819                                 $this->addedLine( $line ) . "</tr>\n";
1820                 }
1821                 wfProfileOut( $fname );
1822         }
1823 }
1824
1825 ?>