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                         $multi = $this->getMultiNotice();
 296                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 297                 }
 298         }
 299
 300         /**
 301          * Get the diff table body, without header
 302          * Results are cached
 303          * Returns false on error
 304          */
 305         function getDiffBody() {
 306                 global $wgMemc;
 307                 $fname = 'DifferenceEngine::getDiffBody';
 308                 wfProfileIn( $fname );
 309
 310                 // Cacheable?
 311                 $key = false;
 312                 if ( $this->mOldid && $this->mNewid ) {
 313                         // Try cache
 314                         $key = wfMemcKey( 'diff', 'oldid', $this->mOldid, 'newid', $this->mNewid );
 315                         $difftext = $wgMemc->get( $key );
 316                         if ( $difftext ) {
 317                                 wfIncrStats( 'diff_cache_hit' );
 318                                 $difftext = $this->localiseLineNumbers( $difftext );
 319                                 $difftext .= "\n<!-- diff cache key $key -->\n";
 320                                 wfProfileOut( $fname );
 321                                 return $difftext;
 322                         }
 323                 }
 324
 325                 if ( !$this->loadText() ) {
 326                         wfProfileOut( $fname );
 327                         return false;
 328                 }
 329
 330                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 331
 332                 // Save to cache for 7 days
 333                 if ( $key !== false && $difftext !== false ) {
 334                         wfIncrStats( 'diff_cache_miss' );
 335                         $wgMemc->set( $key, $difftext, 7*86400 );
 336                 } else {
 337                         wfIncrStats( 'diff_uncacheable' );
 338                 }
 339                 // Replace line numbers with the text in the user's language
 340                 if ( $difftext !== false ) {
 341                         $difftext = $this->localiseLineNumbers( $difftext );
 342                 }
 343                 wfProfileOut( $fname );
 344                 return $difftext;
 345         }
 346
 347         /**
 348          * Generate a diff, no caching
 349          * $otext and $ntext must be already segmented
 350          */
 351         function generateDiffBody( $otext, $ntext ) {
 352                 global $wgExternalDiffEngine, $wgContLang;
 353                 $fname = 'DifferenceEngine::generateDiffBody';
 354
 355                 $otext = str_replace( "\r\n", "\n", $otext );
 356                 $ntext = str_replace( "\r\n", "\n", $ntext );
 357
 358                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 359                         # For historical reasons, external diff engine expects
 360                         # input text to be HTML-escaped already
 361                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 362                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 363                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 364                                 dl('php_wikidiff.so');
 365                         }
 366                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 367                 }
 368
 369                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 370                         # Better external diff engine, the 2 may some day be dropped
 371                         # This one does the escaping and segmenting itself
 372                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 373                                 wfProfileIn( "$fname-dl" );
 374                                 @dl('php_wikidiff2.so');
 375                                 wfProfileOut( "$fname-dl" );
 376                         }
 377                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 378                                 wfProfileIn( 'wikidiff2_do_diff' );
 379                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 380                                 wfProfileOut( 'wikidiff2_do_diff' );
 381                                 return $text;
 382                         }
 383                 }
 384                 if ( $wgExternalDiffEngine !== false ) {
 385                         # Diff via the shell
 386                         global $wgTmpDirectory;
 387                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 388                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 389
 390                         $tempFile1 = fopen( $tempName1, "w" );
 391                         if ( !$tempFile1 ) {
 392                                 wfProfileOut( $fname );
 393                                 return false;
 394                         }
 395                         $tempFile2 = fopen( $tempName2, "w" );
 396                         if ( !$tempFile2 ) {
 397                                 wfProfileOut( $fname );
 398                                 return false;
 399                         }
 400                         fwrite( $tempFile1, $otext );
 401                         fwrite( $tempFile2, $ntext );
 402                         fclose( $tempFile1 );
 403                         fclose( $tempFile2 );
 404                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 405                         wfProfileIn( "$fname-shellexec" );
 406                         $difftext = wfShellExec( $cmd );
 407                         wfProfileOut( "$fname-shellexec" );
 408                         unlink( $tempName1 );
 409                         unlink( $tempName2 );
 410                         return $difftext;
 411                 }
 412
 413                 # Native PHP diff
 414                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 415                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 416                 $diffs = new Diff( $ota, $nta );
 417                 $formatter = new TableDiffFormatter();
 418                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 419         }
 420
 421
 422         /**
 423          * Replace line numbers with the text in the user's language
 424          */
 425         function localiseLineNumbers( $text ) {
 426                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 427                         array( &$this, 'localiseLineNumbersCb' ), $text );
 428         }
 429
 430         function localiseLineNumbersCb( $matches ) {
 431                 global $wgLang;
 432                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 433         }
 434
 435
 436         /**
 437          * If there are revisions between the ones being compared, return a note saying so.
 438          */
 439         function getMultiNotice() {
 440                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 441                         return '';
 442
 443                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 444                         // Comparing two different pages? Count would be meaningless.
 445                         return '';
 446                 }
 447
 448                 $oldid = $this->mOldRev->getId();
 449                 $newid = $this->mNewRev->getId();
 450                 if ( $oldid > $newid ) {
 451                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 452                 }
 453
 454                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 455                 if ( !$n )
 456                         return '';
 457
 458                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 459         }
 460
 461
 462         /**
 463          * Add the header to a diff body
 464          */
 465         function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 466                 $header = "
 467                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 468                         <tr>
 469                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 470                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 471                         </tr>
 472                 ";
 473
 474                 if ( $multi != '' )
 475                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 476
 477                 return $header . $diff . "</table>";
 478         }
 479
 480         /**
 481          * Use specified text instead of loading from the database
 482          */
 483         function setText( $oldText, $newText ) {
 484                 $this->mOldtext = $oldText;
 485                 $this->mNewtext = $newText;
 486                 $this->mTextLoaded = 2;
 487         }
 488
 489         /**
 490          * Load revision metadata for the specified articles. If newid is 0, then compare
 491          * the old article in oldid to the current article; if oldid is 0, then
 492          * compare the current article to the immediately previous one (ignoring the
 493          * value of newid).
 494          *
 495          * If oldid is false, leave the corresponding revision object set
 496          * to false. This is impossible via ordinary user input, and is provided for
 497          * API convenience.
 498          */
 499         function loadRevisionData() {
 500                 global $wgLang;
 501                 if ( $this->mRevisionsLoaded ) {
 502                         return true;
 503                 } else {
 504                         // Whether it succeeds or fails, we don't want to try again
 505                         $this->mRevisionsLoaded = true;
 506                 }
 507
 508                 // Load the new revision object
 509                 if( $this->mNewid ) {
 510                         $this->mNewRev = Revision::newFromId( $this->mNewid );
 511                 } else {
 512                         $this->mNewRev = Revision::newFromTitle( $this->mTitle );
 513                 }
 514
 515                 if( is_null( $this->mNewRev ) ) {
 516                         return false;
 517                 }
 518
 519                 // Set assorted variables
 520                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 521                 $this->mNewPage = $this->mNewRev->getTitle();
 522                 if( $this->mNewRev->isCurrent() ) {
 523                         $newLink = $this->mNewPage->escapeLocalUrl();
 524                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 525                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 526                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undo=' . $this->mNewid );
 527
 528                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
 529                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)"
 530                                 . " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 531
 532                 } else {
 533                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 534                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 535                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undo=' . $this->mNewid );
 536                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 537
 538                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
 539                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)"
 540                                 . " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 541                 }
 542
 543                 // Load the old revision object
 544                 $this->mOldRev = false;
 545                 if( $this->mOldid ) {
 546                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 547                 } elseif ( $this->mOldid === 0 ) {
 548                         $rev = $this->mNewRev->getPrevious();
 549                         if( $rev ) {
 550                                 $this->mOldid = $rev->getId();
 551                                 $this->mOldRev = $rev;
 552                         } else {
 553                                 // No previous revision; mark to show as first-version only.
 554                                 $this->mOldid = false;
 555                                 $this->mOldRev = false;
 556                         }
 557                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 558
 559                 if( is_null( $this->mOldRev ) ) {
 560                         return false;
 561                 }
 562
 563                 if ( $this->mOldRev ) {
 564                         $this->mOldPage = $this->mOldRev->getTitle();
 565
 566                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 567                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 568                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 569                         $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) )
 570                                 . "</a> (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 571                 }
 572
 573                 return true;
 574         }
 575
 576         /**
 577          * Load the text of the revisions, as well as revision data.
 578          */
 579         function loadText() {
 580                 if ( $this->mTextLoaded == 2 ) {
 581                         return true;
 582                 } else {
 583                         // Whether it succeeds or fails, we don't want to try again
 584                         $this->mTextLoaded = 2;
 585                 }
 586
 587                 if ( !$this->loadRevisionData() ) {
 588                         return false;
 589                 }
 590                 if ( $this->mOldRev ) {
 591                         // FIXME: permission tests
 592                         $this->mOldtext = $this->mOldRev->getText();
 593                         if ( $this->mOldtext === false ) {
 594                                 return false;
 595                         }
 596                 }
 597                 if ( $this->mNewRev ) {
 598                         $this->mNewtext = $this->mNewRev->getText();
 599                         if ( $this->mNewtext === false ) {
 600                                 return false;
 601                         }
 602                 }
 603                 return true;
 604         }
 605
 606         /**
 607          * Load the text of the new revision, not the old one
 608          */
 609         function loadNewText() {
 610                 if ( $this->mTextLoaded >= 1 ) {
 611                         return true;
 612                 } else {
 613                         $this->mTextLoaded = 1;
 614                 }
 615                 if ( !$this->loadRevisionData() ) {
 616                         return false;
 617                 }
 618                 $this->mNewtext = $this->mNewRev->getText();
 619                 return true;
 620         }
 621
 622
 623 }
 624
 625 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 626 //
 627 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 628 // You may copy this code freely under the conditions of the GPL.
 629 //
 630
 631 define('USE_ASSERTS', function_exists('assert'));
 632
 633 /**
 634  * @todo document
 635  * @private
 636  * @package MediaWiki
 637  * @subpackage DifferenceEngine
 638  */
 639 class _DiffOp {
 640         var $type;
 641         var $orig;
 642         var $closing;
 643
 644         function reverse() {
 645                 trigger_error('pure virtual', E_USER_ERROR);
 646         }
 647
 648         function norig() {
 649                 return $this->orig ? sizeof($this->orig) : 0;
 650         }
 651
 652         function nclosing() {
 653                 return $this->closing ? sizeof($this->closing) : 0;
 654         }
 655 }
 656
 657 /**
 658  * @todo document
 659  * @private
 660  * @package MediaWiki
 661  * @subpackage DifferenceEngine
 662  */
 663 class _DiffOp_Copy extends _DiffOp {
 664         var $type = 'copy';
 665
 666         function _DiffOp_Copy ($orig, $closing = false) {
 667                 if (!is_array($closing))
 668                         $closing = $orig;
 669                 $this->orig = $orig;
 670                 $this->closing = $closing;
 671         }
 672
 673         function reverse() {
 674                 return new _DiffOp_Copy($this->closing, $this->orig);
 675         }
 676 }
 677
 678 /**
 679  * @todo document
 680  * @private
 681  * @package MediaWiki
 682  * @subpackage DifferenceEngine
 683  */
 684 class _DiffOp_Delete extends _DiffOp {
 685         var $type = 'delete';
 686
 687         function _DiffOp_Delete ($lines) {
 688                 $this->orig = $lines;
 689                 $this->closing = false;
 690         }
 691
 692         function reverse() {
 693                 return new _DiffOp_Add($this->orig);
 694         }
 695 }
 696
 697 /**
 698  * @todo document
 699  * @private
 700  * @package MediaWiki
 701  * @subpackage DifferenceEngine
 702  */
 703 class _DiffOp_Add extends _DiffOp {
 704         var $type = 'add';
 705
 706         function _DiffOp_Add ($lines) {
 707                 $this->closing = $lines;
 708                 $this->orig = false;
 709         }
 710
 711         function reverse() {
 712                 return new _DiffOp_Delete($this->closing);
 713         }
 714 }
 715
 716 /**
 717  * @todo document
 718  * @private
 719  * @package MediaWiki
 720  * @subpackage DifferenceEngine
 721  */
 722 class _DiffOp_Change extends _DiffOp {
 723         var $type = 'change';
 724
 725         function _DiffOp_Change ($orig, $closing) {
 726                 $this->orig = $orig;
 727                 $this->closing = $closing;
 728         }
 729
 730         function reverse() {
 731                 return new _DiffOp_Change($this->closing, $this->orig);
 732         }
 733 }
 734
 735
 736 /**
 737  * Class used internally by Diff to actually compute the diffs.
 738  *
 739  * The algorithm used here is mostly lifted from the perl module
 740  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 741  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 742  *
 743  * More ideas are taken from:
 744  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 745  *
 746  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 747  * diffutils-2.7, which can be found at:
 748  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 749  *
 750  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 751  * are my own.
 752  *
 753  * Line length limits for robustness added by Tim Starling, 2005-08-31
 754  *
 755  * @author Geoffrey T. Dairiki, Tim Starling
 756  * @private
 757  * @package MediaWiki
 758  * @subpackage DifferenceEngine
 759  */
 760 class _DiffEngine
 761 {
 762         const MAX_XREF_LENGTH =  10000;
 763
 764         function diff ($from_lines, $to_lines) {
 765                 $fname = '_DiffEngine::diff';
 766                 wfProfileIn( $fname );
 767
 768                 $n_from = sizeof($from_lines);
 769                 $n_to = sizeof($to_lines);
 770
 771                 $this->xchanged = $this->ychanged = array();
 772                 $this->xv = $this->yv = array();
 773                 $this->xind = $this->yind = array();
 774                 unset($this->seq);
 775                 unset($this->in_seq);
 776                 unset($this->lcs);
 777
 778                 // Skip leading common lines.
 779                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 780                         if ($from_lines[$skip] !== $to_lines[$skip])
 781                                 break;
 782                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 783                 }
 784                 // Skip trailing common lines.
 785                 $xi = $n_from; $yi = $n_to;
 786                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 787                         if ($from_lines[$xi] !== $to_lines[$yi])
 788                                 break;
 789                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 790                 }
 791
 792                 // Ignore lines which do not exist in both files.
 793                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 794                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 795                 }
 796
 797                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 798                         $line = $to_lines[$yi];
 799                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 800                                 continue;
 801                         $yhash[$this->_line_hash($line)] = 1;
 802                         $this->yv[] = $line;
 803                         $this->yind[] = $yi;
 804                 }
 805                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 806                         $line = $from_lines[$xi];
 807                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 808                                 continue;
 809                         $this->xv[] = $line;
 810                         $this->xind[] = $xi;
 811                 }
 812
 813                 // Find the LCS.
 814                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 815
 816                 // Merge edits when possible
 817                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 818                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 819
 820                 // Compute the edit operations.
 821                 $edits = array();
 822                 $xi = $yi = 0;
 823                 while ($xi < $n_from || $yi < $n_to) {
 824                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 825                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 826
 827                         // Skip matching "snake".
 828                         $copy = array();
 829                         while ( $xi < $n_from && $yi < $n_to
 830                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 831                                 $copy[] = $from_lines[$xi++];
 832                                 ++$yi;
 833                         }
 834                         if ($copy)
 835                                 $edits[] = new _DiffOp_Copy($copy);
 836
 837                         // Find deletes & adds.
 838                         $delete = array();
 839                         while ($xi < $n_from && $this->xchanged[$xi])
 840                                 $delete[] = $from_lines[$xi++];
 841
 842                         $add = array();
 843                         while ($yi < $n_to && $this->ychanged[$yi])
 844                                 $add[] = $to_lines[$yi++];
 845
 846                         if ($delete && $add)
 847                                 $edits[] = new _DiffOp_Change($delete, $add);
 848                         elseif ($delete)
 849                                 $edits[] = new _DiffOp_Delete($delete);
 850                         elseif ($add)
 851                                 $edits[] = new _DiffOp_Add($add);
 852                 }
 853                 wfProfileOut( $fname );
 854                 return $edits;
 855         }
 856
 857         /**
 858          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 859          */
 860         function _line_hash( $line ) {
 861                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 862                         return md5( $line );
 863                 } else {
 864                         return $line;
 865                 }
 866         }
 867
 868
 869         /* Divide the Largest Common Subsequence (LCS) of the sequences
 870          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 871          * sized segments.
 872          *
 873          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 874          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 875          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 876          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 877          * that (X0, Y0) == (XOFF, YOFF) and
 878          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 879          *
 880          * This function assumes that the first lines of the specified portions
 881          * of the two files do not match, and likewise that the last lines do not
 882          * match.  The caller must trim matching lines from the beginning and end
 883          * of the portions it is going to specify.
 884          */
 885         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 886                 $fname = '_DiffEngine::_diag';
 887                 wfProfileIn( $fname );
 888                 $flip = false;
 889
 890                 if ($xlim - $xoff > $ylim - $yoff) {
 891                         // Things seems faster (I'm not sure I understand why)
 892                                 // when the shortest sequence in X.
 893                                 $flip = true;
 894                         list ($xoff, $xlim, $yoff, $ylim)
 895                         = array( $yoff, $ylim, $xoff, $xlim);
 896                 }
 897
 898                 if ($flip)
 899                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 900                                 $ymatches[$this->xv[$i]][] = $i;
 901                 else
 902                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 903                                 $ymatches[$this->yv[$i]][] = $i;
 904
 905                 $this->lcs = 0;
 906                 $this->seq[0]= $yoff - 1;
 907                 $this->in_seq = array();
 908                 $ymids[0] = array();
 909
 910                 $numer = $xlim - $xoff + $nchunks - 1;
 911                 $x = $xoff;
 912                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 913                         wfProfileIn( "$fname-chunk" );
 914                         if ($chunk > 0)
 915                                 for ($i = 0; $i <= $this->lcs; $i++)
 916                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 917
 918                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 919                         for ( ; $x < $x1; $x++) {
 920                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 921                                         if (empty($ymatches[$line]))
 922                                                 continue;
 923                                 $matches = $ymatches[$line];
 924                                 reset($matches);
 925                                 while (list ($junk, $y) = each($matches))
 926                                         if (empty($this->in_seq[$y])) {
 927                                                 $k = $this->_lcs_pos($y);
 928                                                 USE_ASSERTS && assert($k > 0);
 929                                                 $ymids[$k] = $ymids[$k-1];
 930                                                 break;
 931                                         }
 932                                 while (list ( /* $junk */, $y) = each($matches)) {
 933                                         if ($y > $this->seq[$k-1]) {
 934                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 935                                                 // Optimization: this is a common case:
 936                                                 //      next match is just replacing previous match.
 937                                                 $this->in_seq[$this->seq[$k]] = false;
 938                                                 $this->seq[$k] = $y;
 939                                                 $this->in_seq[$y] = 1;
 940                                         } else if (empty($this->in_seq[$y])) {
 941                                                 $k = $this->_lcs_pos($y);
 942                                                 USE_ASSERTS && assert($k > 0);
 943                                                 $ymids[$k] = $ymids[$k-1];
 944                                         }
 945                                 }
 946                         }
 947                         wfProfileOut( "$fname-chunk" );
 948                 }
 949
 950                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 951                 $ymid = $ymids[$this->lcs];
 952                 for ($n = 0; $n < $nchunks - 1; $n++) {
 953                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 954                         $y1 = $ymid[$n] + 1;
 955                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 956                 }
 957                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 958
 959                 wfProfileOut( $fname );
 960                 return array($this->lcs, $seps);
 961         }
 962
 963         function _lcs_pos ($ypos) {
 964                 $fname = '_DiffEngine::_lcs_pos';
 965                 wfProfileIn( $fname );
 966
 967                 $end = $this->lcs;
 968                 if ($end == 0 || $ypos > $this->seq[$end]) {
 969                         $this->seq[++$this->lcs] = $ypos;
 970                         $this->in_seq[$ypos] = 1;
 971                         wfProfileOut( $fname );
 972                         return $this->lcs;
 973                 }
 974
 975                 $beg = 1;
 976                 while ($beg < $end) {
 977                         $mid = (int)(($beg + $end) / 2);
 978                         if ( $ypos > $this->seq[$mid] )
 979                                 $beg = $mid + 1;
 980                         else
 981                                 $end = $mid;
 982                 }
 983
 984                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 985
 986                 $this->in_seq[$this->seq[$end]] = false;
 987                 $this->seq[$end] = $ypos;
 988                 $this->in_seq[$ypos] = 1;
 989                 wfProfileOut( $fname );
 990                 return $end;
 991         }
 992
 993         /* Find LCS of two sequences.
 994          *
 995          * The results are recorded in the vectors $this->{x,y}changed[], by
 996          * storing a 1 in the element for each line that is an insertion
 997          * or deletion (ie. is not in the LCS).
 998          *
 999          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1000          *
1001          * Note that XLIM, YLIM are exclusive bounds.
1002          * All line numbers are origin-0 and discarded lines are not counted.
1003          */
1004         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1005                 $fname = '_DiffEngine::_compareseq';
1006                 wfProfileIn( $fname );
1007
1008                 // Slide down the bottom initial diagonal.
1009                 while ($xoff < $xlim && $yoff < $ylim
1010                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1011                         ++$xoff;
1012                         ++$yoff;
1013                 }
1014
1015                 // Slide up the top initial diagonal.
1016                 while ($xlim > $xoff && $ylim > $yoff
1017                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1018                         --$xlim;
1019                         --$ylim;
1020                 }
1021
1022                 if ($xoff == $xlim || $yoff == $ylim)
1023                         $lcs = 0;
1024                 else {
1025                         // This is ad hoc but seems to work well.
1026                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1027                         //$nchunks = max(2,min(8,(int)$nchunks));
1028                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1029                         list ($lcs, $seps)
1030                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1031                 }
1032
1033                 if ($lcs == 0) {
1034                         // X and Y sequences have no common subsequence:
1035                         // mark all changed.
1036                         while ($yoff < $ylim)
1037                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1038                         while ($xoff < $xlim)
1039                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1040                 } else {
1041                         // Use the partitions to split this problem into subproblems.
1042                         reset($seps);
1043                         $pt1 = $seps[0];
1044                         while ($pt2 = next($seps)) {
1045                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1046                                 $pt1 = $pt2;
1047                         }
1048                 }
1049                 wfProfileOut( $fname );
1050         }
1051
1052         /* Adjust inserts/deletes of identical lines to join changes
1053          * as much as possible.
1054          *
1055          * We do something when a run of changed lines include a
1056          * line at one end and has an excluded, identical line at the other.
1057          * We are free to choose which identical line is included.
1058          * `compareseq' usually chooses the one at the beginning,
1059          * but usually it is cleaner to consider the following identical line
1060          * to be the "change".
1061          *
1062          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1063          */
1064         function _shift_boundaries ($lines, &$changed, $other_changed) {
1065                 $fname = '_DiffEngine::_shift_boundaries';
1066                 wfProfileIn( $fname );
1067                 $i = 0;
1068                 $j = 0;
1069
1070                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1071                 $len = sizeof($lines);
1072                 $other_len = sizeof($other_changed);
1073
1074                 while (1) {
1075                         /*
1076                          * Scan forwards to find beginning of another run of changes.
1077                          * Also keep track of the corresponding point in the other file.
1078                          *
1079                          * Throughout this code, $i and $j are adjusted together so that
1080                          * the first $i elements of $changed and the first $j elements
1081                          * of $other_changed both contain the same number of zeros
1082                          * (unchanged lines).
1083                          * Furthermore, $j is always kept so that $j == $other_len or
1084                          * $other_changed[$j] == false.
1085                          */
1086                         while ($j < $other_len && $other_changed[$j])
1087                                 $j++;
1088
1089                         while ($i < $len && ! $changed[$i]) {
1090                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1091                                 $i++; $j++;
1092                                 while ($j < $other_len && $other_changed[$j])
1093                                         $j++;
1094                         }
1095
1096                         if ($i == $len)
1097                                 break;
1098
1099                         $start = $i;
1100
1101                         // Find the end of this run of changes.
1102                         while (++$i < $len && $changed[$i])
1103                                 continue;
1104
1105                         do {
1106                                 /*
1107                                  * Record the length of this run of changes, so that
1108                                  * we can later determine whether the run has grown.
1109                                  */
1110                                 $runlength = $i - $start;
1111
1112                                 /*
1113                                  * Move the changed region back, so long as the
1114                                  * previous unchanged line matches the last changed one.
1115                                  * This merges with previous changed regions.
1116                                  */
1117                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1118                                         $changed[--$start] = 1;
1119                                         $changed[--$i] = false;
1120                                         while ($start > 0 && $changed[$start - 1])
1121                                                 $start--;
1122                                         USE_ASSERTS && assert('$j > 0');
1123                                         while ($other_changed[--$j])
1124                                                 continue;
1125                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1126                                 }
1127
1128                                 /*
1129                                  * Set CORRESPONDING to the end of the changed run, at the last
1130                                  * point where it corresponds to a changed run in the other file.
1131                                  * CORRESPONDING == LEN means no such point has been found.
1132                                  */
1133                                 $corresponding = $j < $other_len ? $i : $len;
1134
1135                                 /*
1136                                  * Move the changed region forward, so long as the
1137                                  * first changed line matches the following unchanged one.
1138                                  * This merges with following changed regions.
1139                                  * Do this second, so that if there are no merges,
1140                                  * the changed region is moved forward as far as possible.
1141                                  */
1142                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1143                                         $changed[$start++] = false;
1144                                         $changed[$i++] = 1;
1145                                         while ($i < $len && $changed[$i])
1146                                                 $i++;
1147
1148                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1149                                         $j++;
1150                                         if ($j < $other_len && $other_changed[$j]) {
1151                                                 $corresponding = $i;
1152                                                 while ($j < $other_len && $other_changed[$j])
1153                                                         $j++;
1154                                         }
1155                                 }
1156                         } while ($runlength != $i - $start);
1157
1158                         /*
1159                          * If possible, move the fully-merged run of changes
1160                          * back to a corresponding run in the other file.
1161                          */
1162                         while ($corresponding < $i) {
1163                                 $changed[--$start] = 1;
1164                                 $changed[--$i] = 0;
1165                                 USE_ASSERTS && assert('$j > 0');
1166                                 while ($other_changed[--$j])
1167                                         continue;
1168                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1169                         }
1170                 }
1171                 wfProfileOut( $fname );
1172         }
1173 }
1174
1175 /**
1176  * Class representing a 'diff' between two sequences of strings.
1177  * @todo document
1178  * @private
1179  * @package MediaWiki
1180  * @subpackage DifferenceEngine
1181  */
1182 class Diff
1183 {
1184         var $edits;
1185
1186         /**
1187          * Constructor.
1188          * Computes diff between sequences of strings.
1189          *
1190          * @param $from_lines array An array of strings.
1191          *                (Typically these are lines from a file.)
1192          * @param $to_lines array An array of strings.
1193          */
1194         function Diff($from_lines, $to_lines) {
1195                 $eng = new _DiffEngine;
1196                 $this->edits = $eng->diff($from_lines, $to_lines);
1197                 //$this->_check($from_lines, $to_lines);
1198         }
1199
1200         /**
1201          * Compute reversed Diff.
1202          *
1203          * SYNOPSIS:
1204          *
1205          *      $diff = new Diff($lines1, $lines2);
1206          *      $rev = $diff->reverse();
1207          * @return object A Diff object representing the inverse of the
1208          *                                original diff.
1209          */
1210         function reverse () {
1211                 $rev = $this;
1212                 $rev->edits = array();
1213                 foreach ($this->edits as $edit) {
1214                         $rev->edits[] = $edit->reverse();
1215                 }
1216                 return $rev;
1217         }
1218
1219         /**
1220          * Check for empty diff.
1221          *
1222          * @return bool True iff two sequences were identical.
1223          */
1224         function isEmpty () {
1225                 foreach ($this->edits as $edit) {
1226                         if ($edit->type != 'copy')
1227                                 return false;
1228                 }
1229                 return true;
1230         }
1231
1232         /**
1233          * Compute the length of the Longest Common Subsequence (LCS).
1234          *
1235          * This is mostly for diagnostic purposed.
1236          *
1237          * @return int The length of the LCS.
1238          */
1239         function lcs () {
1240                 $lcs = 0;
1241                 foreach ($this->edits as $edit) {
1242                         if ($edit->type == 'copy')
1243                                 $lcs += sizeof($edit->orig);
1244                 }
1245                 return $lcs;
1246         }
1247
1248         /**
1249          * Get the original set of lines.
1250          *
1251          * This reconstructs the $from_lines parameter passed to the
1252          * constructor.
1253          *
1254          * @return array The original sequence of strings.
1255          */
1256         function orig() {
1257                 $lines = array();
1258
1259                 foreach ($this->edits as $edit) {
1260                         if ($edit->orig)
1261                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1262                 }
1263                 return $lines;
1264         }
1265
1266         /**
1267          * Get the closing set of lines.
1268          *
1269          * This reconstructs the $to_lines parameter passed to the
1270          * constructor.
1271          *
1272          * @return array The sequence of strings.
1273          */
1274         function closing() {
1275                 $lines = array();
1276
1277                 foreach ($this->edits as $edit) {
1278                         if ($edit->closing)
1279                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1280                 }
1281                 return $lines;
1282         }
1283
1284         /**
1285          * Check a Diff for validity.
1286          *
1287          * This is here only for debugging purposes.
1288          */
1289         function _check ($from_lines, $to_lines) {
1290                 $fname = 'Diff::_check';
1291                 wfProfileIn( $fname );
1292                 if (serialize($from_lines) != serialize($this->orig()))
1293                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1294                 if (serialize($to_lines) != serialize($this->closing()))
1295                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1296
1297                 $rev = $this->reverse();
1298                 if (serialize($to_lines) != serialize($rev->orig()))
1299                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1300                 if (serialize($from_lines) != serialize($rev->closing()))
1301                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1302
1303
1304                 $prevtype = 'none';
1305                 foreach ($this->edits as $edit) {
1306                         if ( $prevtype == $edit->type )
1307                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1308                         $prevtype = $edit->type;
1309                 }
1310
1311                 $lcs = $this->lcs();
1312                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1313                 wfProfileOut( $fname );
1314         }
1315 }
1316
1317 /**
1318  * FIXME: bad name.
1319  * @todo document
1320  * @private
1321  * @package MediaWiki
1322  * @subpackage DifferenceEngine
1323  */
1324 class MappedDiff extends Diff
1325 {
1326         /**
1327          * Constructor.
1328          *
1329          * Computes diff between sequences of strings.
1330          *
1331          * This can be used to compute things like
1332          * case-insensitve diffs, or diffs which ignore
1333          * changes in white-space.
1334          *
1335          * @param $from_lines array An array of strings.
1336          *      (Typically these are lines from a file.)
1337          *
1338          * @param $to_lines array An array of strings.
1339          *
1340          * @param $mapped_from_lines array This array should
1341          *      have the same size number of elements as $from_lines.
1342          *      The elements in $mapped_from_lines and
1343          *      $mapped_to_lines are what is actually compared
1344          *      when computing the diff.
1345          *
1346          * @param $mapped_to_lines array This array should
1347          *      have the same number of elements as $to_lines.
1348          */
1349         function MappedDiff($from_lines, $to_lines,
1350                                                 $mapped_from_lines, $mapped_to_lines) {
1351                 $fname = 'MappedDiff::MappedDiff';
1352                 wfProfileIn( $fname );
1353
1354                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1355                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1356
1357                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1358
1359                 $xi = $yi = 0;
1360                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1361                         $orig = &$this->edits[$i]->orig;
1362                         if (is_array($orig)) {
1363                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1364                                 $xi += sizeof($orig);
1365                         }
1366
1367                         $closing = &$this->edits[$i]->closing;
1368                         if (is_array($closing)) {
1369                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1370                                 $yi += sizeof($closing);
1371                         }
1372                 }
1373                 wfProfileOut( $fname );
1374         }
1375 }
1376
1377 /**
1378  * A class to format Diffs
1379  *
1380  * This class formats the diff in classic diff format.
1381  * It is intended that this class be customized via inheritance,
1382  * to obtain fancier outputs.
1383  * @todo document
1384  * @private
1385  * @package MediaWiki
1386  * @subpackage DifferenceEngine
1387  */
1388 class DiffFormatter
1389 {
1390         /**
1391          * Number of leading context "lines" to preserve.
1392          *
1393          * This should be left at zero for this class, but subclasses
1394          * may want to set this to other values.
1395          */
1396         var $leading_context_lines = 0;
1397
1398         /**
1399          * Number of trailing context "lines" to preserve.
1400          *
1401          * This should be left at zero for this class, but subclasses
1402          * may want to set this to other values.
1403          */
1404         var $trailing_context_lines = 0;
1405
1406         /**
1407          * Format a diff.
1408          *
1409          * @param $diff object A Diff object.
1410          * @return string The formatted output.
1411          */
1412         function format($diff) {
1413                 $fname = 'DiffFormatter::format';
1414                 wfProfileIn( $fname );
1415
1416                 $xi = $yi = 1;
1417                 $block = false;
1418                 $context = array();
1419
1420                 $nlead = $this->leading_context_lines;
1421                 $ntrail = $this->trailing_context_lines;
1422
1423                 $this->_start_diff();
1424
1425                 foreach ($diff->edits as $edit) {
1426                         if ($edit->type == 'copy') {
1427                                 if (is_array($block)) {
1428                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1429                                                 $block[] = $edit;
1430                                         }
1431                                         else{
1432                                                 if ($ntrail) {
1433                                                         $context = array_slice($edit->orig, 0, $ntrail);
1434                                                         $block[] = new _DiffOp_Copy($context);
1435                                                 }
1436                                                 $this->_block($x0, $ntrail + $xi - $x0,
1437                                                                           $y0, $ntrail + $yi - $y0,
1438                                                                           $block);
1439                                                 $block = false;
1440                                         }
1441                                 }
1442                                 $context = $edit->orig;
1443                         }
1444                         else {
1445                                 if (! is_array($block)) {
1446                                         $context = array_slice($context, sizeof($context) - $nlead);
1447                                         $x0 = $xi - sizeof($context);
1448                                         $y0 = $yi - sizeof($context);
1449                                         $block = array();
1450                                         if ($context)
1451                                                 $block[] = new _DiffOp_Copy($context);
1452                                 }
1453                                 $block[] = $edit;
1454                         }
1455
1456                         if ($edit->orig)
1457                                 $xi += sizeof($edit->orig);
1458                         if ($edit->closing)
1459                                 $yi += sizeof($edit->closing);
1460                 }
1461
1462                 if (is_array($block))
1463                         $this->_block($x0, $xi - $x0,
1464                                                   $y0, $yi - $y0,
1465                                                   $block);
1466
1467                 $end = $this->_end_diff();
1468                 wfProfileOut( $fname );
1469                 return $end;
1470         }
1471
1472         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1473                 $fname = 'DiffFormatter::_block';
1474                 wfProfileIn( $fname );
1475                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1476                 foreach ($edits as $edit) {
1477                         if ($edit->type == 'copy')
1478                                 $this->_context($edit->orig);
1479                         elseif ($edit->type == 'add')
1480                                 $this->_added($edit->closing);
1481                         elseif ($edit->type == 'delete')
1482                                 $this->_deleted($edit->orig);
1483                         elseif ($edit->type == 'change')
1484                                 $this->_changed($edit->orig, $edit->closing);
1485                         else
1486                                 trigger_error('Unknown edit type', E_USER_ERROR);
1487                 }
1488                 $this->_end_block();
1489                 wfProfileOut( $fname );
1490         }
1491
1492         function _start_diff() {
1493                 ob_start();
1494         }
1495
1496         function _end_diff() {
1497                 $val = ob_get_contents();
1498                 ob_end_clean();
1499                 return $val;
1500         }
1501
1502         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1503                 if ($xlen > 1)
1504                         $xbeg .= "," . ($xbeg + $xlen - 1);
1505                 if ($ylen > 1)
1506                         $ybeg .= "," . ($ybeg + $ylen - 1);
1507
1508                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1509         }
1510
1511         function _start_block($header) {
1512                 echo $header;
1513         }
1514
1515         function _end_block() {
1516         }
1517
1518         function _lines($lines, $prefix = ' ') {
1519                 foreach ($lines as $line)
1520                         echo "$prefix $line\n";
1521         }
1522
1523         function _context($lines) {
1524                 $this->_lines($lines);
1525         }
1526
1527         function _added($lines) {
1528                 $this->_lines($lines, '>');
1529         }
1530         function _deleted($lines) {
1531                 $this->_lines($lines, '<');
1532         }
1533
1534         function _changed($orig, $closing) {
1535                 $this->_deleted($orig);
1536                 echo "---\n";
1537                 $this->_added($closing);
1538         }
1539 }
1540
1541
1542 /**
1543  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1544  *
1545  */
1546
1547 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1548
1549 /**
1550  * @todo document
1551  * @private
1552  * @package MediaWiki
1553  * @subpackage DifferenceEngine
1554  */
1555 class _HWLDF_WordAccumulator {
1556         function _HWLDF_WordAccumulator () {
1557                 $this->_lines = array();
1558                 $this->_line = '';
1559                 $this->_group = '';
1560                 $this->_tag = '';
1561         }
1562
1563         function _flushGroup ($new_tag) {
1564                 if ($this->_group !== '') {
1565                         if ($this->_tag == 'mark')
1566                                 $this->_line .= '<span class="diffchange">' .
1567                                         htmlspecialchars ( $this->_group ) . '</span>';
1568                         else
1569                                 $this->_line .= htmlspecialchars ( $this->_group );
1570                 }
1571                 $this->_group = '';
1572                 $this->_tag = $new_tag;
1573         }
1574
1575         function _flushLine ($new_tag) {
1576                 $this->_flushGroup($new_tag);
1577                 if ($this->_line != '')
1578                         array_push ( $this->_lines, $this->_line );
1579                 else
1580                         # make empty lines visible by inserting an NBSP
1581                         array_push ( $this->_lines, NBSP );
1582                 $this->_line = '';
1583         }
1584
1585         function addWords ($words, $tag = '') {
1586                 if ($tag != $this->_tag)
1587                         $this->_flushGroup($tag);
1588
1589                 foreach ($words as $word) {
1590                         // new-line should only come as first char of word.
1591                         if ($word == '')
1592                                 continue;
1593                         if ($word[0] == "\n") {
1594                                 $this->_flushLine($tag);
1595                                 $word = substr($word, 1);
1596                         }
1597                         assert(!strstr($word, "\n"));
1598                         $this->_group .= $word;
1599                 }
1600         }
1601
1602         function getLines() {
1603                 $this->_flushLine('~done');
1604                 return $this->_lines;
1605         }
1606 }
1607
1608 /**
1609  * @todo document
1610  * @private
1611  * @package MediaWiki
1612  * @subpackage DifferenceEngine
1613  */
1614 class WordLevelDiff extends MappedDiff
1615 {
1616         const MAX_LINE_LENGTH = 10000;
1617
1618         function WordLevelDiff ($orig_lines, $closing_lines) {
1619                 $fname = 'WordLevelDiff::WordLevelDiff';
1620                 wfProfileIn( $fname );
1621
1622                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1623                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1624
1625                 $this->MappedDiff($orig_words, $closing_words,
1626                                                   $orig_stripped, $closing_stripped);
1627                 wfProfileOut( $fname );
1628         }
1629
1630         function _split($lines) {
1631                 $fname = 'WordLevelDiff::_split';
1632                 wfProfileIn( $fname );
1633
1634                 $words = array();
1635                 $stripped = array();
1636                 $first = true;
1637                 foreach ( $lines as $line ) {
1638                         # If the line is too long, just pretend the entire line is one big word
1639                         # This prevents resource exhaustion problems
1640                         if ( $first ) {
1641                                 $first = false;
1642                         } else {
1643                                 $words[] = "\n";
1644                                 $stripped[] = "\n";
1645                         }
1646                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1647                                 $words[] = $line;
1648                                 $stripped[] = $line;
1649                         } else {
1650                                 $m = array();
1651                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1652                                         $line, $m))
1653                                 {
1654                                         $words = array_merge( $words, $m[0] );
1655                                         $stripped = array_merge( $stripped, $m[1] );
1656                                 }
1657                         }
1658                 }
1659                 wfProfileOut( $fname );
1660                 return array($words, $stripped);
1661         }
1662
1663         function orig () {
1664                 $fname = 'WordLevelDiff::orig';
1665                 wfProfileIn( $fname );
1666                 $orig = new _HWLDF_WordAccumulator;
1667
1668                 foreach ($this->edits as $edit) {
1669                         if ($edit->type == 'copy')
1670                                 $orig->addWords($edit->orig);
1671                         elseif ($edit->orig)
1672                                 $orig->addWords($edit->orig, 'mark');
1673                 }
1674                 $lines = $orig->getLines();
1675                 wfProfileOut( $fname );
1676                 return $lines;
1677         }
1678
1679         function closing () {
1680                 $fname = 'WordLevelDiff::closing';
1681                 wfProfileIn( $fname );
1682                 $closing = new _HWLDF_WordAccumulator;
1683
1684                 foreach ($this->edits as $edit) {
1685                         if ($edit->type == 'copy')
1686                                 $closing->addWords($edit->closing);
1687                         elseif ($edit->closing)
1688                                 $closing->addWords($edit->closing, 'mark');
1689                 }
1690                 $lines = $closing->getLines();
1691                 wfProfileOut( $fname );
1692                 return $lines;
1693         }
1694 }
1695
1696 /**
1697  *      Wikipedia Table style diff formatter.
1698  * @todo document
1699  * @private
1700  * @package MediaWiki
1701  * @subpackage DifferenceEngine
1702  */
1703 class TableDiffFormatter extends DiffFormatter
1704 {
1705         function TableDiffFormatter() {
1706                 $this->leading_context_lines = 2;
1707                 $this->trailing_context_lines = 2;
1708         }
1709
1710         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1711                 $r = '<tr><td colspan="2" align="left"><strong><!--LINE '.$xbeg."--></strong></td>\n" .
1712                   '<td colspan="2" align="left"><strong><!--LINE '.$ybeg."--></strong></td></tr>\n";
1713                 return $r;
1714         }
1715
1716         function _start_block( $header ) {
1717                 echo $header;
1718         }
1719
1720         function _end_block() {
1721         }
1722
1723         function _lines( $lines, $prefix=' ', $color='white' ) {
1724         }
1725
1726         # HTML-escape parameter before calling this
1727         function addedLine( $line ) {
1728                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1729         }
1730
1731         # HTML-escape parameter before calling this
1732         function deletedLine( $line ) {
1733                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1734         }
1735
1736         # HTML-escape parameter before calling this
1737         function contextLine( $line ) {
1738                 return "<td> </td><td class='diff-context'>{$line}</td>";
1739         }
1740
1741         function emptyLine() {
1742                 return '<td colspan="2">&nbsp;</td>';
1743         }
1744
1745         function _added( $lines ) {
1746                 foreach ($lines as $line) {
1747                         echo '<tr>' . $this->emptyLine() .
1748                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1749                 }
1750         }
1751
1752         function _deleted($lines) {
1753                 foreach ($lines as $line) {
1754                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1755                           $this->emptyLine() . "</tr>\n";
1756                 }
1757         }
1758
1759         function _context( $lines ) {
1760                 foreach ($lines as $line) {
1761                         echo '<tr>' .
1762                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1763                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1764                 }
1765         }
1766
1767         function _changed( $orig, $closing ) {
1768                 $fname = 'TableDiffFormatter::_changed';
1769                 wfProfileIn( $fname );
1770
1771                 $diff = new WordLevelDiff( $orig, $closing );
1772                 $del = $diff->orig();
1773                 $add = $diff->closing();
1774
1775                 # Notice that WordLevelDiff returns HTML-escaped output.
1776                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1777
1778                 while ( $line = array_shift( $del ) ) {
1779                         $aline = array_shift( $add );
1780                         echo '<tr>' . $this->deletedLine( $line ) .
1781                                 $this->addedLine( $aline ) . "</tr>\n";
1782                 }
1783                 foreach ($add as $line) {       # If any leftovers
1784                         echo '<tr>' . $this->emptyLine() .
1785                                 $this->addedLine( $line ) . "</tr>\n";
1786                 }
1787                 wfProfileOut( $fname );
1788         }
1789 }
1790
1791 ?>