includes/DifferenceEngine.php

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