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