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