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