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