includes/DifferenceEngine.php

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