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