includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @todo indicate where diff.doc can be found.
   5  * @addtogroup DifferenceEngine
   6  */
   7
   8 /**
   9  * Constant to indicate diff cache compatibility.
  10  * Bump this when changing the diff formatting in a way that
  11  * fixes important bugs or such to force cached diff views to
  12  * clear.
  13  */
  14 define( 'MW_DIFF_VERSION', '1.11a' );
  15
  16 /**
  17  * @todo document
  18  * @public
  19  * @addtogroup DifferenceEngine
  20  */
  21 class DifferenceEngine {
  22         /**#@+
  23          * @private
  24          */
  25         var $mOldid, $mNewid, $mTitle;
  26         var $mOldtitle, $mNewtitle, $mPagetitle;
  27         var $mOldtext, $mNewtext;
  28         var $mOldPage, $mNewPage;
  29         var $mRcidMarkPatrolled;
  30         var $mOldRev, $mNewRev;
  31         var $mRevisionsLoaded = false; // Have the revisions been loaded
  32         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  33         /**#@-*/
  34
  35         /**
  36          * Constructor
  37          * @param $titleObj Title object that the diff is associated with
  38          * @param $old Integer: old ID we want to show and diff with.
  39          * @param $new String: either 'prev' or 'next'.
  40          * @param $rcid Integer: ??? FIXME (default 0)
  41          * @param $refreshCache boolean If set, refreshes the diff cache
  42          */
  43         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0, $refreshCache = false ) {
  44                 $this->mTitle = $titleObj;
  45                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  46
  47                 if ( 'prev' === $new ) {
  48                         # Show diff between revision $old and the previous one.
  49                         # Get previous one from DB.
  50                         #
  51                         $this->mNewid = intval($old);
  52
  53                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  54
  55                 } elseif ( 'next' === $new ) {
  56                         # Show diff between revision $old and the previous one.
  57                         # Get previous one from DB.
  58                         #
  59                         $this->mOldid = intval($old);
  60                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  61                         if ( false === $this->mNewid ) {
  62                                 # if no result, NewId points to the newest old revision. The only newer
  63                                 # revision is cur, which is "0".
  64                                 $this->mNewid = 0;
  65                         }
  66
  67                 } else {
  68                         $this->mOldid = intval($old);
  69                         $this->mNewid = intval($new);
  70                 }
  71                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  72                 $this->mRefreshCache = $refreshCache;
  73         }
  74
  75         function showDiffPage( $diffOnly = false ) {
  76                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  77                 $fname = 'DifferenceEngine::showDiffPage';
  78                 wfProfileIn( $fname );
  79
  80                 # If external diffs are enabled both globally and for the user,
  81                 # we'll use the application/x-external-editor interface to call
  82                 # an external diff tool like kompare, kdiff3, etc.
  83                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  84                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  85                         $wgOut->disable();
  86                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  87                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  88                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  89                         $special=$wgLang->getNsText(NS_SPECIAL);
  90                         $control=<<<CONTROL
  91 [Process]
  92 Type=Diff text
  93 Engine=MediaWiki
  94 Script={$wgServer}{$wgScript}
  95 Special namespace={$special}
  96
  97 [File]
  98 Extension=wiki
  99 URL=$url1
 100
 101 [File 2]
 102 Extension=wiki
 103 URL=$url2
 104 CONTROL;
 105                         echo($control);
 106                         return;
 107                 }
 108
 109                 $wgOut->setArticleFlag( false );
 110                 if ( ! $this->loadRevisionData() ) {
 111                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, {$this->mNewid})";
 112                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 113                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 114                         $wgOut->addWikitext( $mtext );
 115                         wfProfileOut( $fname );
 116                         return;
 117                 }
 118
 119                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 120
 121                 if ( $this->mNewRev->isCurrent() ) {
 122                         $wgOut->setArticleFlag( true );
 123                 }
 124
 125                 # mOldid is false if the difference engine is called with a "vague" query for
 126                 # a diff between a version V and its previous version V' AND the version V
 127                 # is the first version of that article. In that case, V' does not exist.
 128                 if ( $this->mOldid === false ) {
 129                         $this->showFirstRevision();
 130                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 131                         wfProfileOut( $fname );
 132                         return;
 133                 }
 134
 135                 $wgOut->suppressQuickbar();
 136
 137                 $oldTitle = $this->mOldPage->getPrefixedText();
 138                 $newTitle = $this->mNewPage->getPrefixedText();
 139                 if( $oldTitle == $newTitle ) {
 140                         $wgOut->setPageTitle( $newTitle );
 141                 } else {
 142                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 143                 }
 144                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 145                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 146
 147                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 148                         $wgOut->loginToUse();
 149                         $wgOut->output();
 150                         wfProfileOut( $fname );
 151                         exit;
 152                 }
 153
 154                 $sk = $wgUser->getSkin();
 155
 156                 if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 157                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 158                 } else {
 159                         $rollback = '';
 160                 }
 161
 162                 // Prepare a change patrol link, if applicable
 163                 if( $wgUseRCPatrol && $wgUser->isAllowed( 'patrol' ) ) {
 164                         // If we've been given an explicit change identifier, use it; saves time
 165                         if( $this->mRcidMarkPatrolled ) {
 166                                 $rcid = $this->mRcidMarkPatrolled;
 167                         } else {
 168                                 // Look for an unpatrolled change corresponding to this diff
 169                                 $db = wfGetDB( DB_SLAVE );
 170                                 $change = RecentChange::newFromConds(
 171                                         array(
 172                                                 // Add redundant timestamp condition so we can use the
 173                                                 // existing index
 174                                                 'rc_timestamp' => $db->timestamp( $this->mNewRev->getTimestamp() ),
 175                                                 'rc_this_oldid' => $this->mNewid,
 176                                                 'rc_last_oldid' => $this->mOldid,
 177                                                 'rc_patrolled' => 0,
 178                                         ),
 179                                         __METHOD__
 180                                 );
 181                                 if( $change instanceof RecentChange ) {
 182                                         $rcid = $change->mAttribs['rc_id'];
 183                                 } else {
 184                                         // None found
 185                                         $rcid = 0;
 186                                 }
 187                         }
 188                         // Build the link
 189                         if( $rcid ) {
 190                                 $patrol = ' [' . $sk->makeKnownLinkObj(
 191                                         $this->mTitle,
 192                                         wfMsgHtml( 'markaspatrolleddiff' ),
 193                                         "action=markpatrolled&rcid={$rcid}"
 194                                 ) . ']';
 195                         } else {
 196                                 $patrol = '';
 197                         }
 198                 } else {
 199                         $patrol = '';
 200                 }
 201
 202                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 203                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 204                 if ( $this->mNewRev->isCurrent() ) {
 205                         $nextlink = '&nbsp;';
 206                 } else {
 207                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 208                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 209                 }
 210
 211                 $oldminor = '';
 212                 $newminor = '';
 213
 214                 if ($this->mOldRev->mMinorEdit == 1) {
 215                         $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
 216                                 wfMsg( 'minoreditletter') ) . ' ';
 217                 }
 218
 219                 if ($this->mNewRev->mMinorEdit == 1) {
 220                         $newminor = wfElement( 'span', array( 'class' => 'minor' ),
 221                         wfMsg( 'minoreditletter') ) . ' ';
 222                 }
 223
 224                 $oldHeader = '<div id="mw-diff-otitle1"><strong>' . $this->mOldtitle . '</strong></div>' .
 225                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev ) . "</div>" .
 226                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly ) . "</div>" .
 227                         '<div id="mw-diff-otitle4">' . $prevlink . '</div>';
 228                 $newHeader = '<div id="mw-diff-ntitle1"><strong>' .$this->mNewtitle . '</strong></div>' .
 229                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev ) . " $rollback</div>" .
 230                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly ) . "</div>" .
 231                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 232
 233                 $this->showDiff( $oldHeader, $newHeader );
 234
 235                 if ( !$diffOnly )
 236                         $this->renderNewRevision();
 237
 238                 wfProfileOut( $fname );
 239         }
 240
 241         /**
 242          * Show the new revision of the page.
 243          */
 244         function renderNewRevision() {
 245                 global $wgOut;
 246                 $fname = 'DifferenceEngine::renderNewRevision';
 247                 wfProfileIn( $fname );
 248
 249                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 250                 #add deleted rev tag if needed
 251                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 252                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-permission' ) );
 253                 }
 254
 255                 if( !$this->mNewRev->isCurrent() ) {
 256                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 257                 }
 258
 259                 $this->loadNewText();
 260                 if( is_object( $this->mNewRev ) ) {
 261                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 262                 }
 263
 264                 if ($this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage()) {
 265                         // Stolen from Article::view --AG 2007-10-11
 266
 267                         // Give hooks a chance to customise the output
 268                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 269                                 // Wrap the whole lot in a <pre> and don't parse
 270                                 $m = array();
 271                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 272                                 $wgOut->addHtml( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 273                                 $wgOut->addHtml( htmlspecialchars( $this->mNewtext ) );
 274                                 $wgOut->addHtml( "\n</pre>\n" );
 275                         }
 276                 } else
 277                         $wgOut->addWikiTextTidy( $this->mNewtext );
 278
 279                 if( !$this->mNewRev->isCurrent() ) {
 280                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 281                 }
 282
 283                 wfProfileOut( $fname );
 284         }
 285
 286         /**
 287          * Show the first revision of an article. Uses normal diff headers in
 288          * contrast to normal "old revision" display style.
 289          */
 290         function showFirstRevision() {
 291                 global $wgOut, $wgUser;
 292
 293                 $fname = 'DifferenceEngine::showFirstRevision';
 294                 wfProfileIn( $fname );
 295
 296                 # Get article text from the DB
 297                 #
 298                 if ( ! $this->loadNewText() ) {
 299                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 300                           "{$this->mNewid})";
 301                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 302                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 303                         $wgOut->addWikitext( $mtext );
 304                         wfProfileOut( $fname );
 305                         return;
 306                 }
 307                 if ( $this->mNewRev->isCurrent() ) {
 308                         $wgOut->setArticleFlag( true );
 309                 }
 310
 311                 # Check if user is allowed to look at this page. If not, bail out.
 312                 #
 313                 if ( !( $this->mTitle->userCanRead() ) ) {
 314                         $wgOut->loginToUse();
 315                         $wgOut->output();
 316                         wfProfileOut( $fname );
 317                         exit;
 318                 }
 319
 320                 # Prepare the header box
 321                 #
 322                 $sk = $wgUser->getSkin();
 323
 324                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 325                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 326                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 327                         $sk->revComment( $this->mNewRev ) . "<br />" .
 328                         $nextlink . "</div>\n";
 329
 330                 $wgOut->addHTML( $header );
 331
 332                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 333                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 334
 335                 wfProfileOut( $fname );
 336         }
 337
 338         /**
 339          * Get the diff text, send it to $wgOut
 340          * Returns false if the diff could not be generated, otherwise returns true
 341          */
 342         function showDiff( $otitle, $ntitle ) {
 343                 global $wgOut;
 344                 $diff = $this->getDiff( $otitle, $ntitle );
 345                 if ( $diff === false ) {
 346                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
 347                         return false;
 348                 } else {
 349                         $this->showDiffStyle();
 350                         $wgOut->addHTML( $diff );
 351                         return true;
 352                 }
 353         }
 354
 355         /**
 356          * Add style sheets and supporting JS for diff display.
 357          */
 358         function showDiffStyle() {
 359                 global $wgStylePath, $wgStyleVersion, $wgOut;
 360                 $wgOut->addStyle( 'common/diff.css' );
 361
 362                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 363                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 364         }
 365
 366         /**
 367          * Get complete diff table, including header
 368          *
 369          * @param Title $otitle Old title
 370          * @param Title $ntitle New title
 371          * @return mixed
 372          */
 373         function getDiff( $otitle, $ntitle ) {
 374                 $body = $this->getDiffBody();
 375                 if ( $body === false ) {
 376                         return false;
 377                 } else {
 378                         $multi = $this->getMultiNotice();
 379                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 380                 }
 381         }
 382
 383         /**
 384          * Get the diff table body, without header
 385          *
 386          * @return mixed
 387          */
 388         function getDiffBody() {
 389                 global $wgMemc;
 390                 $fname = 'DifferenceEngine::getDiffBody';
 391                 wfProfileIn( $fname );
 392
 393                 // Cacheable?
 394                 $key = false;
 395                 if ( $this->mOldid && $this->mNewid ) {
 396                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 397                         // Try cache
 398                         if ( !$this->mRefreshCache ) {
 399                                 $difftext = $wgMemc->get( $key );
 400                                 if ( $difftext ) {
 401                                         wfIncrStats( 'diff_cache_hit' );
 402                                         $difftext = $this->localiseLineNumbers( $difftext );
 403                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 404                                         wfProfileOut( $fname );
 405                                         return $difftext;
 406                                 }
 407                         } // don't try to load but save the result
 408                 }
 409
 410                 #loadtext is permission safe, this just clears out the diff
 411                 if ( !$this->loadText() ) {
 412                         wfProfileOut( $fname );
 413                         return false;
 414                 } else if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 415                   return '';
 416                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 417                   return '';
 418                 }
 419
 420                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 421
 422                 // Save to cache for 7 days
 423                 if ( $key !== false && $difftext !== false ) {
 424                         wfIncrStats( 'diff_cache_miss' );
 425                         $wgMemc->set( $key, $difftext, 7*86400 );
 426                 } else {
 427                         wfIncrStats( 'diff_uncacheable' );
 428                 }
 429                 // Replace line numbers with the text in the user's language
 430                 if ( $difftext !== false ) {
 431                         $difftext = $this->localiseLineNumbers( $difftext );
 432                 }
 433                 wfProfileOut( $fname );
 434                 return $difftext;
 435         }
 436
 437         /**
 438          * Generate a diff, no caching
 439          * $otext and $ntext must be already segmented
 440          */
 441         function generateDiffBody( $otext, $ntext ) {
 442                 global $wgExternalDiffEngine, $wgContLang;
 443                 $fname = 'DifferenceEngine::generateDiffBody';
 444
 445                 $otext = str_replace( "\r\n", "\n", $otext );
 446                 $ntext = str_replace( "\r\n", "\n", $ntext );
 447
 448                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 449                         # For historical reasons, external diff engine expects
 450                         # input text to be HTML-escaped already
 451                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 452                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 453                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 454                                 dl('php_wikidiff.so');
 455                         }
 456                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 457                 }
 458
 459                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 460                         # Better external diff engine, the 2 may some day be dropped
 461                         # This one does the escaping and segmenting itself
 462                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 463                                 wfProfileIn( "$fname-dl" );
 464                                 @dl('php_wikidiff2.so');
 465                                 wfProfileOut( "$fname-dl" );
 466                         }
 467                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 468                                 wfProfileIn( 'wikidiff2_do_diff' );
 469                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 470                                 wfProfileOut( 'wikidiff2_do_diff' );
 471                                 return $text;
 472                         }
 473                 }
 474                 if ( $wgExternalDiffEngine !== false ) {
 475                         # Diff via the shell
 476                         global $wgTmpDirectory;
 477                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 478                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 479
 480                         $tempFile1 = fopen( $tempName1, "w" );
 481                         if ( !$tempFile1 ) {
 482                                 wfProfileOut( $fname );
 483                                 return false;
 484                         }
 485                         $tempFile2 = fopen( $tempName2, "w" );
 486                         if ( !$tempFile2 ) {
 487                                 wfProfileOut( $fname );
 488                                 return false;
 489                         }
 490                         fwrite( $tempFile1, $otext );
 491                         fwrite( $tempFile2, $ntext );
 492                         fclose( $tempFile1 );
 493                         fclose( $tempFile2 );
 494                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 495                         wfProfileIn( "$fname-shellexec" );
 496                         $difftext = wfShellExec( $cmd );
 497                         wfProfileOut( "$fname-shellexec" );
 498                         unlink( $tempName1 );
 499                         unlink( $tempName2 );
 500                         return $difftext;
 501                 }
 502
 503                 # Native PHP diff
 504                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 505                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 506                 $diffs = new Diff( $ota, $nta );
 507                 $formatter = new TableDiffFormatter();
 508                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 509         }
 510
 511
 512         /**
 513          * Replace line numbers with the text in the user's language
 514          */
 515         function localiseLineNumbers( $text ) {
 516                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 517                         array( &$this, 'localiseLineNumbersCb' ), $text );
 518         }
 519
 520         function localiseLineNumbersCb( $matches ) {
 521                 global $wgLang;
 522                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 523         }
 524
 525
 526         /**
 527          * If there are revisions between the ones being compared, return a note saying so.
 528          */
 529         function getMultiNotice() {
 530                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 531                         return '';
 532
 533                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 534                         // Comparing two different pages? Count would be meaningless.
 535                         return '';
 536                 }
 537
 538                 $oldid = $this->mOldRev->getId();
 539                 $newid = $this->mNewRev->getId();
 540                 if ( $oldid > $newid ) {
 541                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 542                 }
 543
 544                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 545                 if ( !$n )
 546                         return '';
 547
 548                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 549         }
 550
 551
 552         /**
 553          * Add the header to a diff body
 554          */
 555         function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 556                 global $wgOut;
 557
 558                 if ( $this->mOldRev && $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 559                    $otitle = '<span class="history-deleted">'.$otitle.'</span>';
 560                 }
 561                 if ( $this->mNewRev && $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 562                    $ntitle = '<span class="history-deleted">'.$ntitle.'</span>';
 563                 }
 564                 $header = "
 565                         <table class='diff'>
 566                         <col class='diff-marker' />
 567                         <col class='diff-content' />
 568                         <col class='diff-marker' />
 569                         <col class='diff-content' />
 570                         <tr>
 571                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 572                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 573                         </tr>
 574                 ";
 575
 576                 if ( $multi != '' )
 577                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 578
 579                 return $header . $diff . "</table>";
 580         }
 581
 582         /**
 583          * Use specified text instead of loading from the database
 584          */
 585         function setText( $oldText, $newText ) {
 586                 $this->mOldtext = $oldText;
 587                 $this->mNewtext = $newText;
 588                 $this->mTextLoaded = 2;
 589         }
 590
 591         /**
 592          * Load revision metadata for the specified articles. If newid is 0, then compare
 593          * the old article in oldid to the current article; if oldid is 0, then
 594          * compare the current article to the immediately previous one (ignoring the
 595          * value of newid).
 596          *
 597          * If oldid is false, leave the corresponding revision object set
 598          * to false. This is impossible via ordinary user input, and is provided for
 599          * API convenience.
 600          */
 601         function loadRevisionData() {
 602                 global $wgLang;
 603                 if ( $this->mRevisionsLoaded ) {
 604                         return true;
 605                 } else {
 606                         // Whether it succeeds or fails, we don't want to try again
 607                         $this->mRevisionsLoaded = true;
 608                 }
 609
 610                 // Load the new revision object
 611                 $this->mNewRev = $this->mNewid
 612                         ? Revision::newFromId( $this->mNewid )
 613                         : Revision::newFromTitle( $this->mTitle );
 614                 if( !$this->mNewRev instanceof Revision )
 615                         return false;
 616
 617                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 618                 $this->mNewid = $this->mNewRev->getId();
 619
 620                 // Set assorted variables
 621                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 622                 $this->mNewPage = $this->mNewRev->getTitle();
 623                 if( $this->mNewRev->isCurrent() ) {
 624                         $newLink = $this->mNewPage->escapeLocalUrl();
 625                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 626                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 627
 628                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
 629                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 630
 631                 } else {
 632                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 633                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 634                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 635
 636                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
 637                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 638                 }
 639
 640                 // Load the old revision object
 641                 $this->mOldRev = false;
 642                 if( $this->mOldid ) {
 643                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 644                 } elseif ( $this->mOldid === 0 ) {
 645                         $rev = $this->mNewRev->getPrevious();
 646                         if( $rev ) {
 647                                 $this->mOldid = $rev->getId();
 648                                 $this->mOldRev = $rev;
 649                         } else {
 650                                 // No previous revision; mark to show as first-version only.
 651                                 $this->mOldid = false;
 652                                 $this->mOldRev = false;
 653                         }
 654                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 655
 656                 if( is_null( $this->mOldRev ) ) {
 657                         return false;
 658                 }
 659
 660                 if ( $this->mOldRev ) {
 661                         $this->mOldPage = $this->mOldRev->getTitle();
 662
 663                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 664                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 665                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 666                         $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) )
 667                                 . "</a> (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 668
 669                         // Add an "undo" link
 670                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 671                         $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 672                 }
 673
 674                 return true;
 675         }
 676
 677         /**
 678          * Load the text of the revisions, as well as revision data.
 679          */
 680         function loadText() {
 681                 if ( $this->mTextLoaded == 2 ) {
 682                         return true;
 683                 } else {
 684                         // Whether it succeeds or fails, we don't want to try again
 685                         $this->mTextLoaded = 2;
 686                 }
 687
 688                 if ( !$this->loadRevisionData() ) {
 689                         return false;
 690                 }
 691                 if ( $this->mOldRev ) {
 692                         // FIXME: permission tests
 693                         $this->mOldtext = $this->mOldRev->revText();
 694                         if ( $this->mOldtext === false ) {
 695                                 return false;
 696                         }
 697                 }
 698                 if ( $this->mNewRev ) {
 699                         $this->mNewtext = $this->mNewRev->revText();
 700                         if ( $this->mNewtext === false ) {
 701                                 return false;
 702                         }
 703                 }
 704                 return true;
 705         }
 706
 707         /**
 708          * Load the text of the new revision, not the old one
 709          */
 710         function loadNewText() {
 711                 if ( $this->mTextLoaded >= 1 ) {
 712                         return true;
 713                 } else {
 714                         $this->mTextLoaded = 1;
 715                 }
 716                 if ( !$this->loadRevisionData() ) {
 717                         return false;
 718                 }
 719                 $this->mNewtext = $this->mNewRev->getText();
 720                 return true;
 721         }
 722
 723
 724 }
 725
 726 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 727 //
 728 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 729 // You may copy this code freely under the conditions of the GPL.
 730 //
 731
 732 define('USE_ASSERTS', function_exists('assert'));
 733
 734 /**
 735  * @todo document
 736  * @private
 737  * @addtogroup DifferenceEngine
 738  */
 739 class _DiffOp {
 740         var $type;
 741         var $orig;
 742         var $closing;
 743
 744         function reverse() {
 745                 trigger_error('pure virtual', E_USER_ERROR);
 746         }
 747
 748         function norig() {
 749                 return $this->orig ? sizeof($this->orig) : 0;
 750         }
 751
 752         function nclosing() {
 753                 return $this->closing ? sizeof($this->closing) : 0;
 754         }
 755 }
 756
 757 /**
 758  * @todo document
 759  * @private
 760  * @addtogroup DifferenceEngine
 761  */
 762 class _DiffOp_Copy extends _DiffOp {
 763         var $type = 'copy';
 764
 765         function _DiffOp_Copy ($orig, $closing = false) {
 766                 if (!is_array($closing))
 767                         $closing = $orig;
 768                 $this->orig = $orig;
 769                 $this->closing = $closing;
 770         }
 771
 772         function reverse() {
 773                 return new _DiffOp_Copy($this->closing, $this->orig);
 774         }
 775 }
 776
 777 /**
 778  * @todo document
 779  * @private
 780  * @addtogroup DifferenceEngine
 781  */
 782 class _DiffOp_Delete extends _DiffOp {
 783         var $type = 'delete';
 784
 785         function _DiffOp_Delete ($lines) {
 786                 $this->orig = $lines;
 787                 $this->closing = false;
 788         }
 789
 790         function reverse() {
 791                 return new _DiffOp_Add($this->orig);
 792         }
 793 }
 794
 795 /**
 796  * @todo document
 797  * @private
 798  * @addtogroup DifferenceEngine
 799  */
 800 class _DiffOp_Add extends _DiffOp {
 801         var $type = 'add';
 802
 803         function _DiffOp_Add ($lines) {
 804                 $this->closing = $lines;
 805                 $this->orig = false;
 806         }
 807
 808         function reverse() {
 809                 return new _DiffOp_Delete($this->closing);
 810         }
 811 }
 812
 813 /**
 814  * @todo document
 815  * @private
 816  * @addtogroup DifferenceEngine
 817  */
 818 class _DiffOp_Change extends _DiffOp {
 819         var $type = 'change';
 820
 821         function _DiffOp_Change ($orig, $closing) {
 822                 $this->orig = $orig;
 823                 $this->closing = $closing;
 824         }
 825
 826         function reverse() {
 827                 return new _DiffOp_Change($this->closing, $this->orig);
 828         }
 829 }
 830
 831
 832 /**
 833  * Class used internally by Diff to actually compute the diffs.
 834  *
 835  * The algorithm used here is mostly lifted from the perl module
 836  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 837  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 838  *
 839  * More ideas are taken from:
 840  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 841  *
 842  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 843  * diffutils-2.7, which can be found at:
 844  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 845  *
 846  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 847  * are my own.
 848  *
 849  * Line length limits for robustness added by Tim Starling, 2005-08-31
 850  *
 851  * @author Geoffrey T. Dairiki, Tim Starling
 852  * @private
 853  * @addtogroup DifferenceEngine
 854  */
 855 class _DiffEngine
 856 {
 857         const MAX_XREF_LENGTH =  10000;
 858
 859         function diff ($from_lines, $to_lines) {
 860                 $fname = '_DiffEngine::diff';
 861                 wfProfileIn( $fname );
 862
 863                 $n_from = sizeof($from_lines);
 864                 $n_to = sizeof($to_lines);
 865
 866                 $this->xchanged = $this->ychanged = array();
 867                 $this->xv = $this->yv = array();
 868                 $this->xind = $this->yind = array();
 869                 unset($this->seq);
 870                 unset($this->in_seq);
 871                 unset($this->lcs);
 872
 873                 // Skip leading common lines.
 874                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 875                         if ($from_lines[$skip] !== $to_lines[$skip])
 876                                 break;
 877                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 878                 }
 879                 // Skip trailing common lines.
 880                 $xi = $n_from; $yi = $n_to;
 881                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 882                         if ($from_lines[$xi] !== $to_lines[$yi])
 883                                 break;
 884                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 885                 }
 886
 887                 // Ignore lines which do not exist in both files.
 888                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 889                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 890                 }
 891
 892                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 893                         $line = $to_lines[$yi];
 894                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 895                                 continue;
 896                         $yhash[$this->_line_hash($line)] = 1;
 897                         $this->yv[] = $line;
 898                         $this->yind[] = $yi;
 899                 }
 900                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 901                         $line = $from_lines[$xi];
 902                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 903                                 continue;
 904                         $this->xv[] = $line;
 905                         $this->xind[] = $xi;
 906                 }
 907
 908                 // Find the LCS.
 909                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 910
 911                 // Merge edits when possible
 912                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 913                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 914
 915                 // Compute the edit operations.
 916                 $edits = array();
 917                 $xi = $yi = 0;
 918                 while ($xi < $n_from || $yi < $n_to) {
 919                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 920                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 921
 922                         // Skip matching "snake".
 923                         $copy = array();
 924                         while ( $xi < $n_from && $yi < $n_to
 925                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 926                                 $copy[] = $from_lines[$xi++];
 927                                 ++$yi;
 928                         }
 929                         if ($copy)
 930                                 $edits[] = new _DiffOp_Copy($copy);
 931
 932                         // Find deletes & adds.
 933                         $delete = array();
 934                         while ($xi < $n_from && $this->xchanged[$xi])
 935                                 $delete[] = $from_lines[$xi++];
 936
 937                         $add = array();
 938                         while ($yi < $n_to && $this->ychanged[$yi])
 939                                 $add[] = $to_lines[$yi++];
 940
 941                         if ($delete && $add)
 942                                 $edits[] = new _DiffOp_Change($delete, $add);
 943                         elseif ($delete)
 944                                 $edits[] = new _DiffOp_Delete($delete);
 945                         elseif ($add)
 946                                 $edits[] = new _DiffOp_Add($add);
 947                 }
 948                 wfProfileOut( $fname );
 949                 return $edits;
 950         }
 951
 952         /**
 953          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 954          */
 955         function _line_hash( $line ) {
 956                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 957                         return md5( $line );
 958                 } else {
 959                         return $line;
 960                 }
 961         }
 962
 963
 964         /* Divide the Largest Common Subsequence (LCS) of the sequences
 965          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 966          * sized segments.
 967          *
 968          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 969          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 970          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 971          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 972          * that (X0, Y0) == (XOFF, YOFF) and
 973          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 974          *
 975          * This function assumes that the first lines of the specified portions
 976          * of the two files do not match, and likewise that the last lines do not
 977          * match.  The caller must trim matching lines from the beginning and end
 978          * of the portions it is going to specify.
 979          */
 980         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 981                 $fname = '_DiffEngine::_diag';
 982                 wfProfileIn( $fname );
 983                 $flip = false;
 984
 985                 if ($xlim - $xoff > $ylim - $yoff) {
 986                         // Things seems faster (I'm not sure I understand why)
 987                                 // when the shortest sequence in X.
 988                                 $flip = true;
 989                         list ($xoff, $xlim, $yoff, $ylim)
 990                         = array( $yoff, $ylim, $xoff, $xlim);
 991                 }
 992
 993                 if ($flip)
 994                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 995                                 $ymatches[$this->xv[$i]][] = $i;
 996                 else
 997                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 998                                 $ymatches[$this->yv[$i]][] = $i;
 999
1000                 $this->lcs = 0;
1001                 $this->seq[0]= $yoff - 1;
1002                 $this->in_seq = array();
1003                 $ymids[0] = array();
1004
1005                 $numer = $xlim - $xoff + $nchunks - 1;
1006                 $x = $xoff;
1007                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1008                         wfProfileIn( "$fname-chunk" );
1009                         if ($chunk > 0)
1010                                 for ($i = 0; $i <= $this->lcs; $i++)
1011                                         $ymids[$i][$chunk-1] = $this->seq[$i];
1012
1013                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1014                         for ( ; $x < $x1; $x++) {
1015                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1016                                         if (empty($ymatches[$line]))
1017                                                 continue;
1018                                 $matches = $ymatches[$line];
1019                                 reset($matches);
1020                                 while (list ($junk, $y) = each($matches))
1021                                         if (empty($this->in_seq[$y])) {
1022                                                 $k = $this->_lcs_pos($y);
1023                                                 USE_ASSERTS && assert($k > 0);
1024                                                 $ymids[$k] = $ymids[$k-1];
1025                                                 break;
1026                                         }
1027                                 while (list ( /* $junk */, $y) = each($matches)) {
1028                                         if ($y > $this->seq[$k-1]) {
1029                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1030                                                 // Optimization: this is a common case:
1031                                                 //      next match is just replacing previous match.
1032                                                 $this->in_seq[$this->seq[$k]] = false;
1033                                                 $this->seq[$k] = $y;
1034                                                 $this->in_seq[$y] = 1;
1035                                         } else if (empty($this->in_seq[$y])) {
1036                                                 $k = $this->_lcs_pos($y);
1037                                                 USE_ASSERTS && assert($k > 0);
1038                                                 $ymids[$k] = $ymids[$k-1];
1039                                         }
1040                                 }
1041                         }
1042                         wfProfileOut( "$fname-chunk" );
1043                 }
1044
1045                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1046                 $ymid = $ymids[$this->lcs];
1047                 for ($n = 0; $n < $nchunks - 1; $n++) {
1048                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1049                         $y1 = $ymid[$n] + 1;
1050                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1051                 }
1052                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1053
1054                 wfProfileOut( $fname );
1055                 return array($this->lcs, $seps);
1056         }
1057
1058         function _lcs_pos ($ypos) {
1059                 $fname = '_DiffEngine::_lcs_pos';
1060                 wfProfileIn( $fname );
1061
1062                 $end = $this->lcs;
1063                 if ($end == 0 || $ypos > $this->seq[$end]) {
1064                         $this->seq[++$this->lcs] = $ypos;
1065                         $this->in_seq[$ypos] = 1;
1066                         wfProfileOut( $fname );
1067                         return $this->lcs;
1068                 }
1069
1070                 $beg = 1;
1071                 while ($beg < $end) {
1072                         $mid = (int)(($beg + $end) / 2);
1073                         if ( $ypos > $this->seq[$mid] )
1074                                 $beg = $mid + 1;
1075                         else
1076                                 $end = $mid;
1077                 }
1078
1079                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1080
1081                 $this->in_seq[$this->seq[$end]] = false;
1082                 $this->seq[$end] = $ypos;
1083                 $this->in_seq[$ypos] = 1;
1084                 wfProfileOut( $fname );
1085                 return $end;
1086         }
1087
1088         /* Find LCS of two sequences.
1089          *
1090          * The results are recorded in the vectors $this->{x,y}changed[], by
1091          * storing a 1 in the element for each line that is an insertion
1092          * or deletion (ie. is not in the LCS).
1093          *
1094          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1095          *
1096          * Note that XLIM, YLIM are exclusive bounds.
1097          * All line numbers are origin-0 and discarded lines are not counted.
1098          */
1099         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1100                 $fname = '_DiffEngine::_compareseq';
1101                 wfProfileIn( $fname );
1102
1103                 // Slide down the bottom initial diagonal.
1104                 while ($xoff < $xlim && $yoff < $ylim
1105                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1106                         ++$xoff;
1107                         ++$yoff;
1108                 }
1109
1110                 // Slide up the top initial diagonal.
1111                 while ($xlim > $xoff && $ylim > $yoff
1112                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1113                         --$xlim;
1114                         --$ylim;
1115                 }
1116
1117                 if ($xoff == $xlim || $yoff == $ylim)
1118                         $lcs = 0;
1119                 else {
1120                         // This is ad hoc but seems to work well.
1121                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1122                         //$nchunks = max(2,min(8,(int)$nchunks));
1123                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1124                         list ($lcs, $seps)
1125                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1126                 }
1127
1128                 if ($lcs == 0) {
1129                         // X and Y sequences have no common subsequence:
1130                         // mark all changed.
1131                         while ($yoff < $ylim)
1132                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1133                         while ($xoff < $xlim)
1134                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1135                 } else {
1136                         // Use the partitions to split this problem into subproblems.
1137                         reset($seps);
1138                         $pt1 = $seps[0];
1139                         while ($pt2 = next($seps)) {
1140                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1141                                 $pt1 = $pt2;
1142                         }
1143                 }
1144                 wfProfileOut( $fname );
1145         }
1146
1147         /* Adjust inserts/deletes of identical lines to join changes
1148          * as much as possible.
1149          *
1150          * We do something when a run of changed lines include a
1151          * line at one end and has an excluded, identical line at the other.
1152          * We are free to choose which identical line is included.
1153          * `compareseq' usually chooses the one at the beginning,
1154          * but usually it is cleaner to consider the following identical line
1155          * to be the "change".
1156          *
1157          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1158          */
1159         function _shift_boundaries ($lines, &$changed, $other_changed) {
1160                 $fname = '_DiffEngine::_shift_boundaries';
1161                 wfProfileIn( $fname );
1162                 $i = 0;
1163                 $j = 0;
1164
1165                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1166                 $len = sizeof($lines);
1167                 $other_len = sizeof($other_changed);
1168
1169                 while (1) {
1170                         /*
1171                          * Scan forwards to find beginning of another run of changes.
1172                          * Also keep track of the corresponding point in the other file.
1173                          *
1174                          * Throughout this code, $i and $j are adjusted together so that
1175                          * the first $i elements of $changed and the first $j elements
1176                          * of $other_changed both contain the same number of zeros
1177                          * (unchanged lines).
1178                          * Furthermore, $j is always kept so that $j == $other_len or
1179                          * $other_changed[$j] == false.
1180                          */
1181                         while ($j < $other_len && $other_changed[$j])
1182                                 $j++;
1183
1184                         while ($i < $len && ! $changed[$i]) {
1185                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1186                                 $i++; $j++;
1187                                 while ($j < $other_len && $other_changed[$j])
1188                                         $j++;
1189                         }
1190
1191                         if ($i == $len)
1192                                 break;
1193
1194                         $start = $i;
1195
1196                         // Find the end of this run of changes.
1197                         while (++$i < $len && $changed[$i])
1198                                 continue;
1199
1200                         do {
1201                                 /*
1202                                  * Record the length of this run of changes, so that
1203                                  * we can later determine whether the run has grown.
1204                                  */
1205                                 $runlength = $i - $start;
1206
1207                                 /*
1208                                  * Move the changed region back, so long as the
1209                                  * previous unchanged line matches the last changed one.
1210                                  * This merges with previous changed regions.
1211                                  */
1212                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1213                                         $changed[--$start] = 1;
1214                                         $changed[--$i] = false;
1215                                         while ($start > 0 && $changed[$start - 1])
1216                                                 $start--;
1217                                         USE_ASSERTS && assert('$j > 0');
1218                                         while ($other_changed[--$j])
1219                                                 continue;
1220                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1221                                 }
1222
1223                                 /*
1224                                  * Set CORRESPONDING to the end of the changed run, at the last
1225                                  * point where it corresponds to a changed run in the other file.
1226                                  * CORRESPONDING == LEN means no such point has been found.
1227                                  */
1228                                 $corresponding = $j < $other_len ? $i : $len;
1229
1230                                 /*
1231                                  * Move the changed region forward, so long as the
1232                                  * first changed line matches the following unchanged one.
1233                                  * This merges with following changed regions.
1234                                  * Do this second, so that if there are no merges,
1235                                  * the changed region is moved forward as far as possible.
1236                                  */
1237                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1238                                         $changed[$start++] = false;
1239                                         $changed[$i++] = 1;
1240                                         while ($i < $len && $changed[$i])
1241                                                 $i++;
1242
1243                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1244                                         $j++;
1245                                         if ($j < $other_len && $other_changed[$j]) {
1246                                                 $corresponding = $i;
1247                                                 while ($j < $other_len && $other_changed[$j])
1248                                                         $j++;
1249                                         }
1250                                 }
1251                         } while ($runlength != $i - $start);
1252
1253                         /*
1254                          * If possible, move the fully-merged run of changes
1255                          * back to a corresponding run in the other file.
1256                          */
1257                         while ($corresponding < $i) {
1258                                 $changed[--$start] = 1;
1259                                 $changed[--$i] = 0;
1260                                 USE_ASSERTS && assert('$j > 0');
1261                                 while ($other_changed[--$j])
1262                                         continue;
1263                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1264                         }
1265                 }
1266                 wfProfileOut( $fname );
1267         }
1268 }
1269
1270 /**
1271  * Class representing a 'diff' between two sequences of strings.
1272  * @todo document
1273  * @private
1274  * @addtogroup DifferenceEngine
1275  */
1276 class Diff
1277 {
1278         var $edits;
1279
1280         /**
1281          * Constructor.
1282          * Computes diff between sequences of strings.
1283          *
1284          * @param $from_lines array An array of strings.
1285          *                (Typically these are lines from a file.)
1286          * @param $to_lines array An array of strings.
1287          */
1288         function Diff($from_lines, $to_lines) {
1289                 $eng = new _DiffEngine;
1290                 $this->edits = $eng->diff($from_lines, $to_lines);
1291                 //$this->_check($from_lines, $to_lines);
1292         }
1293
1294         /**
1295          * Compute reversed Diff.
1296          *
1297          * SYNOPSIS:
1298          *
1299          *      $diff = new Diff($lines1, $lines2);
1300          *      $rev = $diff->reverse();
1301          * @return object A Diff object representing the inverse of the
1302          *                                original diff.
1303          */
1304         function reverse () {
1305                 $rev = $this;
1306                 $rev->edits = array();
1307                 foreach ($this->edits as $edit) {
1308                         $rev->edits[] = $edit->reverse();
1309                 }
1310                 return $rev;
1311         }
1312
1313         /**
1314          * Check for empty diff.
1315          *
1316          * @return bool True iff two sequences were identical.
1317          */
1318         function isEmpty () {
1319                 foreach ($this->edits as $edit) {
1320                         if ($edit->type != 'copy')
1321                                 return false;
1322                 }
1323                 return true;
1324         }
1325
1326         /**
1327          * Compute the length of the Longest Common Subsequence (LCS).
1328          *
1329          * This is mostly for diagnostic purposed.
1330          *
1331          * @return int The length of the LCS.
1332          */
1333         function lcs () {
1334                 $lcs = 0;
1335                 foreach ($this->edits as $edit) {
1336                         if ($edit->type == 'copy')
1337                                 $lcs += sizeof($edit->orig);
1338                 }
1339                 return $lcs;
1340         }
1341
1342         /**
1343          * Get the original set of lines.
1344          *
1345          * This reconstructs the $from_lines parameter passed to the
1346          * constructor.
1347          *
1348          * @return array The original sequence of strings.
1349          */
1350         function orig() {
1351                 $lines = array();
1352
1353                 foreach ($this->edits as $edit) {
1354                         if ($edit->orig)
1355                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1356                 }
1357                 return $lines;
1358         }
1359
1360         /**
1361          * Get the closing set of lines.
1362          *
1363          * This reconstructs the $to_lines parameter passed to the
1364          * constructor.
1365          *
1366          * @return array The sequence of strings.
1367          */
1368         function closing() {
1369                 $lines = array();
1370
1371                 foreach ($this->edits as $edit) {
1372                         if ($edit->closing)
1373                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1374                 }
1375                 return $lines;
1376         }
1377
1378         /**
1379          * Check a Diff for validity.
1380          *
1381          * This is here only for debugging purposes.
1382          */
1383         function _check ($from_lines, $to_lines) {
1384                 $fname = 'Diff::_check';
1385                 wfProfileIn( $fname );
1386                 if (serialize($from_lines) != serialize($this->orig()))
1387                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1388                 if (serialize($to_lines) != serialize($this->closing()))
1389                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1390
1391                 $rev = $this->reverse();
1392                 if (serialize($to_lines) != serialize($rev->orig()))
1393                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1394                 if (serialize($from_lines) != serialize($rev->closing()))
1395                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1396
1397
1398                 $prevtype = 'none';
1399                 foreach ($this->edits as $edit) {
1400                         if ( $prevtype == $edit->type )
1401                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1402                         $prevtype = $edit->type;
1403                 }
1404
1405                 $lcs = $this->lcs();
1406                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1407                 wfProfileOut( $fname );
1408         }
1409 }
1410
1411 /**
1412  * @todo document, bad name.
1413  * @private
1414  * @addtogroup DifferenceEngine
1415  */
1416 class MappedDiff extends Diff
1417 {
1418         /**
1419          * Constructor.
1420          *
1421          * Computes diff between sequences of strings.
1422          *
1423          * This can be used to compute things like
1424          * case-insensitve diffs, or diffs which ignore
1425          * changes in white-space.
1426          *
1427          * @param $from_lines array An array of strings.
1428          *      (Typically these are lines from a file.)
1429          *
1430          * @param $to_lines array An array of strings.
1431          *
1432          * @param $mapped_from_lines array This array should
1433          *      have the same size number of elements as $from_lines.
1434          *      The elements in $mapped_from_lines and
1435          *      $mapped_to_lines are what is actually compared
1436          *      when computing the diff.
1437          *
1438          * @param $mapped_to_lines array This array should
1439          *      have the same number of elements as $to_lines.
1440          */
1441         function MappedDiff($from_lines, $to_lines,
1442                                                 $mapped_from_lines, $mapped_to_lines) {
1443                 $fname = 'MappedDiff::MappedDiff';
1444                 wfProfileIn( $fname );
1445
1446                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1447                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1448
1449                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1450
1451                 $xi = $yi = 0;
1452                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1453                         $orig = &$this->edits[$i]->orig;
1454                         if (is_array($orig)) {
1455                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1456                                 $xi += sizeof($orig);
1457                         }
1458
1459                         $closing = &$this->edits[$i]->closing;
1460                         if (is_array($closing)) {
1461                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1462                                 $yi += sizeof($closing);
1463                         }
1464                 }
1465                 wfProfileOut( $fname );
1466         }
1467 }
1468
1469 /**
1470  * A class to format Diffs
1471  *
1472  * This class formats the diff in classic diff format.
1473  * It is intended that this class be customized via inheritance,
1474  * to obtain fancier outputs.
1475  * @todo document
1476  * @private
1477  * @addtogroup DifferenceEngine
1478  */
1479 class DiffFormatter
1480 {
1481         /**
1482          * Number of leading context "lines" to preserve.
1483          *
1484          * This should be left at zero for this class, but subclasses
1485          * may want to set this to other values.
1486          */
1487         var $leading_context_lines = 0;
1488
1489         /**
1490          * Number of trailing context "lines" to preserve.
1491          *
1492          * This should be left at zero for this class, but subclasses
1493          * may want to set this to other values.
1494          */
1495         var $trailing_context_lines = 0;
1496
1497         /**
1498          * Format a diff.
1499          *
1500          * @param $diff object A Diff object.
1501          * @return string The formatted output.
1502          */
1503         function format($diff) {
1504                 $fname = 'DiffFormatter::format';
1505                 wfProfileIn( $fname );
1506
1507                 $xi = $yi = 1;
1508                 $block = false;
1509                 $context = array();
1510
1511                 $nlead = $this->leading_context_lines;
1512                 $ntrail = $this->trailing_context_lines;
1513
1514                 $this->_start_diff();
1515
1516                 foreach ($diff->edits as $edit) {
1517                         if ($edit->type == 'copy') {
1518                                 if (is_array($block)) {
1519                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1520                                                 $block[] = $edit;
1521                                         }
1522                                         else{
1523                                                 if ($ntrail) {
1524                                                         $context = array_slice($edit->orig, 0, $ntrail);
1525                                                         $block[] = new _DiffOp_Copy($context);
1526                                                 }
1527                                                 $this->_block($x0, $ntrail + $xi - $x0,
1528                                                                           $y0, $ntrail + $yi - $y0,
1529                                                                           $block);
1530                                                 $block = false;
1531                                         }
1532                                 }
1533                                 $context = $edit->orig;
1534                         }
1535                         else {
1536                                 if (! is_array($block)) {
1537                                         $context = array_slice($context, sizeof($context) - $nlead);
1538                                         $x0 = $xi - sizeof($context);
1539                                         $y0 = $yi - sizeof($context);
1540                                         $block = array();
1541                                         if ($context)
1542                                                 $block[] = new _DiffOp_Copy($context);
1543                                 }
1544                                 $block[] = $edit;
1545                         }
1546
1547                         if ($edit->orig)
1548                                 $xi += sizeof($edit->orig);
1549                         if ($edit->closing)
1550                                 $yi += sizeof($edit->closing);
1551                 }
1552
1553                 if (is_array($block))
1554                         $this->_block($x0, $xi - $x0,
1555                                                   $y0, $yi - $y0,
1556                                                   $block);
1557
1558                 $end = $this->_end_diff();
1559                 wfProfileOut( $fname );
1560                 return $end;
1561         }
1562
1563         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1564                 $fname = 'DiffFormatter::_block';
1565                 wfProfileIn( $fname );
1566                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1567                 foreach ($edits as $edit) {
1568                         if ($edit->type == 'copy')
1569                                 $this->_context($edit->orig);
1570                         elseif ($edit->type == 'add')
1571                                 $this->_added($edit->closing);
1572                         elseif ($edit->type == 'delete')
1573                                 $this->_deleted($edit->orig);
1574                         elseif ($edit->type == 'change')
1575                                 $this->_changed($edit->orig, $edit->closing);
1576                         else
1577                                 trigger_error('Unknown edit type', E_USER_ERROR);
1578                 }
1579                 $this->_end_block();
1580                 wfProfileOut( $fname );
1581         }
1582
1583         function _start_diff() {
1584                 ob_start();
1585         }
1586
1587         function _end_diff() {
1588                 $val = ob_get_contents();
1589                 ob_end_clean();
1590                 return $val;
1591         }
1592
1593         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1594                 if ($xlen > 1)
1595                         $xbeg .= "," . ($xbeg + $xlen - 1);
1596                 if ($ylen > 1)
1597                         $ybeg .= "," . ($ybeg + $ylen - 1);
1598
1599                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1600         }
1601
1602         function _start_block($header) {
1603                 echo $header;
1604         }
1605
1606         function _end_block() {
1607         }
1608
1609         function _lines($lines, $prefix = ' ') {
1610                 foreach ($lines as $line)
1611                         echo "$prefix $line\n";
1612         }
1613
1614         function _context($lines) {
1615                 $this->_lines($lines);
1616         }
1617
1618         function _added($lines) {
1619                 $this->_lines($lines, '>');
1620         }
1621         function _deleted($lines) {
1622                 $this->_lines($lines, '<');
1623         }
1624
1625         function _changed($orig, $closing) {
1626                 $this->_deleted($orig);
1627                 echo "---\n";
1628                 $this->_added($closing);
1629         }
1630 }
1631
1632
1633 /**
1634  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1635  *
1636  */
1637
1638 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1639
1640 /**
1641  * @todo document
1642  * @private
1643  * @addtogroup DifferenceEngine
1644  */
1645 class _HWLDF_WordAccumulator {
1646         function _HWLDF_WordAccumulator () {
1647                 $this->_lines = array();
1648                 $this->_line = '';
1649                 $this->_group = '';
1650                 $this->_tag = '';
1651         }
1652
1653         function _flushGroup ($new_tag) {
1654                 if ($this->_group !== '') {
1655                         if ($this->_tag == 'ins')
1656                                 $this->_line .= '<ins class="diffchange">' .
1657                                         htmlspecialchars ( $this->_group ) . '</ins>';
1658                         elseif ($this->_tag == 'del')
1659                                 $this->_line .= '<del class="diffchange">' .
1660                                         htmlspecialchars ( $this->_group ) . '</del>';
1661                         else
1662                                 $this->_line .= htmlspecialchars ( $this->_group );
1663                 }
1664                 $this->_group = '';
1665                 $this->_tag = $new_tag;
1666         }
1667
1668         function _flushLine ($new_tag) {
1669                 $this->_flushGroup($new_tag);
1670                 if ($this->_line != '')
1671                         array_push ( $this->_lines, $this->_line );
1672                 else
1673                         # make empty lines visible by inserting an NBSP
1674                         array_push ( $this->_lines, NBSP );
1675                 $this->_line = '';
1676         }
1677
1678         function addWords ($words, $tag = '') {
1679                 if ($tag != $this->_tag)
1680                         $this->_flushGroup($tag);
1681
1682                 foreach ($words as $word) {
1683                         // new-line should only come as first char of word.
1684                         if ($word == '')
1685                                 continue;
1686                         if ($word[0] == "\n") {
1687                                 $this->_flushLine($tag);
1688                                 $word = substr($word, 1);
1689                         }
1690                         assert(!strstr($word, "\n"));
1691                         $this->_group .= $word;
1692                 }
1693         }
1694
1695         function getLines() {
1696                 $this->_flushLine('~done');
1697                 return $this->_lines;
1698         }
1699 }
1700
1701 /**
1702  * @todo document
1703  * @private
1704  * @addtogroup DifferenceEngine
1705  */
1706 class WordLevelDiff extends MappedDiff
1707 {
1708         const MAX_LINE_LENGTH = 10000;
1709
1710         function WordLevelDiff ($orig_lines, $closing_lines) {
1711                 $fname = 'WordLevelDiff::WordLevelDiff';
1712                 wfProfileIn( $fname );
1713
1714                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1715                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1716
1717                 $this->MappedDiff($orig_words, $closing_words,
1718                                                   $orig_stripped, $closing_stripped);
1719                 wfProfileOut( $fname );
1720         }
1721
1722         function _split($lines) {
1723                 $fname = 'WordLevelDiff::_split';
1724                 wfProfileIn( $fname );
1725
1726                 $words = array();
1727                 $stripped = array();
1728                 $first = true;
1729                 foreach ( $lines as $line ) {
1730                         # If the line is too long, just pretend the entire line is one big word
1731                         # This prevents resource exhaustion problems
1732                         if ( $first ) {
1733                                 $first = false;
1734                         } else {
1735                                 $words[] = "\n";
1736                                 $stripped[] = "\n";
1737                         }
1738                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1739                                 $words[] = $line;
1740                                 $stripped[] = $line;
1741                         } else {
1742                                 $m = array();
1743                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1744                                         $line, $m))
1745                                 {
1746                                         $words = array_merge( $words, $m[0] );
1747                                         $stripped = array_merge( $stripped, $m[1] );
1748                                 }
1749                         }
1750                 }
1751                 wfProfileOut( $fname );
1752                 return array($words, $stripped);
1753         }
1754
1755         function orig () {
1756                 $fname = 'WordLevelDiff::orig';
1757                 wfProfileIn( $fname );
1758                 $orig = new _HWLDF_WordAccumulator;
1759
1760                 foreach ($this->edits as $edit) {
1761                         if ($edit->type == 'copy')
1762                                 $orig->addWords($edit->orig);
1763                         elseif ($edit->orig)
1764                                 $orig->addWords($edit->orig, 'del');
1765                 }
1766                 $lines = $orig->getLines();
1767                 wfProfileOut( $fname );
1768                 return $lines;
1769         }
1770
1771         function closing () {
1772                 $fname = 'WordLevelDiff::closing';
1773                 wfProfileIn( $fname );
1774                 $closing = new _HWLDF_WordAccumulator;
1775
1776                 foreach ($this->edits as $edit) {
1777                         if ($edit->type == 'copy')
1778                                 $closing->addWords($edit->closing);
1779                         elseif ($edit->closing)
1780                                 $closing->addWords($edit->closing, 'ins');
1781                 }
1782                 $lines = $closing->getLines();
1783                 wfProfileOut( $fname );
1784                 return $lines;
1785         }
1786 }
1787
1788 /**
1789  *      Wikipedia Table style diff formatter.
1790  * @todo document
1791  * @private
1792  * @addtogroup DifferenceEngine
1793  */
1794 class TableDiffFormatter extends DiffFormatter
1795 {
1796         function TableDiffFormatter() {
1797                 $this->leading_context_lines = 2;
1798                 $this->trailing_context_lines = 2;
1799         }
1800
1801         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1802                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1803                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1804                 return $r;
1805         }
1806
1807         function _start_block( $header ) {
1808                 echo $header;
1809         }
1810
1811         function _end_block() {
1812         }
1813
1814         function _lines( $lines, $prefix=' ', $color='white' ) {
1815         }
1816
1817         # HTML-escape parameter before calling this
1818         function addedLine( $line ) {
1819                 return $this->wrapLine( '+', 'diff-addedline', $line );
1820         }
1821
1822         # HTML-escape parameter before calling this
1823         function deletedLine( $line ) {
1824                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1825         }
1826
1827         # HTML-escape parameter before calling this
1828         function contextLine( $line ) {
1829                 return $this->wrapLine( ' ', 'diff-context', $line );
1830         }
1831
1832         private function wrapLine( $marker, $class, $line ) {
1833                 if( $line !== '' ) {
1834                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1835                         $line = "<div>$line</div>";
1836                 }
1837                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1838         }
1839
1840         function emptyLine() {
1841                 return '<td colspan="2">&nbsp;</td>';
1842         }
1843
1844         function _added( $lines ) {
1845                 foreach ($lines as $line) {
1846                         echo '<tr>' . $this->emptyLine() .
1847                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1848                 }
1849         }
1850
1851         function _deleted($lines) {
1852                 foreach ($lines as $line) {
1853                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1854                           $this->emptyLine() . "</tr>\n";
1855                 }
1856         }
1857
1858         function _context( $lines ) {
1859                 foreach ($lines as $line) {
1860                         echo '<tr>' .
1861                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1862                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1863                 }
1864         }
1865
1866         function _changed( $orig, $closing ) {
1867                 $fname = 'TableDiffFormatter::_changed';
1868                 wfProfileIn( $fname );
1869
1870                 $diff = new WordLevelDiff( $orig, $closing );
1871                 $del = $diff->orig();
1872                 $add = $diff->closing();
1873
1874                 # Notice that WordLevelDiff returns HTML-escaped output.
1875                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1876
1877                 while ( $line = array_shift( $del ) ) {
1878                         $aline = array_shift( $add );
1879                         echo '<tr>' . $this->deletedLine( $line ) .
1880                                 $this->addedLine( $aline ) . "</tr>\n";
1881                 }
1882                 foreach ($add as $line) {       # If any leftovers
1883                         echo '<tr>' . $this->emptyLine() .
1884                                 $this->addedLine( $line ) . "</tr>\n";
1885                 }
1886                 wfProfileOut( $fname );
1887         }
1888 }
1889
1890
1891