includes/DifferenceEngine.php

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