includes/DifferenceEngine.php

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