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