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