includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @todo indicate where diff.doc can be found.
   5  * @addtogroup DifferenceEngine
   6  */
   7
   8 /**
   9  * @todo document
  10  * @public
  11  * @addtogroup DifferenceEngine
  12  */
  13 class DifferenceEngine {
  14         /**#@+
  15          * @private
  16          */
  17         var $mOldid, $mNewid, $mTitle;
  18         var $mOldtitle, $mNewtitle, $mPagetitle;
  19         var $mOldtext, $mNewtext;
  20         var $mOldPage, $mNewPage;
  21         var $mRcidMarkPatrolled;
  22         var $mOldRev, $mNewRev;
  23         var $mRevisionsLoaded = false; // Have the revisions been loaded
  24         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  25         /**#@-*/
  26
  27         /**
  28          * Constructor
  29          * @param $titleObj Title object that the diff is associated with
  30          * @param $old Integer: old ID we want to show and diff with.
  31          * @param $new String: either 'prev' or 'next'.
  32          * @param $rcid Integer: ??? FIXME (default 0)
  33          */
  34         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0 ) {
  35                 $this->mTitle = $titleObj;
  36                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  37
  38                 if ( 'prev' === $new ) {
  39                         # Show diff between revision $old and the previous one.
  40                         # Get previous one from DB.
  41                         #
  42                         $this->mNewid = intval($old);
  43
  44                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  45
  46                 } elseif ( 'next' === $new ) {
  47                         # Show diff between revision $old and the previous one.
  48                         # Get previous one from DB.
  49                         #
  50                         $this->mOldid = intval($old);
  51                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  52                         if ( false === $this->mNewid ) {
  53                                 # if no result, NewId points to the newest old revision. The only newer
  54                                 # revision is cur, which is "0".
  55                                 $this->mNewid = 0;
  56                         }
  57
  58                 } else {
  59                         $this->mOldid = intval($old);
  60                         $this->mNewid = intval($new);
  61                 }
  62                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  63         }
  64
  65         function showDiffPage( $diffOnly = false ) {
  66                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  67                 $fname = 'DifferenceEngine::showDiffPage';
  68                 wfProfileIn( $fname );
  69
  70                 # If external diffs are enabled both globally and for the user,
  71                 # we'll use the application/x-external-editor interface to call
  72                 # an external diff tool like kompare, kdiff3, etc.
  73                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  74                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  75                         $wgOut->disable();
  76                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  77                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  78                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  79                         $special=$wgLang->getNsText(NS_SPECIAL);
  80                         $control=<<<CONTROL
  81 [Process]
  82 Type=Diff text
  83 Engine=MediaWiki
  84 Script={$wgServer}{$wgScript}
  85 Special namespace={$special}
  86
  87 [File]
  88 Extension=wiki
  89 URL=$url1
  90
  91 [File 2]
  92 Extension=wiki
  93 URL=$url2
  94 CONTROL;
  95                         echo($control);
  96                         return;
  97                 }
  98
  99                 $wgOut->addStyle( 'common/diff.css' );
 100                 $wgOut->setArticleFlag( false );
 101                 if ( ! $this->loadRevisionData() ) {
 102                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, {$this->mNewid})";
 103                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 104                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 105                         $wgOut->addWikitext( $mtext );
 106                         wfProfileOut( $fname );
 107                         return;
 108                 }
 109
 110                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 111
 112                 if ( $this->mNewRev->isCurrent() ) {
 113                         $wgOut->setArticleFlag( true );
 114                 }
 115
 116                 # mOldid is false if the difference engine is called with a "vague" query for
 117                 # a diff between a version V and its previous version V' AND the version V
 118                 # is the first version of that article. In that case, V' does not exist.
 119                 if ( $this->mOldid === false ) {
 120                         $this->showFirstRevision();
 121                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 122                         wfProfileOut( $fname );
 123                         return;
 124                 }
 125
 126                 $wgOut->suppressQuickbar();
 127
 128                 $oldTitle = $this->mOldPage->getPrefixedText();
 129                 $newTitle = $this->mNewPage->getPrefixedText();
 130                 if( $oldTitle == $newTitle ) {
 131                         $wgOut->setPageTitle( $newTitle );
 132                 } else {
 133                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 134                 }
 135                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 136                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 137
 138                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 139                         $wgOut->loginToUse();
 140                         $wgOut->output();
 141                         wfProfileOut( $fname );
 142                         exit;
 143                 }
 144
 145                 $sk = $wgUser->getSkin();
 146
 147                 if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 148                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 149                 } else {
 150                         $rollback = '';
 151                 }
 152                 if( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isAllowed( 'patrol' ) ) {
 153                         $patrol = ' [' . $sk->makeKnownLinkObj( $this->mTitle, wfMsg( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 154                 } else {
 155                         $patrol = '';
 156                 }
 157
 158                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 159                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 160                 if ( $this->mNewRev->isCurrent() ) {
 161                         $nextlink = '&nbsp;';
 162                 } else {
 163                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 164                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 165                 }
 166
 167                 $oldminor = '';
 168                 $newminor = '';
 169
 170                 if ($this->mOldRev->mMinorEdit == 1) {
 171                         $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
 172                                 wfMsg( 'minoreditletter') ) . ' ';
 173                 }
 174
 175                 if ($this->mNewRev->mMinorEdit == 1) {
 176                         $newminor = wfElement( 'span', array( 'class' => 'minor' ),
 177                         wfMsg( 'minoreditletter') ) . ' ';
 178                 }
 179
 180                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />" .
 181                         $sk->revUserTools( $this->mOldRev ) . "<br />" .
 182                         $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly ) . "<br />" .
 183                         $prevlink;
 184                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />" .
 185                         $sk->revUserTools( $this->mNewRev ) . " $rollback<br />" .
 186                         $newminor . $sk->revComment( $this->mNewRev, !$diffOnly ) . "<br />" .
 187                         $nextlink . $patrol;
 188
 189                 $this->showDiff( $oldHeader, $newHeader );
 190
 191                 if ( !$diffOnly )
 192                         $this->renderNewRevision();
 193
 194                 wfProfileOut( $fname );
 195         }
 196
 197         /**
 198          * Show the new revision of the page.
 199          */
 200         function renderNewRevision() {
 201                 global $wgOut;
 202                 $fname = 'DifferenceEngine::renderNewRevision';
 203                 wfProfileIn( $fname );
 204
 205                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 206                 #add deleted rev tag if needed
 207                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 208                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-permission' ) );
 209                 }
 210
 211                 if( !$this->mNewRev->isCurrent() ) {
 212                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 213                 }
 214
 215                 $this->loadNewText();
 216                 if( is_object( $this->mNewRev ) ) {
 217                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 218                 }
 219
 220                 $wgOut->addWikiTextTidy( $this->mNewtext );
 221
 222                 if( !$this->mNewRev->isCurrent() ) {
 223                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 224                 }
 225
 226                 wfProfileOut( $fname );
 227         }
 228
 229         /**
 230          * Show the first revision of an article. Uses normal diff headers in
 231          * contrast to normal "old revision" display style.
 232          */
 233         function showFirstRevision() {
 234                 global $wgOut, $wgUser;
 235
 236                 $fname = 'DifferenceEngine::showFirstRevision';
 237                 wfProfileIn( $fname );
 238
 239                 # Get article text from the DB
 240                 #
 241                 if ( ! $this->loadNewText() ) {
 242                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 243                           "{$this->mNewid})";
 244                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 245                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 246                         $wgOut->addWikitext( $mtext );
 247                         wfProfileOut( $fname );
 248                         return;
 249                 }
 250                 if ( $this->mNewRev->isCurrent() ) {
 251                         $wgOut->setArticleFlag( true );
 252                 }
 253
 254                 # Check if user is allowed to look at this page. If not, bail out.
 255                 #
 256                 if ( !( $this->mTitle->userCanRead() ) ) {
 257                         $wgOut->loginToUse();
 258                         $wgOut->output();
 259                         wfProfileOut( $fname );
 260                         exit;
 261                 }
 262
 263                 # Prepare the header box
 264                 #
 265                 $sk = $wgUser->getSkin();
 266
 267                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 268                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 269                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 270                         $sk->revComment( $this->mNewRev ) . "<br />" .
 271                         $nextlink . "</div>\n";
 272
 273                 $wgOut->addHTML( $header );
 274
 275                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 276                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 277
 278                 wfProfileOut( $fname );
 279         }
 280
 281         /**
 282          * Get the diff text, send it to $wgOut
 283          * Returns false if the diff could not be generated, otherwise returns true
 284          */
 285         function showDiff( $otitle, $ntitle ) {
 286                 global $wgOut;
 287                 $diff = $this->getDiff( $otitle, $ntitle );
 288                 if ( $diff === false ) {
 289                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
 290                         return false;
 291                 } else {
 292                         $wgOut->addHTML( $diff );
 293                         return true;
 294                 }
 295         }
 296
 297         /**
 298          * Get diff table, including header
 299          * Note that the interface has changed, it's no longer static.
 300          * Returns false on error
 301          */
 302         function getDiff( $otitle, $ntitle ) {
 303                 $body = $this->getDiffBody();
 304                 if ( $body === false ) {
 305                         return false;
 306                 } else {
 307                         $multi = $this->getMultiNotice();
 308                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 309                 }
 310         }
 311
 312         /**
 313          * Get the diff table body, without header
 314          * Results are cached
 315          * Returns false on error
 316          */
 317         function getDiffBody() {
 318                 global $wgMemc;
 319                 $fname = 'DifferenceEngine::getDiffBody';
 320                 wfProfileIn( $fname );
 321
 322                 // Cacheable?
 323                 $key = false;
 324                 if ( $this->mOldid && $this->mNewid ) {
 325                         // Try cache
 326                         $key = wfMemcKey( 'diff', 'oldid', $this->mOldid, 'newid', $this->mNewid );
 327                         $difftext = $wgMemc->get( $key );
 328                         if ( $difftext ) {
 329                                 wfIncrStats( 'diff_cache_hit' );
 330                                 $difftext = $this->localiseLineNumbers( $difftext );
 331                                 $difftext .= "\n<!-- diff cache key $key -->\n";
 332                                 wfProfileOut( $fname );
 333                                 return $difftext;
 334                         }
 335                 }
 336
 337                 #loadtext is permission safe, this just clears out the diff
 338                 if ( !$this->loadText() ) {
 339                         wfProfileOut( $fname );
 340                         return false;
 341                 } else if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 342                   return '';
 343                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 344                   return '';
 345                 }
 346
 347                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 348
 349                 // Save to cache for 7 days
 350                 if ( $key !== false && $difftext !== false ) {
 351                         wfIncrStats( 'diff_cache_miss' );
 352                         $wgMemc->set( $key, $difftext, 7*86400 );
 353                 } else {
 354                         wfIncrStats( 'diff_uncacheable' );
 355                 }
 356                 // Replace line numbers with the text in the user's language
 357                 if ( $difftext !== false ) {
 358                         $difftext = $this->localiseLineNumbers( $difftext );
 359                 }
 360                 wfProfileOut( $fname );
 361                 return $difftext;
 362         }
 363
 364         /**
 365          * Generate a diff, no caching
 366          * $otext and $ntext must be already segmented
 367          */
 368         function generateDiffBody( $otext, $ntext ) {
 369                 global $wgExternalDiffEngine, $wgContLang;
 370                 $fname = 'DifferenceEngine::generateDiffBody';
 371
 372                 $otext = str_replace( "\r\n", "\n", $otext );
 373                 $ntext = str_replace( "\r\n", "\n", $ntext );
 374
 375                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 376                         # For historical reasons, external diff engine expects
 377                         # input text to be HTML-escaped already
 378                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 379                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 380                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 381                                 dl('php_wikidiff.so');
 382                         }
 383                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 384                 }
 385
 386                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 387                         # Better external diff engine, the 2 may some day be dropped
 388                         # This one does the escaping and segmenting itself
 389                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 390                                 wfProfileIn( "$fname-dl" );
 391                                 @dl('php_wikidiff2.so');
 392                                 wfProfileOut( "$fname-dl" );
 393                         }
 394                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 395                                 wfProfileIn( 'wikidiff2_do_diff' );
 396                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 397                                 wfProfileOut( 'wikidiff2_do_diff' );
 398                                 return $text;
 399                         }
 400                 }
 401                 if ( $wgExternalDiffEngine !== false ) {
 402                         # Diff via the shell
 403                         global $wgTmpDirectory;
 404                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 405                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 406
 407                         $tempFile1 = fopen( $tempName1, "w" );
 408                         if ( !$tempFile1 ) {
 409                                 wfProfileOut( $fname );
 410                                 return false;
 411                         }
 412                         $tempFile2 = fopen( $tempName2, "w" );
 413                         if ( !$tempFile2 ) {
 414                                 wfProfileOut( $fname );
 415                                 return false;
 416                         }
 417                         fwrite( $tempFile1, $otext );
 418                         fwrite( $tempFile2, $ntext );
 419                         fclose( $tempFile1 );
 420                         fclose( $tempFile2 );
 421                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 422                         wfProfileIn( "$fname-shellexec" );
 423                         $difftext = wfShellExec( $cmd );
 424                         wfProfileOut( "$fname-shellexec" );
 425                         unlink( $tempName1 );
 426                         unlink( $tempName2 );
 427                         return $difftext;
 428                 }
 429
 430                 # Native PHP diff
 431                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 432                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 433                 $diffs = new Diff( $ota, $nta );
 434                 $formatter = new TableDiffFormatter();
 435                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 436         }
 437
 438
 439         /**
 440          * Replace line numbers with the text in the user's language
 441          */
 442         function localiseLineNumbers( $text ) {
 443                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 444                         array( &$this, 'localiseLineNumbersCb' ), $text );
 445         }
 446
 447         function localiseLineNumbersCb( $matches ) {
 448                 global $wgLang;
 449                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 450         }
 451
 452
 453         /**
 454          * If there are revisions between the ones being compared, return a note saying so.
 455          */
 456         function getMultiNotice() {
 457                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 458                         return '';
 459
 460                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 461                         // Comparing two different pages? Count would be meaningless.
 462                         return '';
 463                 }
 464
 465                 $oldid = $this->mOldRev->getId();
 466                 $newid = $this->mNewRev->getId();
 467                 if ( $oldid > $newid ) {
 468                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 469                 }
 470
 471                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 472                 if ( !$n )
 473                         return '';
 474
 475                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 476         }
 477
 478
 479         /**
 480          * Add the header to a diff body
 481          */
 482         function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 483                 global $wgOut;
 484
 485                 if ( $this->mOldRev && $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 486                    $otitle = '<span class="history-deleted">'.$otitle.'</span>';
 487                 }
 488                 if ( $this->mNewRev && $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 489                    $ntitle = '<span class="history-deleted">'.$ntitle.'</span>';
 490                 }
 491                 $header = "
 492                         <table class='diff'>
 493                         <col class='diff-marker' />
 494                         <col class='diff-content' />
 495                         <col class='diff-marker' />
 496                         <col class='diff-content' />
 497                         <tr>
 498                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 499                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 500                         </tr>
 501                 ";
 502
 503                 if ( $multi != '' )
 504                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 505
 506                 return $header . $diff . "</table>";
 507         }
 508
 509         /**
 510          * Use specified text instead of loading from the database
 511          */
 512         function setText( $oldText, $newText ) {
 513                 $this->mOldtext = $oldText;
 514                 $this->mNewtext = $newText;
 515                 $this->mTextLoaded = 2;
 516         }
 517
 518         /**
 519          * Load revision metadata for the specified articles. If newid is 0, then compare
 520          * the old article in oldid to the current article; if oldid is 0, then
 521          * compare the current article to the immediately previous one (ignoring the
 522          * value of newid).
 523          *
 524          * If oldid is false, leave the corresponding revision object set
 525          * to false. This is impossible via ordinary user input, and is provided for
 526          * API convenience.
 527          */
 528         function loadRevisionData() {
 529                 global $wgLang;
 530                 if ( $this->mRevisionsLoaded ) {
 531                         return true;
 532                 } else {
 533                         // Whether it succeeds or fails, we don't want to try again
 534                         $this->mRevisionsLoaded = true;
 535                 }
 536
 537                 // Load the new revision object
 538                 if( $this->mNewid ) {
 539                         $this->mNewRev = Revision::newFromId( $this->mNewid );
 540                 } else {
 541                         $this->mNewRev = Revision::newFromTitle( $this->mTitle );
 542                 }
 543
 544                 if( is_null( $this->mNewRev ) ) {
 545                         return false;
 546                 }
 547
 548                 // Set assorted variables
 549                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 550                 $this->mNewPage = $this->mNewRev->getTitle();
 551                 if( $this->mNewRev->isCurrent() ) {
 552                         $newLink = $this->mNewPage->escapeLocalUrl();
 553                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 554                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 555
 556                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
 557                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 558
 559                 } else {
 560                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 561                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 562                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 563
 564                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
 565                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 566                 }
 567
 568                 // Load the old revision object
 569                 $this->mOldRev = false;
 570                 if( $this->mOldid ) {
 571                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 572                 } elseif ( $this->mOldid === 0 ) {
 573                         $rev = $this->mNewRev->getPrevious();
 574                         if( $rev ) {
 575                                 $this->mOldid = $rev->getId();
 576                                 $this->mOldRev = $rev;
 577                         } else {
 578                                 // No previous revision; mark to show as first-version only.
 579                                 $this->mOldid = false;
 580                                 $this->mOldRev = false;
 581                         }
 582                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 583
 584                 if( is_null( $this->mOldRev ) ) {
 585                         return false;
 586                 }
 587
 588                 if ( $this->mOldRev ) {
 589                         $this->mOldPage = $this->mOldRev->getTitle();
 590
 591                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 592                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 593                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 594                         $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) )
 595                                 . "</a> (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 596                         //now that we considered old rev, we can make undo link (bug 8133, multi-edit undo)
 597                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 598                         $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 599                 }
 600
 601                 return true;
 602         }
 603
 604         /**
 605          * Load the text of the revisions, as well as revision data.
 606          */
 607         function loadText() {
 608                 if ( $this->mTextLoaded == 2 ) {
 609                         return true;
 610                 } else {
 611                         // Whether it succeeds or fails, we don't want to try again
 612                         $this->mTextLoaded = 2;
 613                 }
 614
 615                 if ( !$this->loadRevisionData() ) {
 616                         return false;
 617                 }
 618                 if ( $this->mOldRev ) {
 619                         // FIXME: permission tests
 620                         $this->mOldtext = $this->mOldRev->revText();
 621                         if ( $this->mOldtext === false ) {
 622                                 return false;
 623                         }
 624                 }
 625                 if ( $this->mNewRev ) {
 626                         $this->mNewtext = $this->mNewRev->revText();
 627                         if ( $this->mNewtext === false ) {
 628                                 return false;
 629                         }
 630                 }
 631                 return true;
 632         }
 633
 634         /**
 635          * Load the text of the new revision, not the old one
 636          */
 637         function loadNewText() {
 638                 if ( $this->mTextLoaded >= 1 ) {
 639                         return true;
 640                 } else {
 641                         $this->mTextLoaded = 1;
 642                 }
 643                 if ( !$this->loadRevisionData() ) {
 644                         return false;
 645                 }
 646                 $this->mNewtext = $this->mNewRev->getText();
 647                 return true;
 648         }
 649
 650
 651 }
 652
 653 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 654 //
 655 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 656 // You may copy this code freely under the conditions of the GPL.
 657 //
 658
 659 define('USE_ASSERTS', function_exists('assert'));
 660
 661 /**
 662  * @todo document
 663  * @private
 664  * @addtogroup DifferenceEngine
 665  */
 666 class _DiffOp {
 667         var $type;
 668         var $orig;
 669         var $closing;
 670
 671         function reverse() {
 672                 trigger_error('pure virtual', E_USER_ERROR);
 673         }
 674
 675         function norig() {
 676                 return $this->orig ? sizeof($this->orig) : 0;
 677         }
 678
 679         function nclosing() {
 680                 return $this->closing ? sizeof($this->closing) : 0;
 681         }
 682 }
 683
 684 /**
 685  * @todo document
 686  * @private
 687  * @addtogroup DifferenceEngine
 688  */
 689 class _DiffOp_Copy extends _DiffOp {
 690         var $type = 'copy';
 691
 692         function _DiffOp_Copy ($orig, $closing = false) {
 693                 if (!is_array($closing))
 694                         $closing = $orig;
 695                 $this->orig = $orig;
 696                 $this->closing = $closing;
 697         }
 698
 699         function reverse() {
 700                 return new _DiffOp_Copy($this->closing, $this->orig);
 701         }
 702 }
 703
 704 /**
 705  * @todo document
 706  * @private
 707  * @addtogroup DifferenceEngine
 708  */
 709 class _DiffOp_Delete extends _DiffOp {
 710         var $type = 'delete';
 711
 712         function _DiffOp_Delete ($lines) {
 713                 $this->orig = $lines;
 714                 $this->closing = false;
 715         }
 716
 717         function reverse() {
 718                 return new _DiffOp_Add($this->orig);
 719         }
 720 }
 721
 722 /**
 723  * @todo document
 724  * @private
 725  * @addtogroup DifferenceEngine
 726  */
 727 class _DiffOp_Add extends _DiffOp {
 728         var $type = 'add';
 729
 730         function _DiffOp_Add ($lines) {
 731                 $this->closing = $lines;
 732                 $this->orig = false;
 733         }
 734
 735         function reverse() {
 736                 return new _DiffOp_Delete($this->closing);
 737         }
 738 }
 739
 740 /**
 741  * @todo document
 742  * @private
 743  * @addtogroup DifferenceEngine
 744  */
 745 class _DiffOp_Change extends _DiffOp {
 746         var $type = 'change';
 747
 748         function _DiffOp_Change ($orig, $closing) {
 749                 $this->orig = $orig;
 750                 $this->closing = $closing;
 751         }
 752
 753         function reverse() {
 754                 return new _DiffOp_Change($this->closing, $this->orig);
 755         }
 756 }
 757
 758
 759 /**
 760  * Class used internally by Diff to actually compute the diffs.
 761  *
 762  * The algorithm used here is mostly lifted from the perl module
 763  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 764  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 765  *
 766  * More ideas are taken from:
 767  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 768  *
 769  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 770  * diffutils-2.7, which can be found at:
 771  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 772  *
 773  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 774  * are my own.
 775  *
 776  * Line length limits for robustness added by Tim Starling, 2005-08-31
 777  *
 778  * @author Geoffrey T. Dairiki, Tim Starling
 779  * @private
 780  * @addtogroup DifferenceEngine
 781  */
 782 class _DiffEngine
 783 {
 784         const MAX_XREF_LENGTH =  10000;
 785
 786         function diff ($from_lines, $to_lines) {
 787                 $fname = '_DiffEngine::diff';
 788                 wfProfileIn( $fname );
 789
 790                 $n_from = sizeof($from_lines);
 791                 $n_to = sizeof($to_lines);
 792
 793                 $this->xchanged = $this->ychanged = array();
 794                 $this->xv = $this->yv = array();
 795                 $this->xind = $this->yind = array();
 796                 unset($this->seq);
 797                 unset($this->in_seq);
 798                 unset($this->lcs);
 799
 800                 // Skip leading common lines.
 801                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 802                         if ($from_lines[$skip] !== $to_lines[$skip])
 803                                 break;
 804                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 805                 }
 806                 // Skip trailing common lines.
 807                 $xi = $n_from; $yi = $n_to;
 808                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 809                         if ($from_lines[$xi] !== $to_lines[$yi])
 810                                 break;
 811                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 812                 }
 813
 814                 // Ignore lines which do not exist in both files.
 815                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 816                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 817                 }
 818
 819                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 820                         $line = $to_lines[$yi];
 821                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 822                                 continue;
 823                         $yhash[$this->_line_hash($line)] = 1;
 824                         $this->yv[] = $line;
 825                         $this->yind[] = $yi;
 826                 }
 827                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 828                         $line = $from_lines[$xi];
 829                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 830                                 continue;
 831                         $this->xv[] = $line;
 832                         $this->xind[] = $xi;
 833                 }
 834
 835                 // Find the LCS.
 836                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 837
 838                 // Merge edits when possible
 839                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 840                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 841
 842                 // Compute the edit operations.
 843                 $edits = array();
 844                 $xi = $yi = 0;
 845                 while ($xi < $n_from || $yi < $n_to) {
 846                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 847                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 848
 849                         // Skip matching "snake".
 850                         $copy = array();
 851                         while ( $xi < $n_from && $yi < $n_to
 852                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 853                                 $copy[] = $from_lines[$xi++];
 854                                 ++$yi;
 855                         }
 856                         if ($copy)
 857                                 $edits[] = new _DiffOp_Copy($copy);
 858
 859                         // Find deletes & adds.
 860                         $delete = array();
 861                         while ($xi < $n_from && $this->xchanged[$xi])
 862                                 $delete[] = $from_lines[$xi++];
 863
 864                         $add = array();
 865                         while ($yi < $n_to && $this->ychanged[$yi])
 866                                 $add[] = $to_lines[$yi++];
 867
 868                         if ($delete && $add)
 869                                 $edits[] = new _DiffOp_Change($delete, $add);
 870                         elseif ($delete)
 871                                 $edits[] = new _DiffOp_Delete($delete);
 872                         elseif ($add)
 873                                 $edits[] = new _DiffOp_Add($add);
 874                 }
 875                 wfProfileOut( $fname );
 876                 return $edits;
 877         }
 878
 879         /**
 880          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 881          */
 882         function _line_hash( $line ) {
 883                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 884                         return md5( $line );
 885                 } else {
 886                         return $line;
 887                 }
 888         }
 889
 890
 891         /* Divide the Largest Common Subsequence (LCS) of the sequences
 892          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 893          * sized segments.
 894          *
 895          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 896          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 897          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 898          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 899          * that (X0, Y0) == (XOFF, YOFF) and
 900          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 901          *
 902          * This function assumes that the first lines of the specified portions
 903          * of the two files do not match, and likewise that the last lines do not
 904          * match.  The caller must trim matching lines from the beginning and end
 905          * of the portions it is going to specify.
 906          */
 907         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 908                 $fname = '_DiffEngine::_diag';
 909                 wfProfileIn( $fname );
 910                 $flip = false;
 911
 912                 if ($xlim - $xoff > $ylim - $yoff) {
 913                         // Things seems faster (I'm not sure I understand why)
 914                                 // when the shortest sequence in X.
 915                                 $flip = true;
 916                         list ($xoff, $xlim, $yoff, $ylim)
 917                         = array( $yoff, $ylim, $xoff, $xlim);
 918                 }
 919
 920                 if ($flip)
 921                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 922                                 $ymatches[$this->xv[$i]][] = $i;
 923                 else
 924                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 925                                 $ymatches[$this->yv[$i]][] = $i;
 926
 927                 $this->lcs = 0;
 928                 $this->seq[0]= $yoff - 1;
 929                 $this->in_seq = array();
 930                 $ymids[0] = array();
 931
 932                 $numer = $xlim - $xoff + $nchunks - 1;
 933                 $x = $xoff;
 934                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 935                         wfProfileIn( "$fname-chunk" );
 936                         if ($chunk > 0)
 937                                 for ($i = 0; $i <= $this->lcs; $i++)
 938                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 939
 940                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 941                         for ( ; $x < $x1; $x++) {
 942                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 943                                         if (empty($ymatches[$line]))
 944                                                 continue;
 945                                 $matches = $ymatches[$line];
 946                                 reset($matches);
 947                                 while (list ($junk, $y) = each($matches))
 948                                         if (empty($this->in_seq[$y])) {
 949                                                 $k = $this->_lcs_pos($y);
 950                                                 USE_ASSERTS && assert($k > 0);
 951                                                 $ymids[$k] = $ymids[$k-1];
 952                                                 break;
 953                                         }
 954                                 while (list ( /* $junk */, $y) = each($matches)) {
 955                                         if ($y > $this->seq[$k-1]) {
 956                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 957                                                 // Optimization: this is a common case:
 958                                                 //      next match is just replacing previous match.
 959                                                 $this->in_seq[$this->seq[$k]] = false;
 960                                                 $this->seq[$k] = $y;
 961                                                 $this->in_seq[$y] = 1;
 962                                         } else if (empty($this->in_seq[$y])) {
 963                                                 $k = $this->_lcs_pos($y);
 964                                                 USE_ASSERTS && assert($k > 0);
 965                                                 $ymids[$k] = $ymids[$k-1];
 966                                         }
 967                                 }
 968                         }
 969                         wfProfileOut( "$fname-chunk" );
 970                 }
 971
 972                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 973                 $ymid = $ymids[$this->lcs];
 974                 for ($n = 0; $n < $nchunks - 1; $n++) {
 975                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 976                         $y1 = $ymid[$n] + 1;
 977                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 978                 }
 979                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 980
 981                 wfProfileOut( $fname );
 982                 return array($this->lcs, $seps);
 983         }
 984
 985         function _lcs_pos ($ypos) {
 986                 $fname = '_DiffEngine::_lcs_pos';
 987                 wfProfileIn( $fname );
 988
 989                 $end = $this->lcs;
 990                 if ($end == 0 || $ypos > $this->seq[$end]) {
 991                         $this->seq[++$this->lcs] = $ypos;
 992                         $this->in_seq[$ypos] = 1;
 993                         wfProfileOut( $fname );
 994                         return $this->lcs;
 995                 }
 996
 997                 $beg = 1;
 998                 while ($beg < $end) {
 999                         $mid = (int)(($beg + $end) / 2);
1000                         if ( $ypos > $this->seq[$mid] )
1001                                 $beg = $mid + 1;
1002                         else
1003                                 $end = $mid;
1004                 }
1005
1006                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1007
1008                 $this->in_seq[$this->seq[$end]] = false;
1009                 $this->seq[$end] = $ypos;
1010                 $this->in_seq[$ypos] = 1;
1011                 wfProfileOut( $fname );
1012                 return $end;
1013         }
1014
1015         /* Find LCS of two sequences.
1016          *
1017          * The results are recorded in the vectors $this->{x,y}changed[], by
1018          * storing a 1 in the element for each line that is an insertion
1019          * or deletion (ie. is not in the LCS).
1020          *
1021          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1022          *
1023          * Note that XLIM, YLIM are exclusive bounds.
1024          * All line numbers are origin-0 and discarded lines are not counted.
1025          */
1026         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1027                 $fname = '_DiffEngine::_compareseq';
1028                 wfProfileIn( $fname );
1029
1030                 // Slide down the bottom initial diagonal.
1031                 while ($xoff < $xlim && $yoff < $ylim
1032                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1033                         ++$xoff;
1034                         ++$yoff;
1035                 }
1036
1037                 // Slide up the top initial diagonal.
1038                 while ($xlim > $xoff && $ylim > $yoff
1039                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1040                         --$xlim;
1041                         --$ylim;
1042                 }
1043
1044                 if ($xoff == $xlim || $yoff == $ylim)
1045                         $lcs = 0;
1046                 else {
1047                         // This is ad hoc but seems to work well.
1048                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1049                         //$nchunks = max(2,min(8,(int)$nchunks));
1050                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1051                         list ($lcs, $seps)
1052                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1053                 }
1054
1055                 if ($lcs == 0) {
1056                         // X and Y sequences have no common subsequence:
1057                         // mark all changed.
1058                         while ($yoff < $ylim)
1059                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1060                         while ($xoff < $xlim)
1061                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1062                 } else {
1063                         // Use the partitions to split this problem into subproblems.
1064                         reset($seps);
1065                         $pt1 = $seps[0];
1066                         while ($pt2 = next($seps)) {
1067                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1068                                 $pt1 = $pt2;
1069                         }
1070                 }
1071                 wfProfileOut( $fname );
1072         }
1073
1074         /* Adjust inserts/deletes of identical lines to join changes
1075          * as much as possible.
1076          *
1077          * We do something when a run of changed lines include a
1078          * line at one end and has an excluded, identical line at the other.
1079          * We are free to choose which identical line is included.
1080          * `compareseq' usually chooses the one at the beginning,
1081          * but usually it is cleaner to consider the following identical line
1082          * to be the "change".
1083          *
1084          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1085          */
1086         function _shift_boundaries ($lines, &$changed, $other_changed) {
1087                 $fname = '_DiffEngine::_shift_boundaries';
1088                 wfProfileIn( $fname );
1089                 $i = 0;
1090                 $j = 0;
1091
1092                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1093                 $len = sizeof($lines);
1094                 $other_len = sizeof($other_changed);
1095
1096                 while (1) {
1097                         /*
1098                          * Scan forwards to find beginning of another run of changes.
1099                          * Also keep track of the corresponding point in the other file.
1100                          *
1101                          * Throughout this code, $i and $j are adjusted together so that
1102                          * the first $i elements of $changed and the first $j elements
1103                          * of $other_changed both contain the same number of zeros
1104                          * (unchanged lines).
1105                          * Furthermore, $j is always kept so that $j == $other_len or
1106                          * $other_changed[$j] == false.
1107                          */
1108                         while ($j < $other_len && $other_changed[$j])
1109                                 $j++;
1110
1111                         while ($i < $len && ! $changed[$i]) {
1112                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1113                                 $i++; $j++;
1114                                 while ($j < $other_len && $other_changed[$j])
1115                                         $j++;
1116                         }
1117
1118                         if ($i == $len)
1119                                 break;
1120
1121                         $start = $i;
1122
1123                         // Find the end of this run of changes.
1124                         while (++$i < $len && $changed[$i])
1125                                 continue;
1126
1127                         do {
1128                                 /*
1129                                  * Record the length of this run of changes, so that
1130                                  * we can later determine whether the run has grown.
1131                                  */
1132                                 $runlength = $i - $start;
1133
1134                                 /*
1135                                  * Move the changed region back, so long as the
1136                                  * previous unchanged line matches the last changed one.
1137                                  * This merges with previous changed regions.
1138                                  */
1139                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1140                                         $changed[--$start] = 1;
1141                                         $changed[--$i] = false;
1142                                         while ($start > 0 && $changed[$start - 1])
1143                                                 $start--;
1144                                         USE_ASSERTS && assert('$j > 0');
1145                                         while ($other_changed[--$j])
1146                                                 continue;
1147                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1148                                 }
1149
1150                                 /*
1151                                  * Set CORRESPONDING to the end of the changed run, at the last
1152                                  * point where it corresponds to a changed run in the other file.
1153                                  * CORRESPONDING == LEN means no such point has been found.
1154                                  */
1155                                 $corresponding = $j < $other_len ? $i : $len;
1156
1157                                 /*
1158                                  * Move the changed region forward, so long as the
1159                                  * first changed line matches the following unchanged one.
1160                                  * This merges with following changed regions.
1161                                  * Do this second, so that if there are no merges,
1162                                  * the changed region is moved forward as far as possible.
1163                                  */
1164                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1165                                         $changed[$start++] = false;
1166                                         $changed[$i++] = 1;
1167                                         while ($i < $len && $changed[$i])
1168                                                 $i++;
1169
1170                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1171                                         $j++;
1172                                         if ($j < $other_len && $other_changed[$j]) {
1173                                                 $corresponding = $i;
1174                                                 while ($j < $other_len && $other_changed[$j])
1175                                                         $j++;
1176                                         }
1177                                 }
1178                         } while ($runlength != $i - $start);
1179
1180                         /*
1181                          * If possible, move the fully-merged run of changes
1182                          * back to a corresponding run in the other file.
1183                          */
1184                         while ($corresponding < $i) {
1185                                 $changed[--$start] = 1;
1186                                 $changed[--$i] = 0;
1187                                 USE_ASSERTS && assert('$j > 0');
1188                                 while ($other_changed[--$j])
1189                                         continue;
1190                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1191                         }
1192                 }
1193                 wfProfileOut( $fname );
1194         }
1195 }
1196
1197 /**
1198  * Class representing a 'diff' between two sequences of strings.
1199  * @todo document
1200  * @private
1201  * @addtogroup DifferenceEngine
1202  */
1203 class Diff
1204 {
1205         var $edits;
1206
1207         /**
1208          * Constructor.
1209          * Computes diff between sequences of strings.
1210          *
1211          * @param $from_lines array An array of strings.
1212          *                (Typically these are lines from a file.)
1213          * @param $to_lines array An array of strings.
1214          */
1215         function Diff($from_lines, $to_lines) {
1216                 $eng = new _DiffEngine;
1217                 $this->edits = $eng->diff($from_lines, $to_lines);
1218                 //$this->_check($from_lines, $to_lines);
1219         }
1220
1221         /**
1222          * Compute reversed Diff.
1223          *
1224          * SYNOPSIS:
1225          *
1226          *      $diff = new Diff($lines1, $lines2);
1227          *      $rev = $diff->reverse();
1228          * @return object A Diff object representing the inverse of the
1229          *                                original diff.
1230          */
1231         function reverse () {
1232                 $rev = $this;
1233                 $rev->edits = array();
1234                 foreach ($this->edits as $edit) {
1235                         $rev->edits[] = $edit->reverse();
1236                 }
1237                 return $rev;
1238         }
1239
1240         /**
1241          * Check for empty diff.
1242          *
1243          * @return bool True iff two sequences were identical.
1244          */
1245         function isEmpty () {
1246                 foreach ($this->edits as $edit) {
1247                         if ($edit->type != 'copy')
1248                                 return false;
1249                 }
1250                 return true;
1251         }
1252
1253         /**
1254          * Compute the length of the Longest Common Subsequence (LCS).
1255          *
1256          * This is mostly for diagnostic purposed.
1257          *
1258          * @return int The length of the LCS.
1259          */
1260         function lcs () {
1261                 $lcs = 0;
1262                 foreach ($this->edits as $edit) {
1263                         if ($edit->type == 'copy')
1264                                 $lcs += sizeof($edit->orig);
1265                 }
1266                 return $lcs;
1267         }
1268
1269         /**
1270          * Get the original set of lines.
1271          *
1272          * This reconstructs the $from_lines parameter passed to the
1273          * constructor.
1274          *
1275          * @return array The original sequence of strings.
1276          */
1277         function orig() {
1278                 $lines = array();
1279
1280                 foreach ($this->edits as $edit) {
1281                         if ($edit->orig)
1282                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1283                 }
1284                 return $lines;
1285         }
1286
1287         /**
1288          * Get the closing set of lines.
1289          *
1290          * This reconstructs the $to_lines parameter passed to the
1291          * constructor.
1292          *
1293          * @return array The sequence of strings.
1294          */
1295         function closing() {
1296                 $lines = array();
1297
1298                 foreach ($this->edits as $edit) {
1299                         if ($edit->closing)
1300                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1301                 }
1302                 return $lines;
1303         }
1304
1305         /**
1306          * Check a Diff for validity.
1307          *
1308          * This is here only for debugging purposes.
1309          */
1310         function _check ($from_lines, $to_lines) {
1311                 $fname = 'Diff::_check';
1312                 wfProfileIn( $fname );
1313                 if (serialize($from_lines) != serialize($this->orig()))
1314                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1315                 if (serialize($to_lines) != serialize($this->closing()))
1316                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1317
1318                 $rev = $this->reverse();
1319                 if (serialize($to_lines) != serialize($rev->orig()))
1320                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1321                 if (serialize($from_lines) != serialize($rev->closing()))
1322                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1323
1324
1325                 $prevtype = 'none';
1326                 foreach ($this->edits as $edit) {
1327                         if ( $prevtype == $edit->type )
1328                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1329                         $prevtype = $edit->type;
1330                 }
1331
1332                 $lcs = $this->lcs();
1333                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1334                 wfProfileOut( $fname );
1335         }
1336 }
1337
1338 /**
1339  * @todo document, bad name.
1340  * @private
1341  * @addtogroup DifferenceEngine
1342  */
1343 class MappedDiff extends Diff
1344 {
1345         /**
1346          * Constructor.
1347          *
1348          * Computes diff between sequences of strings.
1349          *
1350          * This can be used to compute things like
1351          * case-insensitve diffs, or diffs which ignore
1352          * changes in white-space.
1353          *
1354          * @param $from_lines array An array of strings.
1355          *      (Typically these are lines from a file.)
1356          *
1357          * @param $to_lines array An array of strings.
1358          *
1359          * @param $mapped_from_lines array This array should
1360          *      have the same size number of elements as $from_lines.
1361          *      The elements in $mapped_from_lines and
1362          *      $mapped_to_lines are what is actually compared
1363          *      when computing the diff.
1364          *
1365          * @param $mapped_to_lines array This array should
1366          *      have the same number of elements as $to_lines.
1367          */
1368         function MappedDiff($from_lines, $to_lines,
1369                                                 $mapped_from_lines, $mapped_to_lines) {
1370                 $fname = 'MappedDiff::MappedDiff';
1371                 wfProfileIn( $fname );
1372
1373                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1374                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1375
1376                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1377
1378                 $xi = $yi = 0;
1379                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1380                         $orig = &$this->edits[$i]->orig;
1381                         if (is_array($orig)) {
1382                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1383                                 $xi += sizeof($orig);
1384                         }
1385
1386                         $closing = &$this->edits[$i]->closing;
1387                         if (is_array($closing)) {
1388                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1389                                 $yi += sizeof($closing);
1390                         }
1391                 }
1392                 wfProfileOut( $fname );
1393         }
1394 }
1395
1396 /**
1397  * A class to format Diffs
1398  *
1399  * This class formats the diff in classic diff format.
1400  * It is intended that this class be customized via inheritance,
1401  * to obtain fancier outputs.
1402  * @todo document
1403  * @private
1404  * @addtogroup DifferenceEngine
1405  */
1406 class DiffFormatter
1407 {
1408         /**
1409          * Number of leading context "lines" to preserve.
1410          *
1411          * This should be left at zero for this class, but subclasses
1412          * may want to set this to other values.
1413          */
1414         var $leading_context_lines = 0;
1415
1416         /**
1417          * Number of trailing context "lines" to preserve.
1418          *
1419          * This should be left at zero for this class, but subclasses
1420          * may want to set this to other values.
1421          */
1422         var $trailing_context_lines = 0;
1423
1424         /**
1425          * Format a diff.
1426          *
1427          * @param $diff object A Diff object.
1428          * @return string The formatted output.
1429          */
1430         function format($diff) {
1431                 $fname = 'DiffFormatter::format';
1432                 wfProfileIn( $fname );
1433
1434                 $xi = $yi = 1;
1435                 $block = false;
1436                 $context = array();
1437
1438                 $nlead = $this->leading_context_lines;
1439                 $ntrail = $this->trailing_context_lines;
1440
1441                 $this->_start_diff();
1442
1443                 foreach ($diff->edits as $edit) {
1444                         if ($edit->type == 'copy') {
1445                                 if (is_array($block)) {
1446                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1447                                                 $block[] = $edit;
1448                                         }
1449                                         else{
1450                                                 if ($ntrail) {
1451                                                         $context = array_slice($edit->orig, 0, $ntrail);
1452                                                         $block[] = new _DiffOp_Copy($context);
1453                                                 }
1454                                                 $this->_block($x0, $ntrail + $xi - $x0,
1455                                                                           $y0, $ntrail + $yi - $y0,
1456                                                                           $block);
1457                                                 $block = false;
1458                                         }
1459                                 }
1460                                 $context = $edit->orig;
1461                         }
1462                         else {
1463                                 if (! is_array($block)) {
1464                                         $context = array_slice($context, sizeof($context) - $nlead);
1465                                         $x0 = $xi - sizeof($context);
1466                                         $y0 = $yi - sizeof($context);
1467                                         $block = array();
1468                                         if ($context)
1469                                                 $block[] = new _DiffOp_Copy($context);
1470                                 }
1471                                 $block[] = $edit;
1472                         }
1473
1474                         if ($edit->orig)
1475                                 $xi += sizeof($edit->orig);
1476                         if ($edit->closing)
1477                                 $yi += sizeof($edit->closing);
1478                 }
1479
1480                 if (is_array($block))
1481                         $this->_block($x0, $xi - $x0,
1482                                                   $y0, $yi - $y0,
1483                                                   $block);
1484
1485                 $end = $this->_end_diff();
1486                 wfProfileOut( $fname );
1487                 return $end;
1488         }
1489
1490         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1491                 $fname = 'DiffFormatter::_block';
1492                 wfProfileIn( $fname );
1493                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1494                 foreach ($edits as $edit) {
1495                         if ($edit->type == 'copy')
1496                                 $this->_context($edit->orig);
1497                         elseif ($edit->type == 'add')
1498                                 $this->_added($edit->closing);
1499                         elseif ($edit->type == 'delete')
1500                                 $this->_deleted($edit->orig);
1501                         elseif ($edit->type == 'change')
1502                                 $this->_changed($edit->orig, $edit->closing);
1503                         else
1504                                 trigger_error('Unknown edit type', E_USER_ERROR);
1505                 }
1506                 $this->_end_block();
1507                 wfProfileOut( $fname );
1508         }
1509
1510         function _start_diff() {
1511                 ob_start();
1512         }
1513
1514         function _end_diff() {
1515                 $val = ob_get_contents();
1516                 ob_end_clean();
1517                 return $val;
1518         }
1519
1520         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1521                 if ($xlen > 1)
1522                         $xbeg .= "," . ($xbeg + $xlen - 1);
1523                 if ($ylen > 1)
1524                         $ybeg .= "," . ($ybeg + $ylen - 1);
1525
1526                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1527         }
1528
1529         function _start_block($header) {
1530                 echo $header;
1531         }
1532
1533         function _end_block() {
1534         }
1535
1536         function _lines($lines, $prefix = ' ') {
1537                 foreach ($lines as $line)
1538                         echo "$prefix $line\n";
1539         }
1540
1541         function _context($lines) {
1542                 $this->_lines($lines);
1543         }
1544
1545         function _added($lines) {
1546                 $this->_lines($lines, '>');
1547         }
1548         function _deleted($lines) {
1549                 $this->_lines($lines, '<');
1550         }
1551
1552         function _changed($orig, $closing) {
1553                 $this->_deleted($orig);
1554                 echo "---\n";
1555                 $this->_added($closing);
1556         }
1557 }
1558
1559
1560 /**
1561  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1562  *
1563  */
1564
1565 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1566
1567 /**
1568  * @todo document
1569  * @private
1570  * @addtogroup DifferenceEngine
1571  */
1572 class _HWLDF_WordAccumulator {
1573         function _HWLDF_WordAccumulator () {
1574                 $this->_lines = array();
1575                 $this->_line = '';
1576                 $this->_group = '';
1577                 $this->_tag = '';
1578         }
1579
1580         function _flushGroup ($new_tag) {
1581                 if ($this->_group !== '') {
1582                         if ($this->_tag == 'ins')
1583                                 $this->_line .= '<ins class="diffchange">' .
1584                                         htmlspecialchars ( $this->_group ) . '</ins>';
1585                         elseif ($this->_tag == 'del')
1586                                 $this->_line .= '<del class="diffchange">' .
1587                                         htmlspecialchars ( $this->_group ) . '</del>';
1588                         else
1589                                 $this->_line .= htmlspecialchars ( $this->_group );
1590                 }
1591                 $this->_group = '';
1592                 $this->_tag = $new_tag;
1593         }
1594
1595         function _flushLine ($new_tag) {
1596                 $this->_flushGroup($new_tag);
1597                 if ($this->_line != '')
1598                         array_push ( $this->_lines, $this->_line );
1599                 else
1600                         # make empty lines visible by inserting an NBSP
1601                         array_push ( $this->_lines, NBSP );
1602                 $this->_line = '';
1603         }
1604
1605         function addWords ($words, $tag = '') {
1606                 if ($tag != $this->_tag)
1607                         $this->_flushGroup($tag);
1608
1609                 foreach ($words as $word) {
1610                         // new-line should only come as first char of word.
1611                         if ($word == '')
1612                                 continue;
1613                         if ($word[0] == "\n") {
1614                                 $this->_flushLine($tag);
1615                                 $word = substr($word, 1);
1616                         }
1617                         assert(!strstr($word, "\n"));
1618                         $this->_group .= $word;
1619                 }
1620         }
1621
1622         function getLines() {
1623                 $this->_flushLine('~done');
1624                 return $this->_lines;
1625         }
1626 }
1627
1628 /**
1629  * @todo document
1630  * @private
1631  * @addtogroup DifferenceEngine
1632  */
1633 class WordLevelDiff extends MappedDiff
1634 {
1635         const MAX_LINE_LENGTH = 10000;
1636
1637         function WordLevelDiff ($orig_lines, $closing_lines) {
1638                 $fname = 'WordLevelDiff::WordLevelDiff';
1639                 wfProfileIn( $fname );
1640
1641                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1642                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1643
1644                 $this->MappedDiff($orig_words, $closing_words,
1645                                                   $orig_stripped, $closing_stripped);
1646                 wfProfileOut( $fname );
1647         }
1648
1649         function _split($lines) {
1650                 $fname = 'WordLevelDiff::_split';
1651                 wfProfileIn( $fname );
1652
1653                 $words = array();
1654                 $stripped = array();
1655                 $first = true;
1656                 foreach ( $lines as $line ) {
1657                         # If the line is too long, just pretend the entire line is one big word
1658                         # This prevents resource exhaustion problems
1659                         if ( $first ) {
1660                                 $first = false;
1661                         } else {
1662                                 $words[] = "\n";
1663                                 $stripped[] = "\n";
1664                         }
1665                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1666                                 $words[] = $line;
1667                                 $stripped[] = $line;
1668                         } else {
1669                                 $m = array();
1670                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1671                                         $line, $m))
1672                                 {
1673                                         $words = array_merge( $words, $m[0] );
1674                                         $stripped = array_merge( $stripped, $m[1] );
1675                                 }
1676                         }
1677                 }
1678                 wfProfileOut( $fname );
1679                 return array($words, $stripped);
1680         }
1681
1682         function orig () {
1683                 $fname = 'WordLevelDiff::orig';
1684                 wfProfileIn( $fname );
1685                 $orig = new _HWLDF_WordAccumulator;
1686
1687                 foreach ($this->edits as $edit) {
1688                         if ($edit->type == 'copy')
1689                                 $orig->addWords($edit->orig);
1690                         elseif ($edit->orig)
1691                                 $orig->addWords($edit->orig, 'del');
1692                 }
1693                 $lines = $orig->getLines();
1694                 wfProfileOut( $fname );
1695                 return $lines;
1696         }
1697
1698         function closing () {
1699                 $fname = 'WordLevelDiff::closing';
1700                 wfProfileIn( $fname );
1701                 $closing = new _HWLDF_WordAccumulator;
1702
1703                 foreach ($this->edits as $edit) {
1704                         if ($edit->type == 'copy')
1705                                 $closing->addWords($edit->closing);
1706                         elseif ($edit->closing)
1707                                 $closing->addWords($edit->closing, 'ins');
1708                 }
1709                 $lines = $closing->getLines();
1710                 wfProfileOut( $fname );
1711                 return $lines;
1712         }
1713 }
1714
1715 /**
1716  *      Wikipedia Table style diff formatter.
1717  * @todo document
1718  * @private
1719  * @addtogroup DifferenceEngine
1720  */
1721 class TableDiffFormatter extends DiffFormatter
1722 {
1723         function TableDiffFormatter() {
1724                 $this->leading_context_lines = 2;
1725                 $this->trailing_context_lines = 2;
1726         }
1727
1728         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1729                 $r = '<tr><td colspan="2" align="left"><strong><!--LINE '.$xbeg."--></strong></td>\n" .
1730                   '<td colspan="2" align="left"><strong><!--LINE '.$ybeg."--></strong></td></tr>\n";
1731                 return $r;
1732         }
1733
1734         function _start_block( $header ) {
1735                 echo $header;
1736         }
1737
1738         function _end_block() {
1739         }
1740
1741         function _lines( $lines, $prefix=' ', $color='white' ) {
1742         }
1743
1744         # HTML-escape parameter before calling this
1745         function addedLine( $line ) {
1746                 return $this->wrapLine( '+', 'diff-addedline', $line );
1747         }
1748
1749         # HTML-escape parameter before calling this
1750         function deletedLine( $line ) {
1751                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1752         }
1753
1754         # HTML-escape parameter before calling this
1755         function contextLine( $line ) {
1756                 return $this->wrapLine( ' ', 'diff-context', $line );
1757         }
1758
1759         private function wrapLine( $marker, $class, $line ) {
1760                 if( $line !== '' ) {
1761                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1762                         $line = "<div>$line</div>";
1763                 }
1764                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1765         }
1766
1767         function emptyLine() {
1768                 return '<td colspan="2">&nbsp;</td>';
1769         }
1770
1771         function _added( $lines ) {
1772                 foreach ($lines as $line) {
1773                         echo '<tr>' . $this->emptyLine() .
1774                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1775                 }
1776         }
1777
1778         function _deleted($lines) {
1779                 foreach ($lines as $line) {
1780                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1781                           $this->emptyLine() . "</tr>\n";
1782                 }
1783         }
1784
1785         function _context( $lines ) {
1786                 foreach ($lines as $line) {
1787                         echo '<tr>' .
1788                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1789                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1790                 }
1791         }
1792
1793         function _changed( $orig, $closing ) {
1794                 $fname = 'TableDiffFormatter::_changed';
1795                 wfProfileIn( $fname );
1796
1797                 $diff = new WordLevelDiff( $orig, $closing );
1798                 $del = $diff->orig();
1799                 $add = $diff->closing();
1800
1801                 # Notice that WordLevelDiff returns HTML-escaped output.
1802                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1803
1804                 while ( $line = array_shift( $del ) ) {
1805                         $aline = array_shift( $add );
1806                         echo '<tr>' . $this->deletedLine( $line ) .
1807                                 $this->addedLine( $aline ) . "</tr>\n";
1808                 }
1809                 foreach ($add as $line) {       # If any leftovers
1810                         echo '<tr>' . $this->emptyLine() .
1811                                 $this->addedLine( $line ) . "</tr>\n";
1812                 }
1813                 wfProfileOut( $fname );
1814         }
1815 }
1816
1817 ?>