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