includes/DifferenceEngine.php

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