includes/DifferenceEngine.php

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