includes/DifferenceEngine.php

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