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