includes/DifferenceEngine.php

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