includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @todo indicate where diff.doc can be found.
   5  * @addtogroup DifferenceEngine
   6  */
   7
   8 /**
   9  * Constant to indicate diff cache compatibility.
  10  * Bump this when changing the diff formatting in a way that
  11  * fixes important bugs or such to force cached diff views to
  12  * clear.
  13  */
  14 define( 'MW_DIFF_VERSION', '1.11a' );
  15
  16 /**
  17  * @todo document
  18  * @public
  19  * @addtogroup DifferenceEngine
  20  */
  21 class DifferenceEngine {
  22         /**#@+
  23          * @private
  24          */
  25         var $mOldid, $mNewid, $mTitle;
  26         var $mOldtitle, $mNewtitle, $mPagetitle;
  27         var $mOldtext, $mNewtext;
  28         var $mOldPage, $mNewPage;
  29         var $mRcidMarkPatrolled;
  30         var $mOldRev, $mNewRev;
  31         var $mRevisionsLoaded = false; // Have the revisions been loaded
  32         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  33         /**#@-*/
  34
  35         /**
  36          * Constructor
  37          * @param $titleObj Title object that the diff is associated with
  38          * @param $old Integer: old ID we want to show and diff with.
  39          * @param $new String: either 'prev' or 'next'.
  40          * @param $rcid Integer: ??? FIXME (default 0)
  41          * @param $refreshCache boolean If set, refreshes the diff cache
  42          */
  43         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0, $refreshCache = false ) {
  44                 $this->mTitle = $titleObj;
  45                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  46
  47                 if ( 'prev' === $new ) {
  48                         # Show diff between revision $old and the previous one.
  49                         # Get previous one from DB.
  50                         #
  51                         $this->mNewid = intval($old);
  52
  53                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  54
  55                 } elseif ( 'next' === $new ) {
  56                         # Show diff between revision $old and the previous one.
  57                         # Get previous one from DB.
  58                         #
  59                         $this->mOldid = intval($old);
  60                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  61                         if ( false === $this->mNewid ) {
  62                                 # if no result, NewId points to the newest old revision. The only newer
  63                                 # revision is cur, which is "0".
  64                                 $this->mNewid = 0;
  65                         }
  66
  67                 } else {
  68                         $this->mOldid = intval($old);
  69                         $this->mNewid = intval($new);
  70                 }
  71                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  72                 $this->mRefreshCache = $refreshCache;
  73         }
  74
  75         function showDiffPage( $diffOnly = false ) {
  76                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  77                 $fname = 'DifferenceEngine::showDiffPage';
  78                 wfProfileIn( $fname );
  79
  80                 # If external diffs are enabled both globally and for the user,
  81                 # we'll use the application/x-external-editor interface to call
  82                 # an external diff tool like kompare, kdiff3, etc.
  83                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  84                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  85                         $wgOut->disable();
  86                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  87                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  88                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  89                         $special=$wgLang->getNsText(NS_SPECIAL);
  90                         $control=<<<CONTROL
  91 [Process]
  92 Type=Diff text
  93 Engine=MediaWiki
  94 Script={$wgServer}{$wgScript}
  95 Special namespace={$special}
  96
  97 [File]
  98 Extension=wiki
  99 URL=$url1
 100
 101 [File 2]
 102 Extension=wiki
 103 URL=$url2
 104 CONTROL;
 105                         echo($control);
 106                         return;
 107                 }
 108
 109                 $wgOut->setArticleFlag( false );
 110                 if ( ! $this->loadRevisionData() ) {
 111                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, {$this->mNewid})";
 112                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 113                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 114                         $wgOut->addWikitext( $mtext );
 115                         wfProfileOut( $fname );
 116                         return;
 117                 }
 118
 119                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 120
 121                 if ( $this->mNewRev->isCurrent() ) {
 122                         $wgOut->setArticleFlag( true );
 123                 }
 124
 125                 # mOldid is false if the difference engine is called with a "vague" query for
 126                 # a diff between a version V and its previous version V' AND the version V
 127                 # is the first version of that article. In that case, V' does not exist.
 128                 if ( $this->mOldid === false ) {
 129                         $this->showFirstRevision();
 130                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 131                         wfProfileOut( $fname );
 132                         return;
 133                 }
 134
 135                 $wgOut->suppressQuickbar();
 136
 137                 $oldTitle = $this->mOldPage->getPrefixedText();
 138                 $newTitle = $this->mNewPage->getPrefixedText();
 139                 if( $oldTitle == $newTitle ) {
 140                         $wgOut->setPageTitle( $newTitle );
 141                 } else {
 142                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 143                 }
 144                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 145                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 146
 147                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 148                         $wgOut->loginToUse();
 149                         $wgOut->output();
 150                         wfProfileOut( $fname );
 151                         exit;
 152                 }
 153
 154                 $sk = $wgUser->getSkin();
 155
 156                 if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 157                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 158                 } else {
 159                         $rollback = '';
 160                 }
 161
 162                 // Prepare a change patrol link, if applicable
 163                 if( $wgUseRCPatrol && $wgUser->isAllowed( 'patrol' ) ) {
 164                         // If we've been given an explicit change identifier, use it; saves time
 165                         if( $this->mRcidMarkPatrolled ) {
 166                                 $rcid = $this->mRcidMarkPatrolled;
 167                         } else {
 168                                 // Look for an unpatrolled change corresponding to this diff
 169                                 $change = RecentChange::newFromConds(
 170                                         array(
 171                                                 // Add redundant timestamp condition so we can use the
 172                                                 // existing index
 173                                                 'rc_timestamp' => $this->mNewRev->getTimestamp(),
 174                                                 'rc_this_oldid' => $this->mNewid,
 175                                                 'rc_last_oldid' => $this->mOldid,
 176                                                 'rc_patrolled' => 0,
 177                                         ),
 178                                         __METHOD__
 179                                 );
 180                                 if( $change instanceof RecentChange ) {
 181                                         $rcid = $change->mAttribs['rc_id'];
 182                                 } else {
 183                                         // None found
 184                                         $rcid = 0;
 185                                 }
 186                         }
 187                         // Build the link
 188                         if( $rcid ) {
 189                                 $patrol = ' [' . $sk->makeKnownLinkObj(
 190                                         $this->mTitle,
 191                                         wfMsgHtml( 'markaspatrolleddiff' ),
 192                                         "action=markpatrolled&rcid={$rcid}"
 193                                 ) . ']';
 194                         } else {
 195                                 $patrol = '';
 196                         }
 197                 } else {
 198                         $patrol = '';
 199                 }
 200
 201                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 202                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 203                 if ( $this->mNewRev->isCurrent() ) {
 204                         $nextlink = '&nbsp;';
 205                 } else {
 206                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 207                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 208                 }
 209
 210                 $oldminor = '';
 211                 $newminor = '';
 212
 213                 if ($this->mOldRev->mMinorEdit == 1) {
 214                         $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
 215                                 wfMsg( 'minoreditletter') ) . ' ';
 216                 }
 217
 218                 if ($this->mNewRev->mMinorEdit == 1) {
 219                         $newminor = wfElement( 'span', array( 'class' => 'minor' ),
 220                         wfMsg( 'minoreditletter') ) . ' ';
 221                 }
 222
 223                 $oldHeader = '<div id="mw-diff-otitle1"><strong>' . $this->mOldtitle . '</strong></div>' .
 224                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev ) . "</div>" .
 225                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly ) . "</div>" .
 226                         '<div id="mw-diff-otitle4">' . $prevlink . '</div>';
 227                 $newHeader = '<div id="mw-diff-ntitle1"><strong>' .$this->mNewtitle . '</strong></div>' .
 228                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev ) . " $rollback</div>" .
 229                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly ) . "</div>" .
 230                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 231
 232                 $this->showDiff( $oldHeader, $newHeader );
 233
 234                 if ( !$diffOnly )
 235                         $this->renderNewRevision();
 236
 237                 wfProfileOut( $fname );
 238         }
 239
 240         /**
 241          * Show the new revision of the page.
 242          */
 243         function renderNewRevision() {
 244                 global $wgOut;
 245                 $fname = 'DifferenceEngine::renderNewRevision';
 246                 wfProfileIn( $fname );
 247
 248                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 249                 #add deleted rev tag if needed
 250                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 251                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-permission' ) );
 252                 }
 253
 254                 if( !$this->mNewRev->isCurrent() ) {
 255                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 256                 }
 257
 258                 $this->loadNewText();
 259                 if( is_object( $this->mNewRev ) ) {
 260                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 261                 }
 262
 263                 $wgOut->addWikiTextTidy( $this->mNewtext );
 264
 265                 if( !$this->mNewRev->isCurrent() ) {
 266                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 267                 }
 268
 269                 wfProfileOut( $fname );
 270         }
 271
 272         /**
 273          * Show the first revision of an article. Uses normal diff headers in
 274          * contrast to normal "old revision" display style.
 275          */
 276         function showFirstRevision() {
 277                 global $wgOut, $wgUser;
 278
 279                 $fname = 'DifferenceEngine::showFirstRevision';
 280                 wfProfileIn( $fname );
 281
 282                 # Get article text from the DB
 283                 #
 284                 if ( ! $this->loadNewText() ) {
 285                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 286                           "{$this->mNewid})";
 287                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 288                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 289                         $wgOut->addWikitext( $mtext );
 290                         wfProfileOut( $fname );
 291                         return;
 292                 }
 293                 if ( $this->mNewRev->isCurrent() ) {
 294                         $wgOut->setArticleFlag( true );
 295                 }
 296
 297                 # Check if user is allowed to look at this page. If not, bail out.
 298                 #
 299                 if ( !( $this->mTitle->userCanRead() ) ) {
 300                         $wgOut->loginToUse();
 301                         $wgOut->output();
 302                         wfProfileOut( $fname );
 303                         exit;
 304                 }
 305
 306                 # Prepare the header box
 307                 #
 308                 $sk = $wgUser->getSkin();
 309
 310                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 311                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 312                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 313                         $sk->revComment( $this->mNewRev ) . "<br />" .
 314                         $nextlink . "</div>\n";
 315
 316                 $wgOut->addHTML( $header );
 317
 318                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 319                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 320
 321                 wfProfileOut( $fname );
 322         }
 323
 324         /**
 325          * Get the diff text, send it to $wgOut
 326          * Returns false if the diff could not be generated, otherwise returns true
 327          */
 328         function showDiff( $otitle, $ntitle ) {
 329                 global $wgOut;
 330                 $diff = $this->getDiff( $otitle, $ntitle );
 331                 if ( $diff === false ) {
 332                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
 333                         return false;
 334                 } else {
 335                         $this->showDiffStyle();
 336                         $wgOut->addHTML( $diff );
 337                         return true;
 338                 }
 339         }
 340
 341         /**
 342          * Add style sheets and supporting JS for diff display.
 343          */
 344         function showDiffStyle() {
 345                 global $wgStylePath, $wgStyleVersion, $wgOut;
 346                 $wgOut->addStyle( 'common/diff.css' );
 347
 348                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 349                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 350         }
 351
 352         /**
 353          * Get complete diff table, including header
 354          *
 355          * @param Title $otitle Old title
 356          * @param Title $ntitle New title
 357          * @return mixed
 358          */
 359         function getDiff( $otitle, $ntitle ) {
 360                 $body = $this->getDiffBody();
 361                 if ( $body === false ) {
 362                         return false;
 363                 } else {
 364                         $multi = $this->getMultiNotice();
 365                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 366                 }
 367         }
 368
 369         /**
 370          * Get the diff table body, without header
 371          *
 372          * @return mixed
 373          */
 374         function getDiffBody() {
 375                 global $wgMemc;
 376                 $fname = 'DifferenceEngine::getDiffBody';
 377                 wfProfileIn( $fname );
 378
 379                 // Cacheable?
 380                 $key = false;
 381                 if ( $this->mOldid && $this->mNewid ) {
 382                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 383                         // Try cache
 384                         if ( !$this->mRefreshCache ) {
 385                                 $difftext = $wgMemc->get( $key );
 386                                 if ( $difftext ) {
 387                                         wfIncrStats( 'diff_cache_hit' );
 388                                         $difftext = $this->localiseLineNumbers( $difftext );
 389                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 390                                         wfProfileOut( $fname );
 391                                         return $difftext;
 392                                 }
 393                         } // don't try to load but save the result
 394                 }
 395
 396                 #loadtext is permission safe, this just clears out the diff
 397                 if ( !$this->loadText() ) {
 398                         wfProfileOut( $fname );
 399                         return false;
 400                 } else if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 401                   return '';
 402                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 403                   return '';
 404                 }
 405
 406                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 407
 408                 // Save to cache for 7 days
 409                 if ( $key !== false && $difftext !== false ) {
 410                         wfIncrStats( 'diff_cache_miss' );
 411                         $wgMemc->set( $key, $difftext, 7*86400 );
 412                 } else {
 413                         wfIncrStats( 'diff_uncacheable' );
 414                 }
 415                 // Replace line numbers with the text in the user's language
 416                 if ( $difftext !== false ) {
 417                         $difftext = $this->localiseLineNumbers( $difftext );
 418                 }
 419                 wfProfileOut( $fname );
 420                 return $difftext;
 421         }
 422
 423         /**
 424          * Generate a diff, no caching
 425          * $otext and $ntext must be already segmented
 426          */
 427         function generateDiffBody( $otext, $ntext ) {
 428                 global $wgExternalDiffEngine, $wgContLang;
 429                 $fname = 'DifferenceEngine::generateDiffBody';
 430
 431                 $otext = str_replace( "\r\n", "\n", $otext );
 432                 $ntext = str_replace( "\r\n", "\n", $ntext );
 433
 434                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 435                         # For historical reasons, external diff engine expects
 436                         # input text to be HTML-escaped already
 437                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 438                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 439                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 440                                 dl('php_wikidiff.so');
 441                         }
 442                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 443                 }
 444
 445                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 446                         # Better external diff engine, the 2 may some day be dropped
 447                         # This one does the escaping and segmenting itself
 448                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 449                                 wfProfileIn( "$fname-dl" );
 450                                 @dl('php_wikidiff2.so');
 451                                 wfProfileOut( "$fname-dl" );
 452                         }
 453                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 454                                 wfProfileIn( 'wikidiff2_do_diff' );
 455                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 456                                 wfProfileOut( 'wikidiff2_do_diff' );
 457                                 return $text;
 458                         }
 459                 }
 460                 if ( $wgExternalDiffEngine !== false ) {
 461                         # Diff via the shell
 462                         global $wgTmpDirectory;
 463                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 464                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 465
 466                         $tempFile1 = fopen( $tempName1, "w" );
 467                         if ( !$tempFile1 ) {
 468                                 wfProfileOut( $fname );
 469                                 return false;
 470                         }
 471                         $tempFile2 = fopen( $tempName2, "w" );
 472                         if ( !$tempFile2 ) {
 473                                 wfProfileOut( $fname );
 474                                 return false;
 475                         }
 476                         fwrite( $tempFile1, $otext );
 477                         fwrite( $tempFile2, $ntext );
 478                         fclose( $tempFile1 );
 479                         fclose( $tempFile2 );
 480                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 481                         wfProfileIn( "$fname-shellexec" );
 482                         $difftext = wfShellExec( $cmd );
 483                         wfProfileOut( "$fname-shellexec" );
 484                         unlink( $tempName1 );
 485                         unlink( $tempName2 );
 486                         return $difftext;
 487                 }
 488
 489                 # Native PHP diff
 490                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 491                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 492                 $diffs = new Diff( $ota, $nta );
 493                 $formatter = new TableDiffFormatter();
 494                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 495         }
 496
 497
 498         /**
 499          * Replace line numbers with the text in the user's language
 500          */
 501         function localiseLineNumbers( $text ) {
 502                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 503                         array( &$this, 'localiseLineNumbersCb' ), $text );
 504         }
 505
 506         function localiseLineNumbersCb( $matches ) {
 507                 global $wgLang;
 508                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 509         }
 510
 511
 512         /**
 513          * If there are revisions between the ones being compared, return a note saying so.
 514          */
 515         function getMultiNotice() {
 516                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 517                         return '';
 518
 519                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 520                         // Comparing two different pages? Count would be meaningless.
 521                         return '';
 522                 }
 523
 524                 $oldid = $this->mOldRev->getId();
 525                 $newid = $this->mNewRev->getId();
 526                 if ( $oldid > $newid ) {
 527                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 528                 }
 529
 530                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 531                 if ( !$n )
 532                         return '';
 533
 534                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 535         }
 536
 537
 538         /**
 539          * Add the header to a diff body
 540          */
 541         function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 542                 global $wgOut;
 543
 544                 if ( $this->mOldRev && $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 545                    $otitle = '<span class="history-deleted">'.$otitle.'</span>';
 546                 }
 547                 if ( $this->mNewRev && $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 548                    $ntitle = '<span class="history-deleted">'.$ntitle.'</span>';
 549                 }
 550                 $header = "
 551                         <table class='diff'>
 552                         <col class='diff-marker' />
 553                         <col class='diff-content' />
 554                         <col class='diff-marker' />
 555                         <col class='diff-content' />
 556                         <tr>
 557                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 558                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 559                         </tr>
 560                 ";
 561
 562                 if ( $multi != '' )
 563                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 564
 565                 return $header . $diff . "</table>";
 566         }
 567
 568         /**
 569          * Use specified text instead of loading from the database
 570          */
 571         function setText( $oldText, $newText ) {
 572                 $this->mOldtext = $oldText;
 573                 $this->mNewtext = $newText;
 574                 $this->mTextLoaded = 2;
 575         }
 576
 577         /**
 578          * Load revision metadata for the specified articles. If newid is 0, then compare
 579          * the old article in oldid to the current article; if oldid is 0, then
 580          * compare the current article to the immediately previous one (ignoring the
 581          * value of newid).
 582          *
 583          * If oldid is false, leave the corresponding revision object set
 584          * to false. This is impossible via ordinary user input, and is provided for
 585          * API convenience.
 586          */
 587         function loadRevisionData() {
 588                 global $wgLang;
 589                 if ( $this->mRevisionsLoaded ) {
 590                         return true;
 591                 } else {
 592                         // Whether it succeeds or fails, we don't want to try again
 593                         $this->mRevisionsLoaded = true;
 594                 }
 595
 596                 // Load the new revision object
 597                 $this->mNewRev = $this->mNewid
 598                         ? Revision::newFromId( $this->mNewid )
 599                         : Revision::newFromTitle( $this->mTitle );
 600                 if( !$this->mNewRev instanceof Revision )
 601                         return false;
 602
 603                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 604                 $this->mNewid = $this->mNewRev->getId();
 605
 606                 // Set assorted variables
 607                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 608                 $this->mNewPage = $this->mNewRev->getTitle();
 609                 if( $this->mNewRev->isCurrent() ) {
 610                         $newLink = $this->mNewPage->escapeLocalUrl();
 611                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 612                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 613
 614                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
 615                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 616
 617                 } else {
 618                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 619                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 620                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 621
 622                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
 623                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 624                 }
 625
 626                 // Load the old revision object
 627                 $this->mOldRev = false;
 628                 if( $this->mOldid ) {
 629                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 630                 } elseif ( $this->mOldid === 0 ) {
 631                         $rev = $this->mNewRev->getPrevious();
 632                         if( $rev ) {
 633                                 $this->mOldid = $rev->getId();
 634                                 $this->mOldRev = $rev;
 635                         } else {
 636                                 // No previous revision; mark to show as first-version only.
 637                                 $this->mOldid = false;
 638                                 $this->mOldRev = false;
 639                         }
 640                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 641
 642                 if( is_null( $this->mOldRev ) ) {
 643                         return false;
 644                 }
 645
 646                 if ( $this->mOldRev ) {
 647                         $this->mOldPage = $this->mOldRev->getTitle();
 648
 649                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 650                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 651                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 652                         $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) )
 653                                 . "</a> (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 654
 655                         // Add an "undo" link
 656                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 657                         $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 658                 }
 659
 660                 return true;
 661         }
 662
 663         /**
 664          * Load the text of the revisions, as well as revision data.
 665          */
 666         function loadText() {
 667                 if ( $this->mTextLoaded == 2 ) {
 668                         return true;
 669                 } else {
 670                         // Whether it succeeds or fails, we don't want to try again
 671                         $this->mTextLoaded = 2;
 672                 }
 673
 674                 if ( !$this->loadRevisionData() ) {
 675                         return false;
 676                 }
 677                 if ( $this->mOldRev ) {
 678                         // FIXME: permission tests
 679                         $this->mOldtext = $this->mOldRev->revText();
 680                         if ( $this->mOldtext === false ) {
 681                                 return false;
 682                         }
 683                 }
 684                 if ( $this->mNewRev ) {
 685                         $this->mNewtext = $this->mNewRev->revText();
 686                         if ( $this->mNewtext === false ) {
 687                                 return false;
 688                         }
 689                 }
 690                 return true;
 691         }
 692
 693         /**
 694          * Load the text of the new revision, not the old one
 695          */
 696         function loadNewText() {
 697                 if ( $this->mTextLoaded >= 1 ) {
 698                         return true;
 699                 } else {
 700                         $this->mTextLoaded = 1;
 701                 }
 702                 if ( !$this->loadRevisionData() ) {
 703                         return false;
 704                 }
 705                 $this->mNewtext = $this->mNewRev->getText();
 706                 return true;
 707         }
 708
 709
 710 }
 711
 712 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 713 //
 714 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 715 // You may copy this code freely under the conditions of the GPL.
 716 //
 717
 718 define('USE_ASSERTS', function_exists('assert'));
 719
 720 /**
 721  * @todo document
 722  * @private
 723  * @addtogroup DifferenceEngine
 724  */
 725 class _DiffOp {
 726         var $type;
 727         var $orig;
 728         var $closing;
 729
 730         function reverse() {
 731                 trigger_error('pure virtual', E_USER_ERROR);
 732         }
 733
 734         function norig() {
 735                 return $this->orig ? sizeof($this->orig) : 0;
 736         }
 737
 738         function nclosing() {
 739                 return $this->closing ? sizeof($this->closing) : 0;
 740         }
 741 }
 742
 743 /**
 744  * @todo document
 745  * @private
 746  * @addtogroup DifferenceEngine
 747  */
 748 class _DiffOp_Copy extends _DiffOp {
 749         var $type = 'copy';
 750
 751         function _DiffOp_Copy ($orig, $closing = false) {
 752                 if (!is_array($closing))
 753                         $closing = $orig;
 754                 $this->orig = $orig;
 755                 $this->closing = $closing;
 756         }
 757
 758         function reverse() {
 759                 return new _DiffOp_Copy($this->closing, $this->orig);
 760         }
 761 }
 762
 763 /**
 764  * @todo document
 765  * @private
 766  * @addtogroup DifferenceEngine
 767  */
 768 class _DiffOp_Delete extends _DiffOp {
 769         var $type = 'delete';
 770
 771         function _DiffOp_Delete ($lines) {
 772                 $this->orig = $lines;
 773                 $this->closing = false;
 774         }
 775
 776         function reverse() {
 777                 return new _DiffOp_Add($this->orig);
 778         }
 779 }
 780
 781 /**
 782  * @todo document
 783  * @private
 784  * @addtogroup DifferenceEngine
 785  */
 786 class _DiffOp_Add extends _DiffOp {
 787         var $type = 'add';
 788
 789         function _DiffOp_Add ($lines) {
 790                 $this->closing = $lines;
 791                 $this->orig = false;
 792         }
 793
 794         function reverse() {
 795                 return new _DiffOp_Delete($this->closing);
 796         }
 797 }
 798
 799 /**
 800  * @todo document
 801  * @private
 802  * @addtogroup DifferenceEngine
 803  */
 804 class _DiffOp_Change extends _DiffOp {
 805         var $type = 'change';
 806
 807         function _DiffOp_Change ($orig, $closing) {
 808                 $this->orig = $orig;
 809                 $this->closing = $closing;
 810         }
 811
 812         function reverse() {
 813                 return new _DiffOp_Change($this->closing, $this->orig);
 814         }
 815 }
 816
 817
 818 /**
 819  * Class used internally by Diff to actually compute the diffs.
 820  *
 821  * The algorithm used here is mostly lifted from the perl module
 822  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 823  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 824  *
 825  * More ideas are taken from:
 826  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 827  *
 828  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 829  * diffutils-2.7, which can be found at:
 830  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 831  *
 832  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 833  * are my own.
 834  *
 835  * Line length limits for robustness added by Tim Starling, 2005-08-31
 836  *
 837  * @author Geoffrey T. Dairiki, Tim Starling
 838  * @private
 839  * @addtogroup DifferenceEngine
 840  */
 841 class _DiffEngine
 842 {
 843         const MAX_XREF_LENGTH =  10000;
 844
 845         function diff ($from_lines, $to_lines) {
 846                 $fname = '_DiffEngine::diff';
 847                 wfProfileIn( $fname );
 848
 849                 $n_from = sizeof($from_lines);
 850                 $n_to = sizeof($to_lines);
 851
 852                 $this->xchanged = $this->ychanged = array();
 853                 $this->xv = $this->yv = array();
 854                 $this->xind = $this->yind = array();
 855                 unset($this->seq);
 856                 unset($this->in_seq);
 857                 unset($this->lcs);
 858
 859                 // Skip leading common lines.
 860                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 861                         if ($from_lines[$skip] !== $to_lines[$skip])
 862                                 break;
 863                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 864                 }
 865                 // Skip trailing common lines.
 866                 $xi = $n_from; $yi = $n_to;
 867                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 868                         if ($from_lines[$xi] !== $to_lines[$yi])
 869                                 break;
 870                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 871                 }
 872
 873                 // Ignore lines which do not exist in both files.
 874                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 875                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 876                 }
 877
 878                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 879                         $line = $to_lines[$yi];
 880                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 881                                 continue;
 882                         $yhash[$this->_line_hash($line)] = 1;
 883                         $this->yv[] = $line;
 884                         $this->yind[] = $yi;
 885                 }
 886                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 887                         $line = $from_lines[$xi];
 888                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 889                                 continue;
 890                         $this->xv[] = $line;
 891                         $this->xind[] = $xi;
 892                 }
 893
 894                 // Find the LCS.
 895                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 896
 897                 // Merge edits when possible
 898                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 899                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 900
 901                 // Compute the edit operations.
 902                 $edits = array();
 903                 $xi = $yi = 0;
 904                 while ($xi < $n_from || $yi < $n_to) {
 905                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 906                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 907
 908                         // Skip matching "snake".
 909                         $copy = array();
 910                         while ( $xi < $n_from && $yi < $n_to
 911                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 912                                 $copy[] = $from_lines[$xi++];
 913                                 ++$yi;
 914                         }
 915                         if ($copy)
 916                                 $edits[] = new _DiffOp_Copy($copy);
 917
 918                         // Find deletes & adds.
 919                         $delete = array();
 920                         while ($xi < $n_from && $this->xchanged[$xi])
 921                                 $delete[] = $from_lines[$xi++];
 922
 923                         $add = array();
 924                         while ($yi < $n_to && $this->ychanged[$yi])
 925                                 $add[] = $to_lines[$yi++];
 926
 927                         if ($delete && $add)
 928                                 $edits[] = new _DiffOp_Change($delete, $add);
 929                         elseif ($delete)
 930                                 $edits[] = new _DiffOp_Delete($delete);
 931                         elseif ($add)
 932                                 $edits[] = new _DiffOp_Add($add);
 933                 }
 934                 wfProfileOut( $fname );
 935                 return $edits;
 936         }
 937
 938         /**
 939          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 940          */
 941         function _line_hash( $line ) {
 942                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 943                         return md5( $line );
 944                 } else {
 945                         return $line;
 946                 }
 947         }
 948
 949
 950         /* Divide the Largest Common Subsequence (LCS) of the sequences
 951          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 952          * sized segments.
 953          *
 954          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 955          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 956          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 957          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 958          * that (X0, Y0) == (XOFF, YOFF) and
 959          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 960          *
 961          * This function assumes that the first lines of the specified portions
 962          * of the two files do not match, and likewise that the last lines do not
 963          * match.  The caller must trim matching lines from the beginning and end
 964          * of the portions it is going to specify.
 965          */
 966         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 967                 $fname = '_DiffEngine::_diag';
 968                 wfProfileIn( $fname );
 969                 $flip = false;
 970
 971                 if ($xlim - $xoff > $ylim - $yoff) {
 972                         // Things seems faster (I'm not sure I understand why)
 973                                 // when the shortest sequence in X.
 974                                 $flip = true;
 975                         list ($xoff, $xlim, $yoff, $ylim)
 976                         = array( $yoff, $ylim, $xoff, $xlim);
 977                 }
 978
 979                 if ($flip)
 980                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 981                                 $ymatches[$this->xv[$i]][] = $i;
 982                 else
 983                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 984                                 $ymatches[$this->yv[$i]][] = $i;
 985
 986                 $this->lcs = 0;
 987                 $this->seq[0]= $yoff - 1;
 988                 $this->in_seq = array();
 989                 $ymids[0] = array();
 990
 991                 $numer = $xlim - $xoff + $nchunks - 1;
 992                 $x = $xoff;
 993                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 994                         wfProfileIn( "$fname-chunk" );
 995                         if ($chunk > 0)
 996                                 for ($i = 0; $i <= $this->lcs; $i++)
 997                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 998
 999                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1000                         for ( ; $x < $x1; $x++) {
1001                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1002                                         if (empty($ymatches[$line]))
1003                                                 continue;
1004                                 $matches = $ymatches[$line];
1005                                 reset($matches);
1006                                 while (list ($junk, $y) = each($matches))
1007                                         if (empty($this->in_seq[$y])) {
1008                                                 $k = $this->_lcs_pos($y);
1009                                                 USE_ASSERTS && assert($k > 0);
1010                                                 $ymids[$k] = $ymids[$k-1];
1011                                                 break;
1012                                         }
1013                                 while (list ( /* $junk */, $y) = each($matches)) {
1014                                         if ($y > $this->seq[$k-1]) {
1015                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1016                                                 // Optimization: this is a common case:
1017                                                 //      next match is just replacing previous match.
1018                                                 $this->in_seq[$this->seq[$k]] = false;
1019                                                 $this->seq[$k] = $y;
1020                                                 $this->in_seq[$y] = 1;
1021                                         } else if (empty($this->in_seq[$y])) {
1022                                                 $k = $this->_lcs_pos($y);
1023                                                 USE_ASSERTS && assert($k > 0);
1024                                                 $ymids[$k] = $ymids[$k-1];
1025                                         }
1026                                 }
1027                         }
1028                         wfProfileOut( "$fname-chunk" );
1029                 }
1030
1031                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1032                 $ymid = $ymids[$this->lcs];
1033                 for ($n = 0; $n < $nchunks - 1; $n++) {
1034                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1035                         $y1 = $ymid[$n] + 1;
1036                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1037                 }
1038                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1039
1040                 wfProfileOut( $fname );
1041                 return array($this->lcs, $seps);
1042         }
1043
1044         function _lcs_pos ($ypos) {
1045                 $fname = '_DiffEngine::_lcs_pos';
1046                 wfProfileIn( $fname );
1047
1048                 $end = $this->lcs;
1049                 if ($end == 0 || $ypos > $this->seq[$end]) {
1050                         $this->seq[++$this->lcs] = $ypos;
1051                         $this->in_seq[$ypos] = 1;
1052                         wfProfileOut( $fname );
1053                         return $this->lcs;
1054                 }
1055
1056                 $beg = 1;
1057                 while ($beg < $end) {
1058                         $mid = (int)(($beg + $end) / 2);
1059                         if ( $ypos > $this->seq[$mid] )
1060                                 $beg = $mid + 1;
1061                         else
1062                                 $end = $mid;
1063                 }
1064
1065                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1066
1067                 $this->in_seq[$this->seq[$end]] = false;
1068                 $this->seq[$end] = $ypos;
1069                 $this->in_seq[$ypos] = 1;
1070                 wfProfileOut( $fname );
1071                 return $end;
1072         }
1073
1074         /* Find LCS of two sequences.
1075          *
1076          * The results are recorded in the vectors $this->{x,y}changed[], by
1077          * storing a 1 in the element for each line that is an insertion
1078          * or deletion (ie. is not in the LCS).
1079          *
1080          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1081          *
1082          * Note that XLIM, YLIM are exclusive bounds.
1083          * All line numbers are origin-0 and discarded lines are not counted.
1084          */
1085         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1086                 $fname = '_DiffEngine::_compareseq';
1087                 wfProfileIn( $fname );
1088
1089                 // Slide down the bottom initial diagonal.
1090                 while ($xoff < $xlim && $yoff < $ylim
1091                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1092                         ++$xoff;
1093                         ++$yoff;
1094                 }
1095
1096                 // Slide up the top initial diagonal.
1097                 while ($xlim > $xoff && $ylim > $yoff
1098                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1099                         --$xlim;
1100                         --$ylim;
1101                 }
1102
1103                 if ($xoff == $xlim || $yoff == $ylim)
1104                         $lcs = 0;
1105                 else {
1106                         // This is ad hoc but seems to work well.
1107                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1108                         //$nchunks = max(2,min(8,(int)$nchunks));
1109                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1110                         list ($lcs, $seps)
1111                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1112                 }
1113
1114                 if ($lcs == 0) {
1115                         // X and Y sequences have no common subsequence:
1116                         // mark all changed.
1117                         while ($yoff < $ylim)
1118                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1119                         while ($xoff < $xlim)
1120                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1121                 } else {
1122                         // Use the partitions to split this problem into subproblems.
1123                         reset($seps);
1124                         $pt1 = $seps[0];
1125                         while ($pt2 = next($seps)) {
1126                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1127                                 $pt1 = $pt2;
1128                         }
1129                 }
1130                 wfProfileOut( $fname );
1131         }
1132
1133         /* Adjust inserts/deletes of identical lines to join changes
1134          * as much as possible.
1135          *
1136          * We do something when a run of changed lines include a
1137          * line at one end and has an excluded, identical line at the other.
1138          * We are free to choose which identical line is included.
1139          * `compareseq' usually chooses the one at the beginning,
1140          * but usually it is cleaner to consider the following identical line
1141          * to be the "change".
1142          *
1143          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1144          */
1145         function _shift_boundaries ($lines, &$changed, $other_changed) {
1146                 $fname = '_DiffEngine::_shift_boundaries';
1147                 wfProfileIn( $fname );
1148                 $i = 0;
1149                 $j = 0;
1150
1151                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1152                 $len = sizeof($lines);
1153                 $other_len = sizeof($other_changed);
1154
1155                 while (1) {
1156                         /*
1157                          * Scan forwards to find beginning of another run of changes.
1158                          * Also keep track of the corresponding point in the other file.
1159                          *
1160                          * Throughout this code, $i and $j are adjusted together so that
1161                          * the first $i elements of $changed and the first $j elements
1162                          * of $other_changed both contain the same number of zeros
1163                          * (unchanged lines).
1164                          * Furthermore, $j is always kept so that $j == $other_len or
1165                          * $other_changed[$j] == false.
1166                          */
1167                         while ($j < $other_len && $other_changed[$j])
1168                                 $j++;
1169
1170                         while ($i < $len && ! $changed[$i]) {
1171                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1172                                 $i++; $j++;
1173                                 while ($j < $other_len && $other_changed[$j])
1174                                         $j++;
1175                         }
1176
1177                         if ($i == $len)
1178                                 break;
1179
1180                         $start = $i;
1181
1182                         // Find the end of this run of changes.
1183                         while (++$i < $len && $changed[$i])
1184                                 continue;
1185
1186                         do {
1187                                 /*
1188                                  * Record the length of this run of changes, so that
1189                                  * we can later determine whether the run has grown.
1190                                  */
1191                                 $runlength = $i - $start;
1192
1193                                 /*
1194                                  * Move the changed region back, so long as the
1195                                  * previous unchanged line matches the last changed one.
1196                                  * This merges with previous changed regions.
1197                                  */
1198                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1199                                         $changed[--$start] = 1;
1200                                         $changed[--$i] = false;
1201                                         while ($start > 0 && $changed[$start - 1])
1202                                                 $start--;
1203                                         USE_ASSERTS && assert('$j > 0');
1204                                         while ($other_changed[--$j])
1205                                                 continue;
1206                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1207                                 }
1208
1209                                 /*
1210                                  * Set CORRESPONDING to the end of the changed run, at the last
1211                                  * point where it corresponds to a changed run in the other file.
1212                                  * CORRESPONDING == LEN means no such point has been found.
1213                                  */
1214                                 $corresponding = $j < $other_len ? $i : $len;
1215
1216                                 /*
1217                                  * Move the changed region forward, so long as the
1218                                  * first changed line matches the following unchanged one.
1219                                  * This merges with following changed regions.
1220                                  * Do this second, so that if there are no merges,
1221                                  * the changed region is moved forward as far as possible.
1222                                  */
1223                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1224                                         $changed[$start++] = false;
1225                                         $changed[$i++] = 1;
1226                                         while ($i < $len && $changed[$i])
1227                                                 $i++;
1228
1229                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1230                                         $j++;
1231                                         if ($j < $other_len && $other_changed[$j]) {
1232                                                 $corresponding = $i;
1233                                                 while ($j < $other_len && $other_changed[$j])
1234                                                         $j++;
1235                                         }
1236                                 }
1237                         } while ($runlength != $i - $start);
1238
1239                         /*
1240                          * If possible, move the fully-merged run of changes
1241                          * back to a corresponding run in the other file.
1242                          */
1243                         while ($corresponding < $i) {
1244                                 $changed[--$start] = 1;
1245                                 $changed[--$i] = 0;
1246                                 USE_ASSERTS && assert('$j > 0');
1247                                 while ($other_changed[--$j])
1248                                         continue;
1249                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1250                         }
1251                 }
1252                 wfProfileOut( $fname );
1253         }
1254 }
1255
1256 /**
1257  * Class representing a 'diff' between two sequences of strings.
1258  * @todo document
1259  * @private
1260  * @addtogroup DifferenceEngine
1261  */
1262 class Diff
1263 {
1264         var $edits;
1265
1266         /**
1267          * Constructor.
1268          * Computes diff between sequences of strings.
1269          *
1270          * @param $from_lines array An array of strings.
1271          *                (Typically these are lines from a file.)
1272          * @param $to_lines array An array of strings.
1273          */
1274         function Diff($from_lines, $to_lines) {
1275                 $eng = new _DiffEngine;
1276                 $this->edits = $eng->diff($from_lines, $to_lines);
1277                 //$this->_check($from_lines, $to_lines);
1278         }
1279
1280         /**
1281          * Compute reversed Diff.
1282          *
1283          * SYNOPSIS:
1284          *
1285          *      $diff = new Diff($lines1, $lines2);
1286          *      $rev = $diff->reverse();
1287          * @return object A Diff object representing the inverse of the
1288          *                                original diff.
1289          */
1290         function reverse () {
1291                 $rev = $this;
1292                 $rev->edits = array();
1293                 foreach ($this->edits as $edit) {
1294                         $rev->edits[] = $edit->reverse();
1295                 }
1296                 return $rev;
1297         }
1298
1299         /**
1300          * Check for empty diff.
1301          *
1302          * @return bool True iff two sequences were identical.
1303          */
1304         function isEmpty () {
1305                 foreach ($this->edits as $edit) {
1306                         if ($edit->type != 'copy')
1307                                 return false;
1308                 }
1309                 return true;
1310         }
1311
1312         /**
1313          * Compute the length of the Longest Common Subsequence (LCS).
1314          *
1315          * This is mostly for diagnostic purposed.
1316          *
1317          * @return int The length of the LCS.
1318          */
1319         function lcs () {
1320                 $lcs = 0;
1321                 foreach ($this->edits as $edit) {
1322                         if ($edit->type == 'copy')
1323                                 $lcs += sizeof($edit->orig);
1324                 }
1325                 return $lcs;
1326         }
1327
1328         /**
1329          * Get the original set of lines.
1330          *
1331          * This reconstructs the $from_lines parameter passed to the
1332          * constructor.
1333          *
1334          * @return array The original sequence of strings.
1335          */
1336         function orig() {
1337                 $lines = array();
1338
1339                 foreach ($this->edits as $edit) {
1340                         if ($edit->orig)
1341                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1342                 }
1343                 return $lines;
1344         }
1345
1346         /**
1347          * Get the closing set of lines.
1348          *
1349          * This reconstructs the $to_lines parameter passed to the
1350          * constructor.
1351          *
1352          * @return array The sequence of strings.
1353          */
1354         function closing() {
1355                 $lines = array();
1356
1357                 foreach ($this->edits as $edit) {
1358                         if ($edit->closing)
1359                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1360                 }
1361                 return $lines;
1362         }
1363
1364         /**
1365          * Check a Diff for validity.
1366          *
1367          * This is here only for debugging purposes.
1368          */
1369         function _check ($from_lines, $to_lines) {
1370                 $fname = 'Diff::_check';
1371                 wfProfileIn( $fname );
1372                 if (serialize($from_lines) != serialize($this->orig()))
1373                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1374                 if (serialize($to_lines) != serialize($this->closing()))
1375                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1376
1377                 $rev = $this->reverse();
1378                 if (serialize($to_lines) != serialize($rev->orig()))
1379                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1380                 if (serialize($from_lines) != serialize($rev->closing()))
1381                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1382
1383
1384                 $prevtype = 'none';
1385                 foreach ($this->edits as $edit) {
1386                         if ( $prevtype == $edit->type )
1387                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1388                         $prevtype = $edit->type;
1389                 }
1390
1391                 $lcs = $this->lcs();
1392                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1393                 wfProfileOut( $fname );
1394         }
1395 }
1396
1397 /**
1398  * @todo document, bad name.
1399  * @private
1400  * @addtogroup DifferenceEngine
1401  */
1402 class MappedDiff extends Diff
1403 {
1404         /**
1405          * Constructor.
1406          *
1407          * Computes diff between sequences of strings.
1408          *
1409          * This can be used to compute things like
1410          * case-insensitve diffs, or diffs which ignore
1411          * changes in white-space.
1412          *
1413          * @param $from_lines array An array of strings.
1414          *      (Typically these are lines from a file.)
1415          *
1416          * @param $to_lines array An array of strings.
1417          *
1418          * @param $mapped_from_lines array This array should
1419          *      have the same size number of elements as $from_lines.
1420          *      The elements in $mapped_from_lines and
1421          *      $mapped_to_lines are what is actually compared
1422          *      when computing the diff.
1423          *
1424          * @param $mapped_to_lines array This array should
1425          *      have the same number of elements as $to_lines.
1426          */
1427         function MappedDiff($from_lines, $to_lines,
1428                                                 $mapped_from_lines, $mapped_to_lines) {
1429                 $fname = 'MappedDiff::MappedDiff';
1430                 wfProfileIn( $fname );
1431
1432                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1433                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1434
1435                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1436
1437                 $xi = $yi = 0;
1438                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1439                         $orig = &$this->edits[$i]->orig;
1440                         if (is_array($orig)) {
1441                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1442                                 $xi += sizeof($orig);
1443                         }
1444
1445                         $closing = &$this->edits[$i]->closing;
1446                         if (is_array($closing)) {
1447                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1448                                 $yi += sizeof($closing);
1449                         }
1450                 }
1451                 wfProfileOut( $fname );
1452         }
1453 }
1454
1455 /**
1456  * A class to format Diffs
1457  *
1458  * This class formats the diff in classic diff format.
1459  * It is intended that this class be customized via inheritance,
1460  * to obtain fancier outputs.
1461  * @todo document
1462  * @private
1463  * @addtogroup DifferenceEngine
1464  */
1465 class DiffFormatter
1466 {
1467         /**
1468          * Number of leading context "lines" to preserve.
1469          *
1470          * This should be left at zero for this class, but subclasses
1471          * may want to set this to other values.
1472          */
1473         var $leading_context_lines = 0;
1474
1475         /**
1476          * Number of trailing context "lines" to preserve.
1477          *
1478          * This should be left at zero for this class, but subclasses
1479          * may want to set this to other values.
1480          */
1481         var $trailing_context_lines = 0;
1482
1483         /**
1484          * Format a diff.
1485          *
1486          * @param $diff object A Diff object.
1487          * @return string The formatted output.
1488          */
1489         function format($diff) {
1490                 $fname = 'DiffFormatter::format';
1491                 wfProfileIn( $fname );
1492
1493                 $xi = $yi = 1;
1494                 $block = false;
1495                 $context = array();
1496
1497                 $nlead = $this->leading_context_lines;
1498                 $ntrail = $this->trailing_context_lines;
1499
1500                 $this->_start_diff();
1501
1502                 foreach ($diff->edits as $edit) {
1503                         if ($edit->type == 'copy') {
1504                                 if (is_array($block)) {
1505                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1506                                                 $block[] = $edit;
1507                                         }
1508                                         else{
1509                                                 if ($ntrail) {
1510                                                         $context = array_slice($edit->orig, 0, $ntrail);
1511                                                         $block[] = new _DiffOp_Copy($context);
1512                                                 }
1513                                                 $this->_block($x0, $ntrail + $xi - $x0,
1514                                                                           $y0, $ntrail + $yi - $y0,
1515                                                                           $block);
1516                                                 $block = false;
1517                                         }
1518                                 }
1519                                 $context = $edit->orig;
1520                         }
1521                         else {
1522                                 if (! is_array($block)) {
1523                                         $context = array_slice($context, sizeof($context) - $nlead);
1524                                         $x0 = $xi - sizeof($context);
1525                                         $y0 = $yi - sizeof($context);
1526                                         $block = array();
1527                                         if ($context)
1528                                                 $block[] = new _DiffOp_Copy($context);
1529                                 }
1530                                 $block[] = $edit;
1531                         }
1532
1533                         if ($edit->orig)
1534                                 $xi += sizeof($edit->orig);
1535                         if ($edit->closing)
1536                                 $yi += sizeof($edit->closing);
1537                 }
1538
1539                 if (is_array($block))
1540                         $this->_block($x0, $xi - $x0,
1541                                                   $y0, $yi - $y0,
1542                                                   $block);
1543
1544                 $end = $this->_end_diff();
1545                 wfProfileOut( $fname );
1546                 return $end;
1547         }
1548
1549         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1550                 $fname = 'DiffFormatter::_block';
1551                 wfProfileIn( $fname );
1552                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1553                 foreach ($edits as $edit) {
1554                         if ($edit->type == 'copy')
1555                                 $this->_context($edit->orig);
1556                         elseif ($edit->type == 'add')
1557                                 $this->_added($edit->closing);
1558                         elseif ($edit->type == 'delete')
1559                                 $this->_deleted($edit->orig);
1560                         elseif ($edit->type == 'change')
1561                                 $this->_changed($edit->orig, $edit->closing);
1562                         else
1563                                 trigger_error('Unknown edit type', E_USER_ERROR);
1564                 }
1565                 $this->_end_block();
1566                 wfProfileOut( $fname );
1567         }
1568
1569         function _start_diff() {
1570                 ob_start();
1571         }
1572
1573         function _end_diff() {
1574                 $val = ob_get_contents();
1575                 ob_end_clean();
1576                 return $val;
1577         }
1578
1579         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1580                 if ($xlen > 1)
1581                         $xbeg .= "," . ($xbeg + $xlen - 1);
1582                 if ($ylen > 1)
1583                         $ybeg .= "," . ($ybeg + $ylen - 1);
1584
1585                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1586         }
1587
1588         function _start_block($header) {
1589                 echo $header;
1590         }
1591
1592         function _end_block() {
1593         }
1594
1595         function _lines($lines, $prefix = ' ') {
1596                 foreach ($lines as $line)
1597                         echo "$prefix $line\n";
1598         }
1599
1600         function _context($lines) {
1601                 $this->_lines($lines);
1602         }
1603
1604         function _added($lines) {
1605                 $this->_lines($lines, '>');
1606         }
1607         function _deleted($lines) {
1608                 $this->_lines($lines, '<');
1609         }
1610
1611         function _changed($orig, $closing) {
1612                 $this->_deleted($orig);
1613                 echo "---\n";
1614                 $this->_added($closing);
1615         }
1616 }
1617
1618
1619 /**
1620  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1621  *
1622  */
1623
1624 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1625
1626 /**
1627  * @todo document
1628  * @private
1629  * @addtogroup DifferenceEngine
1630  */
1631 class _HWLDF_WordAccumulator {
1632         function _HWLDF_WordAccumulator () {
1633                 $this->_lines = array();
1634                 $this->_line = '';
1635                 $this->_group = '';
1636                 $this->_tag = '';
1637         }
1638
1639         function _flushGroup ($new_tag) {
1640                 if ($this->_group !== '') {
1641                         if ($this->_tag == 'ins')
1642                                 $this->_line .= '<ins class="diffchange">' .
1643                                         htmlspecialchars ( $this->_group ) . '</ins>';
1644                         elseif ($this->_tag == 'del')
1645                                 $this->_line .= '<del class="diffchange">' .
1646                                         htmlspecialchars ( $this->_group ) . '</del>';
1647                         else
1648                                 $this->_line .= htmlspecialchars ( $this->_group );
1649                 }
1650                 $this->_group = '';
1651                 $this->_tag = $new_tag;
1652         }
1653
1654         function _flushLine ($new_tag) {
1655                 $this->_flushGroup($new_tag);
1656                 if ($this->_line != '')
1657                         array_push ( $this->_lines, $this->_line );
1658                 else
1659                         # make empty lines visible by inserting an NBSP
1660                         array_push ( $this->_lines, NBSP );
1661                 $this->_line = '';
1662         }
1663
1664         function addWords ($words, $tag = '') {
1665                 if ($tag != $this->_tag)
1666                         $this->_flushGroup($tag);
1667
1668                 foreach ($words as $word) {
1669                         // new-line should only come as first char of word.
1670                         if ($word == '')
1671                                 continue;
1672                         if ($word[0] == "\n") {
1673                                 $this->_flushLine($tag);
1674                                 $word = substr($word, 1);
1675                         }
1676                         assert(!strstr($word, "\n"));
1677                         $this->_group .= $word;
1678                 }
1679         }
1680
1681         function getLines() {
1682                 $this->_flushLine('~done');
1683                 return $this->_lines;
1684         }
1685 }
1686
1687 /**
1688  * @todo document
1689  * @private
1690  * @addtogroup DifferenceEngine
1691  */
1692 class WordLevelDiff extends MappedDiff
1693 {
1694         const MAX_LINE_LENGTH = 10000;
1695
1696         function WordLevelDiff ($orig_lines, $closing_lines) {
1697                 $fname = 'WordLevelDiff::WordLevelDiff';
1698                 wfProfileIn( $fname );
1699
1700                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1701                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1702
1703                 $this->MappedDiff($orig_words, $closing_words,
1704                                                   $orig_stripped, $closing_stripped);
1705                 wfProfileOut( $fname );
1706         }
1707
1708         function _split($lines) {
1709                 $fname = 'WordLevelDiff::_split';
1710                 wfProfileIn( $fname );
1711
1712                 $words = array();
1713                 $stripped = array();
1714                 $first = true;
1715                 foreach ( $lines as $line ) {
1716                         # If the line is too long, just pretend the entire line is one big word
1717                         # This prevents resource exhaustion problems
1718                         if ( $first ) {
1719                                 $first = false;
1720                         } else {
1721                                 $words[] = "\n";
1722                                 $stripped[] = "\n";
1723                         }
1724                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1725                                 $words[] = $line;
1726                                 $stripped[] = $line;
1727                         } else {
1728                                 $m = array();
1729                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1730                                         $line, $m))
1731                                 {
1732                                         $words = array_merge( $words, $m[0] );
1733                                         $stripped = array_merge( $stripped, $m[1] );
1734                                 }
1735                         }
1736                 }
1737                 wfProfileOut( $fname );
1738                 return array($words, $stripped);
1739         }
1740
1741         function orig () {
1742                 $fname = 'WordLevelDiff::orig';
1743                 wfProfileIn( $fname );
1744                 $orig = new _HWLDF_WordAccumulator;
1745
1746                 foreach ($this->edits as $edit) {
1747                         if ($edit->type == 'copy')
1748                                 $orig->addWords($edit->orig);
1749                         elseif ($edit->orig)
1750                                 $orig->addWords($edit->orig, 'del');
1751                 }
1752                 $lines = $orig->getLines();
1753                 wfProfileOut( $fname );
1754                 return $lines;
1755         }
1756
1757         function closing () {
1758                 $fname = 'WordLevelDiff::closing';
1759                 wfProfileIn( $fname );
1760                 $closing = new _HWLDF_WordAccumulator;
1761
1762                 foreach ($this->edits as $edit) {
1763                         if ($edit->type == 'copy')
1764                                 $closing->addWords($edit->closing);
1765                         elseif ($edit->closing)
1766                                 $closing->addWords($edit->closing, 'ins');
1767                 }
1768                 $lines = $closing->getLines();
1769                 wfProfileOut( $fname );
1770                 return $lines;
1771         }
1772 }
1773
1774 /**
1775  *      Wikipedia Table style diff formatter.
1776  * @todo document
1777  * @private
1778  * @addtogroup DifferenceEngine
1779  */
1780 class TableDiffFormatter extends DiffFormatter
1781 {
1782         function TableDiffFormatter() {
1783                 $this->leading_context_lines = 2;
1784                 $this->trailing_context_lines = 2;
1785         }
1786
1787         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1788                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1789                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1790                 return $r;
1791         }
1792
1793         function _start_block( $header ) {
1794                 echo $header;
1795         }
1796
1797         function _end_block() {
1798         }
1799
1800         function _lines( $lines, $prefix=' ', $color='white' ) {
1801         }
1802
1803         # HTML-escape parameter before calling this
1804         function addedLine( $line ) {
1805                 return $this->wrapLine( '+', 'diff-addedline', $line );
1806         }
1807
1808         # HTML-escape parameter before calling this
1809         function deletedLine( $line ) {
1810                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1811         }
1812
1813         # HTML-escape parameter before calling this
1814         function contextLine( $line ) {
1815                 return $this->wrapLine( ' ', 'diff-context', $line );
1816         }
1817
1818         private function wrapLine( $marker, $class, $line ) {
1819                 if( $line !== '' ) {
1820                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1821                         $line = "<div>$line</div>";
1822                 }
1823                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1824         }
1825
1826         function emptyLine() {
1827                 return '<td colspan="2">&nbsp;</td>';
1828         }
1829
1830         function _added( $lines ) {
1831                 foreach ($lines as $line) {
1832                         echo '<tr>' . $this->emptyLine() .
1833                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1834                 }
1835         }
1836
1837         function _deleted($lines) {
1838                 foreach ($lines as $line) {
1839                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1840                           $this->emptyLine() . "</tr>\n";
1841                 }
1842         }
1843
1844         function _context( $lines ) {
1845                 foreach ($lines as $line) {
1846                         echo '<tr>' .
1847                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1848                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1849                 }
1850         }
1851
1852         function _changed( $orig, $closing ) {
1853                 $fname = 'TableDiffFormatter::_changed';
1854                 wfProfileIn( $fname );
1855
1856                 $diff = new WordLevelDiff( $orig, $closing );
1857                 $del = $diff->orig();
1858                 $add = $diff->closing();
1859
1860                 # Notice that WordLevelDiff returns HTML-escaped output.
1861                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1862
1863                 while ( $line = array_shift( $del ) ) {
1864                         $aline = array_shift( $add );
1865                         echo '<tr>' . $this->deletedLine( $line ) .
1866                                 $this->addedLine( $aline ) . "</tr>\n";
1867                 }
1868                 foreach ($add as $line) {       # If any leftovers
1869                         echo '<tr>' . $this->emptyLine() .
1870                                 $this->addedLine( $line ) . "</tr>\n";
1871                 }
1872                 wfProfileOut( $fname );
1873         }
1874 }
1875
1876
1877