includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @package MediaWiki
   5  * @subpackage DifferenceEngine
   6  */
   7
   8 /**
   9  * @todo document
  10  * @public
  11  * @package MediaWiki
  12  * @subpackage DifferenceEngine
  13  */
  14 class DifferenceEngine {
  15         /**#@+
  16          * @private
  17          */
  18         var $mOldid, $mNewid, $mTitle;
  19         var $mOldtitle, $mNewtitle, $mPagetitle;
  20         var $mOldtext, $mNewtext;
  21         var $mOldPage, $mNewPage;
  22         var $mRcidMarkPatrolled;
  23         var $mOldRev, $mNewRev;
  24         var $mRevisionsLoaded = false; // Have the revisions been loaded
  25         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  26         /**#@-*/
  27
  28         /**
  29          * Constructor
  30          * @param $titleObj Title object that the diff is associated with
  31          * @param $old Integer: old ID we want to show and diff with.
  32          * @param $new String: either 'prev' or 'next'.
  33          * @param $rcid Integer: ??? FIXME (default 0)
  34          */
  35         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0 ) {
  36                 $this->mTitle = $titleObj;
  37                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  38
  39                 if ( 'prev' === $new ) {
  40                         # Show diff between revision $old and the previous one.
  41                         # Get previous one from DB.
  42                         #
  43                         $this->mNewid = intval($old);
  44
  45                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  46
  47                 } elseif ( 'next' === $new ) {
  48                         # Show diff between revision $old and the previous one.
  49                         # Get previous one from DB.
  50                         #
  51                         $this->mOldid = intval($old);
  52                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  53                         if ( false === $this->mNewid ) {
  54                                 # if no result, NewId points to the newest old revision. The only newer
  55                                 # revision is cur, which is "0".
  56                                 $this->mNewid = 0;
  57                         }
  58
  59                 } else {
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = intval($new);
  62                 }
  63                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  64         }
  65
  66         function showDiffPage() {
  67                 global $wgUser, $wgOut, $wgContLang, $wgUseExternalEditor, $wgUseRCPatrol;
  68                 $fname = 'DifferenceEngine::showDiffPage';
  69                 wfProfileIn( $fname );
  70
  71                 # If external diffs are enabled both globally and for the user,
  72                 # we'll use the application/x-external-editor interface to call
  73                 # an external diff tool like kompare, kdiff3, etc.
  74                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  75                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  76                         $wgOut->disable();
  77                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  78                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  79                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  80                         $special=$wgLang->getNsText(NS_SPECIAL);
  81                         $control=<<<CONTROL
  82 [Process]
  83 Type=Diff text
  84 Engine=MediaWiki
  85 Script={$wgServer}{$wgScript}
  86 Special namespace={$special}
  87
  88 [File]
  89 Extension=wiki
  90 URL=$url1
  91
  92 [File 2]
  93 Extension=wiki
  94 URL=$url2
  95 CONTROL;
  96                         echo($control);
  97                         return;
  98                 }
  99
 100                 $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 101                   "{$this->mNewid})";
 102                 $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 103
 104                 $wgOut->setArticleFlag( false );
 105                 if ( ! $this->loadRevisionData() ) {
 106                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 107                         $wgOut->addWikitext( $mtext );
 108                         wfProfileOut( $fname );
 109                         return;
 110                 }
 111
 112                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 113
 114                 if ( $this->mNewRev->isCurrent() ) {
 115                         $wgOut->setArticleFlag( true );
 116                 }
 117
 118                 # mOldid is false if the difference engine is called with a "vague" query for
 119                 # a diff between a version V and its previous version V' AND the version V
 120                 # is the first version of that article. In that case, V' does not exist.
 121                 if ( $this->mOldid === false ) {
 122                         $this->showFirstRevision();
 123                         wfProfileOut( $fname );
 124                         return;
 125                 }
 126
 127                 $wgOut->suppressQuickbar();
 128
 129                 $oldTitle = $this->mOldPage->getPrefixedText();
 130                 $newTitle = $this->mNewPage->getPrefixedText();
 131                 if( $oldTitle == $newTitle ) {
 132                         $wgOut->setPageTitle( $newTitle );
 133                 } else {
 134                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 135                 }
 136                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 137                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 138
 139                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 140                         $wgOut->loginToUse();
 141                         $wgOut->output();
 142                         wfProfileOut( $fname );
 143                         exit;
 144                 }
 145
 146                 $talk = $wgContLang->getNsText( NS_TALK );
 147                 $contribs = wfMsg( 'contribslink' );
 148
 149                 if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 150                         $username = $this->mNewRev->getUserText();
 151                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . Linker::makeKnownLinkObj( $this->mTitle, wfMsg( 'rollbacklink' ),
 152                                 'action=rollback&from=' . urlencode( $username ) .
 153                                 '&token=' . urlencode( $wgUser->editToken( array( $this->mTitle->getPrefixedText(), $username ) ) ) ) .
 154                                 ']</strong>';
 155                 } else {
 156                         $rollback = '';
 157                 }
 158                 if( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isAllowed( 'patrol' ) ) {
 159                         $patrol = ' [' . Linker::makeKnownLinkObj( $this->mTitle, wfMsg( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 160                 } else {
 161                         $patrol = '';
 162                 }
 163
 164                 $prevlink = Linker::makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 165                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 166                 if ( $this->mNewRev->isCurrent() ) {
 167                         $nextlink = '&nbsp;';
 168                 } else {
 169                         $nextlink = Linker::makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 170                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 171                 }
 172
 173                 $oldminor = '';
 174                 $newminor = '';
 175
 176                 if ($this->mOldRev->mMinorEdit == 1) {
 177                         $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
 178                                 wfMsg( 'minoreditletter') ) . ' ';
 179                 }
 180
 181                 if ($this->mNewRev->mMinorEdit == 1) {
 182                         $newminor = wfElement( 'span', array( 'class' => 'minor' ),
 183                                 wfMsg( 'minoreditletter') ) . ' ';
 184                 }
 185
 186                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />" .
 187                         Linker::revUserTools( $this->mOldRev ) . "<br />" .
 188                         $oldminor . Linker::revComment( $this->mOldRev ) . "<br />" .
 189                         $prevlink;
 190                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />" .
 191                         Linker::revUserTools( $this->mNewRev ) . " $rollback<br />" .
 192                         $newminor . Linker::revComment( $this->mNewRev ) . "<br />" .
 193                         $nextlink . $patrol;
 194
 195                 $this->showDiff( $oldHeader, $newHeader );
 196                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 197
 198                 if( !$this->mNewRev->isCurrent() ) {
 199                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 200                 }
 201
 202                 $this->loadNewText();
 203                 if( is_object( $this->mNewRev ) ) {
 204                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 205                 }
 206                 $wgOut->addSecondaryWikiText( $this->mNewtext );
 207
 208                 if( !$this->mNewRev->isCurrent() ) {
 209                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 210                 }
 211
 212                 wfProfileOut( $fname );
 213         }
 214
 215         /**
 216          * Show the first revision of an article. Uses normal diff headers in
 217          * contrast to normal "old revision" display style.
 218          */
 219         function showFirstRevision() {
 220                 global $wgOut;
 221
 222                 $fname = 'DifferenceEngine::showFirstRevision';
 223                 wfProfileIn( $fname );
 224
 225                 # Get article text from the DB
 226                 #
 227                 if ( ! $this->loadNewText() ) {
 228                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 229                           "{$this->mNewid})";
 230                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 231                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 232                         $wgOut->addWikitext( $mtext );
 233                         wfProfileOut( $fname );
 234                         return;
 235                 }
 236                 if ( $this->mNewRev->isCurrent() ) {
 237                         $wgOut->setArticleFlag( true );
 238                 }
 239
 240                 # Check if user is allowed to look at this page. If not, bail out.
 241                 #
 242                 if ( !( $this->mTitle->userCanRead() ) ) {
 243                         $wgOut->loginToUse();
 244                         $wgOut->output();
 245                         wfProfileOut( $fname );
 246                         exit;
 247                 }
 248
 249                 # Prepare the header box
 250                 #
 251                 $nextlink = Linker::makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 252                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 253                         Linker::revUserTools( $this->mNewRev ) . "<br />" .
 254                         Linker::revComment( $this->mNewRev ) . "<br />" .
 255                         $nextlink . "</div>\n";
 256
 257                 $wgOut->addHTML( $header );
 258
 259                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 260                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 261
 262
 263                 # Show current revision
 264                 #
 265                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 266                 if( is_object( $this->mNewRev ) ) {
 267                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 268                 }
 269                 $wgOut->addSecondaryWikiText( $this->mNewtext );
 270
 271                 wfProfileOut( $fname );
 272         }
 273
 274         /**
 275          * Get the diff text, send it to $wgOut
 276          * Returns false if the diff could not be generated, otherwise returns true
 277          */
 278         function showDiff( $otitle, $ntitle ) {
 279                 global $wgOut;
 280                 $diff = $this->getDiff( $otitle, $ntitle );
 281                 if ( $diff === false ) {
 282                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
 283                         return false;
 284                 } else {
 285                         $wgOut->addHTML( $diff );
 286                         return true;
 287                 }
 288         }
 289
 290         /**
 291          * Get diff table, including header
 292          * Note that the interface has changed, it's no longer static.
 293          * Returns false on error
 294          */
 295         function getDiff( $otitle, $ntitle ) {
 296                 $body = $this->getDiffBody();
 297                 if ( $body === false ) {
 298                         return false;
 299                 } else {
 300                         return $this->addHeader( $body, $otitle, $ntitle );
 301                 }
 302         }
 303
 304         /**
 305          * Get the diff table body, without header
 306          * Results are cached
 307          * Returns false on error
 308          */
 309         function getDiffBody() {
 310                 global $wgMemc;
 311                 $fname = 'DifferenceEngine::getDiffBody';
 312                 wfProfileIn( $fname );
 313
 314                 // Cacheable?
 315                 $key = false;
 316                 if ( $this->mOldid && $this->mNewid ) {
 317                         // Try cache
 318                         $key = wfMemcKey( 'diff', 'oldid', $this->mOldid, 'newid', $this->mNewid );
 319                         $difftext = $wgMemc->get( $key );
 320                         if ( $difftext ) {
 321                                 wfIncrStats( 'diff_cache_hit' );
 322                                 $difftext = $this->localiseLineNumbers( $difftext );
 323                                 $difftext .= "\n<!-- diff cache key $key -->\n";
 324                                 wfProfileOut( $fname );
 325                                 return $difftext;
 326                         }
 327                 }
 328
 329                 if ( !$this->loadText() ) {
 330                         wfProfileOut( $fname );
 331                         return false;
 332                 }
 333
 334                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 335
 336                 // Save to cache for 7 days
 337                 if ( $key !== false && $difftext !== false ) {
 338                         wfIncrStats( 'diff_cache_miss' );
 339                         $wgMemc->set( $key, $difftext, 7*86400 );
 340                 } else {
 341                         wfIncrStats( 'diff_uncacheable' );
 342                 }
 343                 // Replace line numbers with the text in the user's language
 344                 if ( $difftext !== false ) {
 345                         $difftext = $this->localiseLineNumbers( $difftext );
 346                 }
 347                 wfProfileOut( $fname );
 348                 return $difftext;
 349         }
 350
 351         /**
 352          * Generate a diff, no caching
 353          * $otext and $ntext must be already segmented
 354          */
 355         function generateDiffBody( $otext, $ntext ) {
 356                 global $wgExternalDiffEngine, $wgContLang;
 357                 $fname = 'DifferenceEngine::generateDiffBody';
 358
 359                 $otext = str_replace( "\r\n", "\n", $otext );
 360                 $ntext = str_replace( "\r\n", "\n", $ntext );
 361
 362                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 363                         # For historical reasons, external diff engine expects
 364                         # input text to be HTML-escaped already
 365                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 366                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 367                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 368                                 dl('php_wikidiff.so');
 369                         }
 370                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 371                 }
 372
 373                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 374                         # Better external diff engine, the 2 may some day be dropped
 375                         # This one does the escaping and segmenting itself
 376                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 377                                 wfProfileIn( "$fname-dl" );
 378                                 @dl('php_wikidiff2.so');
 379                                 wfProfileOut( "$fname-dl" );
 380                         }
 381                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 382                                 wfProfileIn( 'wikidiff2_do_diff' );
 383                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 384                                 wfProfileOut( 'wikidiff2_do_diff' );
 385                                 return $text;
 386                         }
 387                 }
 388                 if ( $wgExternalDiffEngine !== false ) {
 389                         # Diff via the shell
 390                         global $wgTmpDirectory;
 391                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 392                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 393
 394                         $tempFile1 = fopen( $tempName1, "w" );
 395                         if ( !$tempFile1 ) {
 396                                 wfProfileOut( $fname );
 397                                 return false;
 398                         }
 399                         $tempFile2 = fopen( $tempName2, "w" );
 400                         if ( !$tempFile2 ) {
 401                                 wfProfileOut( $fname );
 402                                 return false;
 403                         }
 404                         fwrite( $tempFile1, $otext );
 405                         fwrite( $tempFile2, $ntext );
 406                         fclose( $tempFile1 );
 407                         fclose( $tempFile2 );
 408                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 409                         wfProfileIn( "$fname-shellexec" );
 410                         $difftext = wfShellExec( $cmd );
 411                         wfProfileOut( "$fname-shellexec" );
 412                         unlink( $tempName1 );
 413                         unlink( $tempName2 );
 414                         return $difftext;
 415                 }
 416
 417                 # Native PHP diff
 418                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 419                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 420                 $diffs = new Diff( $ota, $nta );
 421                 $formatter = new TableDiffFormatter();
 422                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 423         }
 424
 425
 426         /**
 427          * Replace line numbers with the text in the user's language
 428          */
 429         function localiseLineNumbers( $text ) {
 430                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 431                         array( &$this, 'localiseLineNumbersCb' ), $text );
 432         }
 433
 434         function localiseLineNumbersCb( $matches ) {
 435                 global $wgLang;
 436                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 437         }
 438
 439         /**
 440          * Add the header to a diff body
 441          */
 442         function addHeader( $diff, $otitle, $ntitle ) {
 443                 $out = "
 444                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 445                         <tr>
 446                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 447                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 448                         </tr>
 449                         $diff
 450                         </table>
 451                 ";
 452                 return $out;
 453         }
 454
 455         /**
 456          * Use specified text instead of loading from the database
 457          */
 458         function setText( $oldText, $newText ) {
 459                 $this->mOldtext = $oldText;
 460                 $this->mNewtext = $newText;
 461                 $this->mTextLoaded = 2;
 462         }
 463
 464         /**
 465          * Load revision metadata for the specified articles. If newid is 0, then compare
 466          * the old article in oldid to the current article; if oldid is 0, then
 467          * compare the current article to the immediately previous one (ignoring the
 468          * value of newid).
 469          *
 470          * If oldid is false, leave the corresponding revision object set
 471          * to false. This is impossible via ordinary user input, and is provided for
 472          * API convenience.
 473          */
 474         function loadRevisionData() {
 475                 global $wgLang;
 476                 if ( $this->mRevisionsLoaded ) {
 477                         return true;
 478                 } else {
 479                         // Whether it succeeds or fails, we don't want to try again
 480                         $this->mRevisionsLoaded = true;
 481                 }
 482
 483                 // Load the new revision object
 484                 if( $this->mNewid ) {
 485                         $this->mNewRev = Revision::newFromId( $this->mNewid );
 486                 } else {
 487                         $this->mNewRev = Revision::newFromTitle( $this->mTitle );
 488                 }
 489
 490                 if( is_null( $this->mNewRev ) ) {
 491                         return false;
 492                 }
 493
 494                 // Set assorted variables
 495                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 496                 $this->mNewPage = $this->mNewRev->getTitle();
 497                 if( $this->mNewRev->isCurrent() ) {
 498                         $newLink = $this->mNewPage->escapeLocalUrl();
 499                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 500                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 501
 502                         $this->mNewtitle = "<strong><a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)</strong>"
 503                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 504
 505                 } else {
 506                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 507                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 508                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 509
 510                         $this->mNewtitle = "<strong><a href='$newLink'>{$this->mPagetitle}</a></strong>"
 511                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 512                 }
 513
 514                 // Load the old revision object
 515                 $this->mOldRev = false;
 516                 if( $this->mOldid ) {
 517                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 518                 } elseif ( $this->mOldid === 0 ) {
 519                         $rev = $this->mNewRev->getPrevious();
 520                         if( $rev ) {
 521                                 $this->mOldid = $rev->getId();
 522                                 $this->mOldRev = $rev;
 523                         } else {
 524                                 // No previous revision; mark to show as first-version only.
 525                                 $this->mOldid = false;
 526                                 $this->mOldRev = false;
 527                         }
 528                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 529
 530                 if( is_null( $this->mOldRev ) ) {
 531                         return false;
 532                 }
 533
 534                 if ( $this->mOldRev ) {
 535                         $this->mOldPage = $this->mOldRev->getTitle();
 536
 537                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 538                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 539                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 540                         $this->mOldtitle = "<strong><a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) )
 541                                 . "</a></strong> (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 542                 }
 543
 544                 return true;
 545         }
 546
 547         /**
 548          * Load the text of the revisions, as well as revision data.
 549          */
 550         function loadText() {
 551                 if ( $this->mTextLoaded == 2 ) {
 552                         return true;
 553                 } else {
 554                         // Whether it succeeds or fails, we don't want to try again
 555                         $this->mTextLoaded = 2;
 556                 }
 557
 558                 if ( !$this->loadRevisionData() ) {
 559                         return false;
 560                 }
 561                 if ( $this->mOldRev ) {
 562                         // FIXME: permission tests
 563                         $this->mOldtext = $this->mOldRev->getText();
 564                         if ( $this->mOldtext === false ) {
 565                                 return false;
 566                         }
 567                 }
 568                 if ( $this->mNewRev ) {
 569                         $this->mNewtext = $this->mNewRev->getText();
 570                         if ( $this->mNewtext === false ) {
 571                                 return false;
 572                         }
 573                 }
 574                 return true;
 575         }
 576
 577         /**
 578          * Load the text of the new revision, not the old one
 579          */
 580         function loadNewText() {
 581                 if ( $this->mTextLoaded >= 1 ) {
 582                         return true;
 583                 } else {
 584                         $this->mTextLoaded = 1;
 585                 }
 586                 if ( !$this->loadRevisionData() ) {
 587                         return false;
 588                 }
 589                 $this->mNewtext = $this->mNewRev->getText();
 590                 return true;
 591         }
 592
 593
 594 }
 595
 596 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 597 //
 598 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 599 // You may copy this code freely under the conditions of the GPL.
 600 //
 601
 602 define('USE_ASSERTS', function_exists('assert'));
 603
 604 /**
 605  * @todo document
 606  * @private
 607  * @package MediaWiki
 608  * @subpackage DifferenceEngine
 609  */
 610 class _DiffOp {
 611         var $type;
 612         var $orig;
 613         var $closing;
 614
 615         function reverse() {
 616                 trigger_error('pure virtual', E_USER_ERROR);
 617         }
 618
 619         function norig() {
 620                 return $this->orig ? sizeof($this->orig) : 0;
 621         }
 622
 623         function nclosing() {
 624                 return $this->closing ? sizeof($this->closing) : 0;
 625         }
 626 }
 627
 628 /**
 629  * @todo document
 630  * @private
 631  * @package MediaWiki
 632  * @subpackage DifferenceEngine
 633  */
 634 class _DiffOp_Copy extends _DiffOp {
 635         var $type = 'copy';
 636
 637         function _DiffOp_Copy ($orig, $closing = false) {
 638                 if (!is_array($closing))
 639                         $closing = $orig;
 640                 $this->orig = $orig;
 641                 $this->closing = $closing;
 642         }
 643
 644         function reverse() {
 645                 return new _DiffOp_Copy($this->closing, $this->orig);
 646         }
 647 }
 648
 649 /**
 650  * @todo document
 651  * @private
 652  * @package MediaWiki
 653  * @subpackage DifferenceEngine
 654  */
 655 class _DiffOp_Delete extends _DiffOp {
 656         var $type = 'delete';
 657
 658         function _DiffOp_Delete ($lines) {
 659                 $this->orig = $lines;
 660                 $this->closing = false;
 661         }
 662
 663         function reverse() {
 664                 return new _DiffOp_Add($this->orig);
 665         }
 666 }
 667
 668 /**
 669  * @todo document
 670  * @private
 671  * @package MediaWiki
 672  * @subpackage DifferenceEngine
 673  */
 674 class _DiffOp_Add extends _DiffOp {
 675         var $type = 'add';
 676
 677         function _DiffOp_Add ($lines) {
 678                 $this->closing = $lines;
 679                 $this->orig = false;
 680         }
 681
 682         function reverse() {
 683                 return new _DiffOp_Delete($this->closing);
 684         }
 685 }
 686
 687 /**
 688  * @todo document
 689  * @private
 690  * @package MediaWiki
 691  * @subpackage DifferenceEngine
 692  */
 693 class _DiffOp_Change extends _DiffOp {
 694         var $type = 'change';
 695
 696         function _DiffOp_Change ($orig, $closing) {
 697                 $this->orig = $orig;
 698                 $this->closing = $closing;
 699         }
 700
 701         function reverse() {
 702                 return new _DiffOp_Change($this->closing, $this->orig);
 703         }
 704 }
 705
 706
 707 /**
 708  * Class used internally by Diff to actually compute the diffs.
 709  *
 710  * The algorithm used here is mostly lifted from the perl module
 711  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 712  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 713  *
 714  * More ideas are taken from:
 715  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 716  *
 717  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 718  * diffutils-2.7, which can be found at:
 719  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 720  *
 721  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 722  * are my own.
 723  *
 724  * Line length limits for robustness added by Tim Starling, 2005-08-31
 725  *
 726  * @author Geoffrey T. Dairiki, Tim Starling
 727  * @private
 728  * @package MediaWiki
 729  * @subpackage DifferenceEngine
 730  */
 731 class _DiffEngine
 732 {
 733         const MAX_XREF_LENGTH =  10000;
 734
 735         function diff ($from_lines, $to_lines) {
 736                 $fname = '_DiffEngine::diff';
 737                 wfProfileIn( $fname );
 738
 739                 $n_from = sizeof($from_lines);
 740                 $n_to = sizeof($to_lines);
 741
 742                 $this->xchanged = $this->ychanged = array();
 743                 $this->xv = $this->yv = array();
 744                 $this->xind = $this->yind = array();
 745                 unset($this->seq);
 746                 unset($this->in_seq);
 747                 unset($this->lcs);
 748
 749                 // Skip leading common lines.
 750                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 751                         if ($from_lines[$skip] !== $to_lines[$skip])
 752                                 break;
 753                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 754                 }
 755                 // Skip trailing common lines.
 756                 $xi = $n_from; $yi = $n_to;
 757                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 758                         if ($from_lines[$xi] !== $to_lines[$yi])
 759                                 break;
 760                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 761                 }
 762
 763                 // Ignore lines which do not exist in both files.
 764                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 765                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 766                 }
 767
 768                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 769                         $line = $to_lines[$yi];
 770                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 771                                 continue;
 772                         $yhash[$this->_line_hash($line)] = 1;
 773                         $this->yv[] = $line;
 774                         $this->yind[] = $yi;
 775                 }
 776                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 777                         $line = $from_lines[$xi];
 778                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 779                                 continue;
 780                         $this->xv[] = $line;
 781                         $this->xind[] = $xi;
 782                 }
 783
 784                 // Find the LCS.
 785                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 786
 787                 // Merge edits when possible
 788                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 789                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 790
 791                 // Compute the edit operations.
 792                 $edits = array();
 793                 $xi = $yi = 0;
 794                 while ($xi < $n_from || $yi < $n_to) {
 795                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 796                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 797
 798                         // Skip matching "snake".
 799                         $copy = array();
 800                         while ( $xi < $n_from && $yi < $n_to
 801                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 802                                 $copy[] = $from_lines[$xi++];
 803                                 ++$yi;
 804                         }
 805                         if ($copy)
 806                                 $edits[] = new _DiffOp_Copy($copy);
 807
 808                         // Find deletes & adds.
 809                         $delete = array();
 810                         while ($xi < $n_from && $this->xchanged[$xi])
 811                                 $delete[] = $from_lines[$xi++];
 812
 813                         $add = array();
 814                         while ($yi < $n_to && $this->ychanged[$yi])
 815                                 $add[] = $to_lines[$yi++];
 816
 817                         if ($delete && $add)
 818                                 $edits[] = new _DiffOp_Change($delete, $add);
 819                         elseif ($delete)
 820                                 $edits[] = new _DiffOp_Delete($delete);
 821                         elseif ($add)
 822                                 $edits[] = new _DiffOp_Add($add);
 823                 }
 824                 wfProfileOut( $fname );
 825                 return $edits;
 826         }
 827
 828         /**
 829          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 830          */
 831         function _line_hash( $line ) {
 832                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 833                         return md5( $line );
 834                 } else {
 835                         return $line;
 836                 }
 837         }
 838
 839
 840         /* Divide the Largest Common Subsequence (LCS) of the sequences
 841          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 842          * sized segments.
 843          *
 844          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 845          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 846          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 847          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 848          * that (X0, Y0) == (XOFF, YOFF) and
 849          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 850          *
 851          * This function assumes that the first lines of the specified portions
 852          * of the two files do not match, and likewise that the last lines do not
 853          * match.  The caller must trim matching lines from the beginning and end
 854          * of the portions it is going to specify.
 855          */
 856         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 857                 $fname = '_DiffEngine::_diag';
 858                 wfProfileIn( $fname );
 859                 $flip = false;
 860
 861                 if ($xlim - $xoff > $ylim - $yoff) {
 862                         // Things seems faster (I'm not sure I understand why)
 863                                 // when the shortest sequence in X.
 864                                 $flip = true;
 865                         list ($xoff, $xlim, $yoff, $ylim)
 866                         = array( $yoff, $ylim, $xoff, $xlim);
 867                 }
 868
 869                 if ($flip)
 870                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 871                                 $ymatches[$this->xv[$i]][] = $i;
 872                 else
 873                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 874                                 $ymatches[$this->yv[$i]][] = $i;
 875
 876                 $this->lcs = 0;
 877                 $this->seq[0]= $yoff - 1;
 878                 $this->in_seq = array();
 879                 $ymids[0] = array();
 880
 881                 $numer = $xlim - $xoff + $nchunks - 1;
 882                 $x = $xoff;
 883                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 884                         wfProfileIn( "$fname-chunk" );
 885                         if ($chunk > 0)
 886                                 for ($i = 0; $i <= $this->lcs; $i++)
 887                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 888
 889                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 890                         for ( ; $x < $x1; $x++) {
 891                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 892                                         if (empty($ymatches[$line]))
 893                                                 continue;
 894                                 $matches = $ymatches[$line];
 895                                 reset($matches);
 896                                 while (list ($junk, $y) = each($matches))
 897                                         if (empty($this->in_seq[$y])) {
 898                                                 $k = $this->_lcs_pos($y);
 899                                                 USE_ASSERTS && assert($k > 0);
 900                                                 $ymids[$k] = $ymids[$k-1];
 901                                                 break;
 902                                         }
 903                                 while (list ($junk, $y) = each($matches)) {
 904                                         if ($y > $this->seq[$k-1]) {
 905                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 906                                                 // Optimization: this is a common case:
 907                                                 //      next match is just replacing previous match.
 908                                                 $this->in_seq[$this->seq[$k]] = false;
 909                                                 $this->seq[$k] = $y;
 910                                                 $this->in_seq[$y] = 1;
 911                                         } else if (empty($this->in_seq[$y])) {
 912                                                 $k = $this->_lcs_pos($y);
 913                                                 USE_ASSERTS && assert($k > 0);
 914                                                 $ymids[$k] = $ymids[$k-1];
 915                                         }
 916                                 }
 917                         }
 918                         wfProfileOut( "$fname-chunk" );
 919                 }
 920
 921                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 922                 $ymid = $ymids[$this->lcs];
 923                 for ($n = 0; $n < $nchunks - 1; $n++) {
 924                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 925                         $y1 = $ymid[$n] + 1;
 926                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 927                 }
 928                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 929
 930                 wfProfileOut( $fname );
 931                 return array($this->lcs, $seps);
 932         }
 933
 934         function _lcs_pos ($ypos) {
 935                 $fname = '_DiffEngine::_lcs_pos';
 936                 wfProfileIn( $fname );
 937
 938                 $end = $this->lcs;
 939                 if ($end == 0 || $ypos > $this->seq[$end]) {
 940                         $this->seq[++$this->lcs] = $ypos;
 941                         $this->in_seq[$ypos] = 1;
 942                         wfProfileOut( $fname );
 943                         return $this->lcs;
 944                 }
 945
 946                 $beg = 1;
 947                 while ($beg < $end) {
 948                         $mid = (int)(($beg + $end) / 2);
 949                         if ( $ypos > $this->seq[$mid] )
 950                                 $beg = $mid + 1;
 951                         else
 952                                 $end = $mid;
 953                 }
 954
 955                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 956
 957                 $this->in_seq[$this->seq[$end]] = false;
 958                 $this->seq[$end] = $ypos;
 959                 $this->in_seq[$ypos] = 1;
 960                 wfProfileOut( $fname );
 961                 return $end;
 962         }
 963
 964         /* Find LCS of two sequences.
 965          *
 966          * The results are recorded in the vectors $this->{x,y}changed[], by
 967          * storing a 1 in the element for each line that is an insertion
 968          * or deletion (ie. is not in the LCS).
 969          *
 970          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 971          *
 972          * Note that XLIM, YLIM are exclusive bounds.
 973          * All line numbers are origin-0 and discarded lines are not counted.
 974          */
 975         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 976                 $fname = '_DiffEngine::_compareseq';
 977                 wfProfileIn( $fname );
 978
 979                 // Slide down the bottom initial diagonal.
 980                 while ($xoff < $xlim && $yoff < $ylim
 981                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 982                         ++$xoff;
 983                         ++$yoff;
 984                 }
 985
 986                 // Slide up the top initial diagonal.
 987                 while ($xlim > $xoff && $ylim > $yoff
 988                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 989                         --$xlim;
 990                         --$ylim;
 991                 }
 992
 993                 if ($xoff == $xlim || $yoff == $ylim)
 994                         $lcs = 0;
 995                 else {
 996                         // This is ad hoc but seems to work well.
 997                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 998                         //$nchunks = max(2,min(8,(int)$nchunks));
 999                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1000                         list ($lcs, $seps)
1001                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1002                 }
1003
1004                 if ($lcs == 0) {
1005                         // X and Y sequences have no common subsequence:
1006                         // mark all changed.
1007                         while ($yoff < $ylim)
1008                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1009                         while ($xoff < $xlim)
1010                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1011                 } else {
1012                         // Use the partitions to split this problem into subproblems.
1013                         reset($seps);
1014                         $pt1 = $seps[0];
1015                         while ($pt2 = next($seps)) {
1016                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1017                                 $pt1 = $pt2;
1018                         }
1019                 }
1020                 wfProfileOut( $fname );
1021         }
1022
1023         /* Adjust inserts/deletes of identical lines to join changes
1024          * as much as possible.
1025          *
1026          * We do something when a run of changed lines include a
1027          * line at one end and has an excluded, identical line at the other.
1028          * We are free to choose which identical line is included.
1029          * `compareseq' usually chooses the one at the beginning,
1030          * but usually it is cleaner to consider the following identical line
1031          * to be the "change".
1032          *
1033          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1034          */
1035         function _shift_boundaries ($lines, &$changed, $other_changed) {
1036                 $fname = '_DiffEngine::_shift_boundaries';
1037                 wfProfileIn( $fname );
1038                 $i = 0;
1039                 $j = 0;
1040
1041                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1042                 $len = sizeof($lines);
1043                 $other_len = sizeof($other_changed);
1044
1045                 while (1) {
1046                         /*
1047                          * Scan forwards to find beginning of another run of changes.
1048                          * Also keep track of the corresponding point in the other file.
1049                          *
1050                          * Throughout this code, $i and $j are adjusted together so that
1051                          * the first $i elements of $changed and the first $j elements
1052                          * of $other_changed both contain the same number of zeros
1053                          * (unchanged lines).
1054                          * Furthermore, $j is always kept so that $j == $other_len or
1055                          * $other_changed[$j] == false.
1056                          */
1057                         while ($j < $other_len && $other_changed[$j])
1058                                 $j++;
1059
1060                         while ($i < $len && ! $changed[$i]) {
1061                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1062                                 $i++; $j++;
1063                                 while ($j < $other_len && $other_changed[$j])
1064                                         $j++;
1065                         }
1066
1067                         if ($i == $len)
1068                                 break;
1069
1070                         $start = $i;
1071
1072                         // Find the end of this run of changes.
1073                         while (++$i < $len && $changed[$i])
1074                                 continue;
1075
1076                         do {
1077                                 /*
1078                                  * Record the length of this run of changes, so that
1079                                  * we can later determine whether the run has grown.
1080                                  */
1081                                 $runlength = $i - $start;
1082
1083                                 /*
1084                                  * Move the changed region back, so long as the
1085                                  * previous unchanged line matches the last changed one.
1086                                  * This merges with previous changed regions.
1087                                  */
1088                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1089                                         $changed[--$start] = 1;
1090                                         $changed[--$i] = false;
1091                                         while ($start > 0 && $changed[$start - 1])
1092                                                 $start--;
1093                                         USE_ASSERTS && assert('$j > 0');
1094                                         while ($other_changed[--$j])
1095                                                 continue;
1096                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1097                                 }
1098
1099                                 /*
1100                                  * Set CORRESPONDING to the end of the changed run, at the last
1101                                  * point where it corresponds to a changed run in the other file.
1102                                  * CORRESPONDING == LEN means no such point has been found.
1103                                  */
1104                                 $corresponding = $j < $other_len ? $i : $len;
1105
1106                                 /*
1107                                  * Move the changed region forward, so long as the
1108                                  * first changed line matches the following unchanged one.
1109                                  * This merges with following changed regions.
1110                                  * Do this second, so that if there are no merges,
1111                                  * the changed region is moved forward as far as possible.
1112                                  */
1113                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1114                                         $changed[$start++] = false;
1115                                         $changed[$i++] = 1;
1116                                         while ($i < $len && $changed[$i])
1117                                                 $i++;
1118
1119                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1120                                         $j++;
1121                                         if ($j < $other_len && $other_changed[$j]) {
1122                                                 $corresponding = $i;
1123                                                 while ($j < $other_len && $other_changed[$j])
1124                                                         $j++;
1125                                         }
1126                                 }
1127                         } while ($runlength != $i - $start);
1128
1129                         /*
1130                          * If possible, move the fully-merged run of changes
1131                          * back to a corresponding run in the other file.
1132                          */
1133                         while ($corresponding < $i) {
1134                                 $changed[--$start] = 1;
1135                                 $changed[--$i] = 0;
1136                                 USE_ASSERTS && assert('$j > 0');
1137                                 while ($other_changed[--$j])
1138                                         continue;
1139                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1140                         }
1141                 }
1142                 wfProfileOut( $fname );
1143         }
1144 }
1145
1146 /**
1147  * Class representing a 'diff' between two sequences of strings.
1148  * @todo document
1149  * @private
1150  * @package MediaWiki
1151  * @subpackage DifferenceEngine
1152  */
1153 class Diff
1154 {
1155         var $edits;
1156
1157         /**
1158          * Constructor.
1159          * Computes diff between sequences of strings.
1160          *
1161          * @param $from_lines array An array of strings.
1162          *                (Typically these are lines from a file.)
1163          * @param $to_lines array An array of strings.
1164          */
1165         function Diff($from_lines, $to_lines) {
1166                 $eng = new _DiffEngine;
1167                 $this->edits = $eng->diff($from_lines, $to_lines);
1168                 //$this->_check($from_lines, $to_lines);
1169         }
1170
1171         /**
1172          * Compute reversed Diff.
1173          *
1174          * SYNOPSIS:
1175          *
1176          *      $diff = new Diff($lines1, $lines2);
1177          *      $rev = $diff->reverse();
1178          * @return object A Diff object representing the inverse of the
1179          *                                original diff.
1180          */
1181         function reverse () {
1182                 $rev = $this;
1183                 $rev->edits = array();
1184                 foreach ($this->edits as $edit) {
1185                         $rev->edits[] = $edit->reverse();
1186                 }
1187                 return $rev;
1188         }
1189
1190         /**
1191          * Check for empty diff.
1192          *
1193          * @return bool True iff two sequences were identical.
1194          */
1195         function isEmpty () {
1196                 foreach ($this->edits as $edit) {
1197                         if ($edit->type != 'copy')
1198                                 return false;
1199                 }
1200                 return true;
1201         }
1202
1203         /**
1204          * Compute the length of the Longest Common Subsequence (LCS).
1205          *
1206          * This is mostly for diagnostic purposed.
1207          *
1208          * @return int The length of the LCS.
1209          */
1210         function lcs () {
1211                 $lcs = 0;
1212                 foreach ($this->edits as $edit) {
1213                         if ($edit->type == 'copy')
1214                                 $lcs += sizeof($edit->orig);
1215                 }
1216                 return $lcs;
1217         }
1218
1219         /**
1220          * Get the original set of lines.
1221          *
1222          * This reconstructs the $from_lines parameter passed to the
1223          * constructor.
1224          *
1225          * @return array The original sequence of strings.
1226          */
1227         function orig() {
1228                 $lines = array();
1229
1230                 foreach ($this->edits as $edit) {
1231                         if ($edit->orig)
1232                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1233                 }
1234                 return $lines;
1235         }
1236
1237         /**
1238          * Get the closing set of lines.
1239          *
1240          * This reconstructs the $to_lines parameter passed to the
1241          * constructor.
1242          *
1243          * @return array The sequence of strings.
1244          */
1245         function closing() {
1246                 $lines = array();
1247
1248                 foreach ($this->edits as $edit) {
1249                         if ($edit->closing)
1250                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1251                 }
1252                 return $lines;
1253         }
1254
1255         /**
1256          * Check a Diff for validity.
1257          *
1258          * This is here only for debugging purposes.
1259          */
1260         function _check ($from_lines, $to_lines) {
1261                 $fname = 'Diff::_check';
1262                 wfProfileIn( $fname );
1263                 if (serialize($from_lines) != serialize($this->orig()))
1264                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1265                 if (serialize($to_lines) != serialize($this->closing()))
1266                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1267
1268                 $rev = $this->reverse();
1269                 if (serialize($to_lines) != serialize($rev->orig()))
1270                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1271                 if (serialize($from_lines) != serialize($rev->closing()))
1272                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1273
1274
1275                 $prevtype = 'none';
1276                 foreach ($this->edits as $edit) {
1277                         if ( $prevtype == $edit->type )
1278                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1279                         $prevtype = $edit->type;
1280                 }
1281
1282                 $lcs = $this->lcs();
1283                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1284                 wfProfileOut( $fname );
1285         }
1286 }
1287
1288 /**
1289  * FIXME: bad name.
1290  * @todo document
1291  * @private
1292  * @package MediaWiki
1293  * @subpackage DifferenceEngine
1294  */
1295 class MappedDiff extends Diff
1296 {
1297         /**
1298          * Constructor.
1299          *
1300          * Computes diff between sequences of strings.
1301          *
1302          * This can be used to compute things like
1303          * case-insensitve diffs, or diffs which ignore
1304          * changes in white-space.
1305          *
1306          * @param $from_lines array An array of strings.
1307          *      (Typically these are lines from a file.)
1308          *
1309          * @param $to_lines array An array of strings.
1310          *
1311          * @param $mapped_from_lines array This array should
1312          *      have the same size number of elements as $from_lines.
1313          *      The elements in $mapped_from_lines and
1314          *      $mapped_to_lines are what is actually compared
1315          *      when computing the diff.
1316          *
1317          * @param $mapped_to_lines array This array should
1318          *      have the same number of elements as $to_lines.
1319          */
1320         function MappedDiff($from_lines, $to_lines,
1321                                                 $mapped_from_lines, $mapped_to_lines) {
1322                 $fname = 'MappedDiff::MappedDiff';
1323                 wfProfileIn( $fname );
1324
1325                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1326                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1327
1328                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1329
1330                 $xi = $yi = 0;
1331                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1332                         $orig = &$this->edits[$i]->orig;
1333                         if (is_array($orig)) {
1334                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1335                                 $xi += sizeof($orig);
1336                         }
1337
1338                         $closing = &$this->edits[$i]->closing;
1339                         if (is_array($closing)) {
1340                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1341                                 $yi += sizeof($closing);
1342                         }
1343                 }
1344                 wfProfileOut( $fname );
1345         }
1346 }
1347
1348 /**
1349  * A class to format Diffs
1350  *
1351  * This class formats the diff in classic diff format.
1352  * It is intended that this class be customized via inheritance,
1353  * to obtain fancier outputs.
1354  * @todo document
1355  * @private
1356  * @package MediaWiki
1357  * @subpackage DifferenceEngine
1358  */
1359 class DiffFormatter
1360 {
1361         /**
1362          * Number of leading context "lines" to preserve.
1363          *
1364          * This should be left at zero for this class, but subclasses
1365          * may want to set this to other values.
1366          */
1367         var $leading_context_lines = 0;
1368
1369         /**
1370          * Number of trailing context "lines" to preserve.
1371          *
1372          * This should be left at zero for this class, but subclasses
1373          * may want to set this to other values.
1374          */
1375         var $trailing_context_lines = 0;
1376
1377         /**
1378          * Format a diff.
1379          *
1380          * @param $diff object A Diff object.
1381          * @return string The formatted output.
1382          */
1383         function format($diff) {
1384                 $fname = 'DiffFormatter::format';
1385                 wfProfileIn( $fname );
1386
1387                 $xi = $yi = 1;
1388                 $block = false;
1389                 $context = array();
1390
1391                 $nlead = $this->leading_context_lines;
1392                 $ntrail = $this->trailing_context_lines;
1393
1394                 $this->_start_diff();
1395
1396                 foreach ($diff->edits as $edit) {
1397                         if ($edit->type == 'copy') {
1398                                 if (is_array($block)) {
1399                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1400                                                 $block[] = $edit;
1401                                         }
1402                                         else{
1403                                                 if ($ntrail) {
1404                                                         $context = array_slice($edit->orig, 0, $ntrail);
1405                                                         $block[] = new _DiffOp_Copy($context);
1406                                                 }
1407                                                 $this->_block($x0, $ntrail + $xi - $x0,
1408                                                                           $y0, $ntrail + $yi - $y0,
1409                                                                           $block);
1410                                                 $block = false;
1411                                         }
1412                                 }
1413                                 $context = $edit->orig;
1414                         }
1415                         else {
1416                                 if (! is_array($block)) {
1417                                         $context = array_slice($context, sizeof($context) - $nlead);
1418                                         $x0 = $xi - sizeof($context);
1419                                         $y0 = $yi - sizeof($context);
1420                                         $block = array();
1421                                         if ($context)
1422                                                 $block[] = new _DiffOp_Copy($context);
1423                                 }
1424                                 $block[] = $edit;
1425                         }
1426
1427                         if ($edit->orig)
1428                                 $xi += sizeof($edit->orig);
1429                         if ($edit->closing)
1430                                 $yi += sizeof($edit->closing);
1431                 }
1432
1433                 if (is_array($block))
1434                         $this->_block($x0, $xi - $x0,
1435                                                   $y0, $yi - $y0,
1436                                                   $block);
1437
1438                 $end = $this->_end_diff();
1439                 wfProfileOut( $fname );
1440                 return $end;
1441         }
1442
1443         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1444                 $fname = 'DiffFormatter::_block';
1445                 wfProfileIn( $fname );
1446                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1447                 foreach ($edits as $edit) {
1448                         if ($edit->type == 'copy')
1449                                 $this->_context($edit->orig);
1450                         elseif ($edit->type == 'add')
1451                                 $this->_added($edit->closing);
1452                         elseif ($edit->type == 'delete')
1453                                 $this->_deleted($edit->orig);
1454                         elseif ($edit->type == 'change')
1455                                 $this->_changed($edit->orig, $edit->closing);
1456                         else
1457                                 trigger_error('Unknown edit type', E_USER_ERROR);
1458                 }
1459                 $this->_end_block();
1460                 wfProfileOut( $fname );
1461         }
1462
1463         function _start_diff() {
1464                 ob_start();
1465         }
1466
1467         function _end_diff() {
1468                 $val = ob_get_contents();
1469                 ob_end_clean();
1470                 return $val;
1471         }
1472
1473         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1474                 if ($xlen > 1)
1475                         $xbeg .= "," . ($xbeg + $xlen - 1);
1476                 if ($ylen > 1)
1477                         $ybeg .= "," . ($ybeg + $ylen - 1);
1478
1479                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1480         }
1481
1482         function _start_block($header) {
1483                 echo $header;
1484         }
1485
1486         function _end_block() {
1487         }
1488
1489         function _lines($lines, $prefix = ' ') {
1490                 foreach ($lines as $line)
1491                         echo "$prefix $line\n";
1492         }
1493
1494         function _context($lines) {
1495                 $this->_lines($lines);
1496         }
1497
1498         function _added($lines) {
1499                 $this->_lines($lines, '>');
1500         }
1501         function _deleted($lines) {
1502                 $this->_lines($lines, '<');
1503         }
1504
1505         function _changed($orig, $closing) {
1506                 $this->_deleted($orig);
1507                 echo "---\n";
1508                 $this->_added($closing);
1509         }
1510 }
1511
1512
1513 /**
1514  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1515  *
1516  */
1517
1518 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1519
1520 /**
1521  * @todo document
1522  * @private
1523  * @package MediaWiki
1524  * @subpackage DifferenceEngine
1525  */
1526 class _HWLDF_WordAccumulator {
1527         function _HWLDF_WordAccumulator () {
1528                 $this->_lines = array();
1529                 $this->_line = '';
1530                 $this->_group = '';
1531                 $this->_tag = '';
1532         }
1533
1534         function _flushGroup ($new_tag) {
1535                 if ($this->_group !== '') {
1536                         if ($this->_tag == 'mark')
1537                                 $this->_line .= '<span class="diffchange">' .
1538                                         htmlspecialchars ( $this->_group ) . '</span>';
1539                         else
1540                                 $this->_line .= htmlspecialchars ( $this->_group );
1541                 }
1542                 $this->_group = '';
1543                 $this->_tag = $new_tag;
1544         }
1545
1546         function _flushLine ($new_tag) {
1547                 $this->_flushGroup($new_tag);
1548                 if ($this->_line != '')
1549                         array_push ( $this->_lines, $this->_line );
1550                 else
1551                         # make empty lines visible by inserting an NBSP
1552                         array_push ( $this->_lines, NBSP );
1553                 $this->_line = '';
1554         }
1555
1556         function addWords ($words, $tag = '') {
1557                 if ($tag != $this->_tag)
1558                         $this->_flushGroup($tag);
1559
1560                 foreach ($words as $word) {
1561                         // new-line should only come as first char of word.
1562                         if ($word == '')
1563                                 continue;
1564                         if ($word[0] == "\n") {
1565                                 $this->_flushLine($tag);
1566                                 $word = substr($word, 1);
1567                         }
1568                         assert(!strstr($word, "\n"));
1569                         $this->_group .= $word;
1570                 }
1571         }
1572
1573         function getLines() {
1574                 $this->_flushLine('~done');
1575                 return $this->_lines;
1576         }
1577 }
1578
1579 /**
1580  * @todo document
1581  * @private
1582  * @package MediaWiki
1583  * @subpackage DifferenceEngine
1584  */
1585 class WordLevelDiff extends MappedDiff
1586 {
1587         const MAX_LINE_LENGTH = 10000;
1588
1589         function WordLevelDiff ($orig_lines, $closing_lines) {
1590                 $fname = 'WordLevelDiff::WordLevelDiff';
1591                 wfProfileIn( $fname );
1592
1593                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1594                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1595
1596                 $this->MappedDiff($orig_words, $closing_words,
1597                                                   $orig_stripped, $closing_stripped);
1598                 wfProfileOut( $fname );
1599         }
1600
1601         function _split($lines) {
1602                 $fname = 'WordLevelDiff::_split';
1603                 wfProfileIn( $fname );
1604
1605                 $words = array();
1606                 $stripped = array();
1607                 $first = true;
1608                 foreach ( $lines as $line ) {
1609                         # If the line is too long, just pretend the entire line is one big word
1610                         # This prevents resource exhaustion problems
1611                         if ( $first ) {
1612                                 $first = false;
1613                         } else {
1614                                 $words[] = "\n";
1615                                 $stripped[] = "\n";
1616                         }
1617                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1618                                 $words[] = $line;
1619                                 $stripped[] = $line;
1620                         } else {
1621                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1622                                         $line, $m))
1623                                 {
1624                                         $words = array_merge( $words, $m[0] );
1625                                         $stripped = array_merge( $stripped, $m[1] );
1626                                 }
1627                         }
1628                 }
1629                 wfProfileOut( $fname );
1630                 return array($words, $stripped);
1631         }
1632
1633         function orig () {
1634                 $fname = 'WordLevelDiff::orig';
1635                 wfProfileIn( $fname );
1636                 $orig = new _HWLDF_WordAccumulator;
1637
1638                 foreach ($this->edits as $edit) {
1639                         if ($edit->type == 'copy')
1640                                 $orig->addWords($edit->orig);
1641                         elseif ($edit->orig)
1642                                 $orig->addWords($edit->orig, 'mark');
1643                 }
1644                 $lines = $orig->getLines();
1645                 wfProfileOut( $fname );
1646                 return $lines;
1647         }
1648
1649         function closing () {
1650                 $fname = 'WordLevelDiff::closing';
1651                 wfProfileIn( $fname );
1652                 $closing = new _HWLDF_WordAccumulator;
1653
1654                 foreach ($this->edits as $edit) {
1655                         if ($edit->type == 'copy')
1656                                 $closing->addWords($edit->closing);
1657                         elseif ($edit->closing)
1658                                 $closing->addWords($edit->closing, 'mark');
1659                 }
1660                 $lines = $closing->getLines();
1661                 wfProfileOut( $fname );
1662                 return $lines;
1663         }
1664 }
1665
1666 /**
1667  *      Wikipedia Table style diff formatter.
1668  * @todo document
1669  * @private
1670  * @package MediaWiki
1671  * @subpackage DifferenceEngine
1672  */
1673 class TableDiffFormatter extends DiffFormatter
1674 {
1675         function TableDiffFormatter() {
1676                 $this->leading_context_lines = 2;
1677                 $this->trailing_context_lines = 2;
1678         }
1679
1680         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1681                 $r = '<tr><td colspan="2" align="left"><strong><!--LINE '.$xbeg."--></strong></td>\n" .
1682                   '<td colspan="2" align="left"><strong><!--LINE '.$ybeg."--></strong></td></tr>\n";
1683                 return $r;
1684         }
1685
1686         function _start_block( $header ) {
1687                 echo $header;
1688         }
1689
1690         function _end_block() {
1691         }
1692
1693         function _lines( $lines, $prefix=' ', $color='white' ) {
1694         }
1695
1696         # HTML-escape parameter before calling this
1697         function addedLine( $line ) {
1698                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1699         }
1700
1701         # HTML-escape parameter before calling this
1702         function deletedLine( $line ) {
1703                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1704         }
1705
1706         # HTML-escape parameter before calling this
1707         function contextLine( $line ) {
1708                 return "<td> </td><td class='diff-context'>{$line}</td>";
1709         }
1710
1711         function emptyLine() {
1712                 return '<td colspan="2">&nbsp;</td>';
1713         }
1714
1715         function _added( $lines ) {
1716                 foreach ($lines as $line) {
1717                         echo '<tr>' . $this->emptyLine() .
1718                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1719                 }
1720         }
1721
1722         function _deleted($lines) {
1723                 foreach ($lines as $line) {
1724                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1725                           $this->emptyLine() . "</tr>\n";
1726                 }
1727         }
1728
1729         function _context( $lines ) {
1730                 foreach ($lines as $line) {
1731                         echo '<tr>' .
1732                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1733                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1734                 }
1735         }
1736
1737         function _changed( $orig, $closing ) {
1738                 $fname = 'TableDiffFormatter::_changed';
1739                 wfProfileIn( $fname );
1740
1741                 $diff = new WordLevelDiff( $orig, $closing );
1742                 $del = $diff->orig();
1743                 $add = $diff->closing();
1744
1745                 # Notice that WordLevelDiff returns HTML-escaped output.
1746                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1747
1748                 while ( $line = array_shift( $del ) ) {
1749                         $aline = array_shift( $add );
1750                         echo '<tr>' . $this->deletedLine( $line ) .
1751                                 $this->addedLine( $aline ) . "</tr>\n";
1752                 }
1753                 foreach ($add as $line) {       # If any leftovers
1754                         echo '<tr>' . $this->emptyLine() .
1755                                 $this->addedLine( $line ) . "</tr>\n";
1756                 }
1757                 wfProfileOut( $fname );
1758         }
1759 }
1760
1761 ?>