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 /**
  12  * @todo document
  13  * @access public
  14  * @package MediaWiki
  15  * @subpackage DifferenceEngine
  16  */
  17 class DifferenceEngine {
  18         /**#@+
  19          * @access private
  20          */
  21         var $mOldid, $mNewid;
  22         var $mOldtitle, $mNewtitle, $mPagetitle;
  23         var $mOldtext, $mNewtext;
  24         var $mOldUser, $mNewUser;
  25         var $mOldComment, $mNewComment;
  26         var $mOldPage, $mNewPage;
  27         var $mRcidMarkPatrolled;
  28         /**#@-*/
  29
  30         /**
  31          * Constructor
  32          * @param integer $old Old ID we want to show and diff with.
  33          * @param string $new Either 'prev' or 'next'.
  34          * @param integer $rcid ??? (default 0)
  35          */
  36         function DifferenceEngine( $old, $new, $rcid = 0 ) {
  37                 global $wgTitle;
  38
  39                 if ( 'prev' == $new ) {
  40                         # Show diff between revision $old and the previous one.
  41                         # Get previous one from DB.
  42                         #
  43                         $this->mNewid = intval($old);
  44
  45                         $this->mOldid = $wgTitle->getPreviousRevisionID( $this->mNewid );
  46
  47                 } elseif ( 'next' == $new ) {
  48                         # Show diff between revision $old and the previous one.
  49                         # Get previous one from DB.
  50                         #
  51                         $this->mOldid = intval($old);
  52                         $this->mNewid = $wgTitle->getNextRevisionID( $this->mOldid );
  53                         if ( false === $this->mNewid ) {
  54                                 # if no result, NewId points to the newest old revision. The only newer
  55                                 # revision is cur, which is "0".
  56                                 $this->mNewid = 0;
  57                         }
  58
  59                 } else {
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = intval($new);
  62                 }
  63                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  64         }
  65
  66         function showDiffPage() {
  67                 global $wgUser, $wgTitle, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol,
  68                        $wgUseExternalEditor, $wgUseRCPatrol;
  69                 $fname = 'DifferenceEngine::showDiffPage';
  70                 wfProfileIn( $fname );
  71
  72                 # If external diffs are enabled both globally and for the user,
  73                 # we'll use the application/x-external-editor interface to call
  74                 # an external diff tool like kompare, kdiff3, etc.
  75                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  76                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  77                         $wgOut->disable();
  78                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  79                         $url1=$wgTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  80                         $url2=$wgTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  81                         $special=$wgLang->getNsText(NS_SPECIAL);
  82                         $control=<<<CONTROL
  83 [Process]
  84 Type=Diff text
  85 Engine=MediaWiki
  86 Script={$wgServer}{$wgScript}
  87 Special namespace={$special}
  88
  89 [File]
  90 Extension=wiki
  91 URL=$url1
  92
  93 [File 2]
  94 Extension=wiki
  95 URL=$url2
  96 CONTROL;
  97                         echo($control);
  98                         return;
  99                 }
 100
 101                 # mOldid is false if the difference engine is called with a "vague" query for
 102                 # a diff between a version V and its previous version V' AND the version V
 103                 # is the first version of that article. In that case, V' does not exist.
 104                 if ( $this->mOldid === false ) {
 105                         $this->showFirstRevision();
 106                         wfProfileOut( $fname );
 107                         return;
 108                 }
 109
 110                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 111                   "{$this->mNewid})";
 112                 $mtext = wfMsg( 'missingarticle', $t );
 113
 114                 $wgOut->setArticleFlag( false );
 115                 if ( ! $this->loadText() ) {
 116                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 117                         $wgOut->addHTML( $mtext );
 118                         wfProfileOut( $fname );
 119                         return;
 120                 }
 121                 $wgOut->suppressQuickbar();
 122
 123                 $oldTitle = $this->mOldPage->getPrefixedText();
 124                 $newTitle = $this->mNewPage->getPrefixedText();
 125                 if( $oldTitle == $newTitle ) {
 126                         $wgOut->setPageTitle( $newTitle );
 127                 } else {
 128                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 129                 }
 130                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 131                 $wgOut->setRobotpolicy( 'noindex,follow' );
 132
 133                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 134                         $wgOut->loginToUse();
 135                         $wgOut->output();
 136                         wfProfileOut( $fname );
 137                         exit;
 138                 }
 139
 140                 $sk = $wgUser->getSkin();
 141                 $talk = $wgContLang->getNsText( NS_TALK );
 142                 $contribs = wfMsg( 'contribslink' );
 143
 144                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 145                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 146
 147                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 148                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 149                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 150                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 151                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 152                         'target=' . urlencode($this->mOldUser) );
 153                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 154                         'target=' . urlencode($this->mNewUser) );
 155                 if ( $this->newRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 156                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
 157                                 'action=rollback&from=' . urlencode($this->mNewUser) .
 158                                 '&token=' . urlencode( $wgUser->editToken( array( $wgTitle->getPrefixedText(), $this->mNewUser ) ) ) ) .
 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( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 167                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 168                 } else {
 169                         $patrol = '';
 170                 }
 171
 172                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsgHtml( 'previousdiff' ),
 173                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 174                 if ( $this->newRev->isCurrent() ) {
 175                         $nextlink = '';
 176                 } else {
 177                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsgHtml( 'nextdiff' ),
 178                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 179                 }
 180
 181                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 182                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 183                         '<br />' . $prevlink;
 184                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 185                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 186                         '<br />' . $nextlink . $patrol;
 187
 188                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 189                   $oldHeader, $newHeader );
 190                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 191
 192                 if( !$this->newRev->isCurrent() ) {
 193                         $oldEditSectionSetting = $wgOut->mParserOptions->setEditSection( false );
 194                 }
 195                 $wgOut->addWikiText( $this->mNewtext );
 196                 if( !$this->newRev->isCurrent() ) {
 197                         $wgOut->mParserOptions->setEditSection( $oldEditSectionSetting );
 198                 }
 199
 200                 wfProfileOut( $fname );
 201         }
 202
 203         /**
 204          * Show the first revision of an article. Uses normal diff headers in
 205          * contrast to normal "old revision" display style.
 206          */
 207         function showFirstRevision() {
 208                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 209
 210                 $fname = 'DifferenceEngine::showFirstRevision';
 211                 wfProfileIn( $fname );
 212
 213
 214                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 215
 216                 # Get article text from the DB
 217                 #
 218                 if ( ! $this->loadText() ) {
 219                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 220                           "{$this->mNewid})";
 221                         $mtext = wfMsg( 'missingarticle', $t );
 222                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 223                         $wgOut->addHTML( $mtext );
 224                         wfProfileOut( $fname );
 225                         return;
 226                 }
 227
 228                 # Check if user is allowed to look at this page. If not, bail out.
 229                 #
 230                 if ( !( $this->mOldPage->userCanRead() ) ) {
 231                         $wgOut->loginToUse();
 232                         $wgOut->output();
 233                         wfProfileOut( $fname );
 234                         exit;
 235                 }
 236
 237                 # Prepare the header box
 238                 #
 239                 $sk = $wgUser->getSkin();
 240
 241                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 242                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 243                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 244                         'target=' . urlencode($this->mOldUser) );
 245                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 246                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 247                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 248                         '<br />' . $nextlink. "</div>\n";
 249
 250                 $wgOut->addHTML( $header );
 251
 252                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 253                 $wgOut->setRobotpolicy( 'noindex,follow' );
 254
 255
 256                 # Show current revision
 257                 #
 258                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 259                 $wgOut->addWikiText( $this->mNewtext );
 260
 261                 wfProfileOut( $fname );
 262         }
 263
 264         function showDiff( $otext, $ntext, $otitle, $ntitle ) {
 265                 global $wgOut;
 266                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 267         }
 268
 269         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 270                 global $wgUseExternalDiffEngine, $wgContLang;
 271                 $out = "
 272                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 273                         <tr>
 274                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 275                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 276                         </tr>
 277                 ";
 278                 $otext = $wgContLang->segmentForDiff($otext);
 279                 $ntext = $wgContLang->segmentForDiff($ntext);
 280                 $difftext='';
 281                 if ( $wgUseExternalDiffEngine ) {
 282                         # For historical reasons, external diff engine expects
 283                         # input text to be HTML-escaped already
 284                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 285                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 286                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 287                                 dl('php_wikidiff.so');
 288                         }
 289                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 290                 } else {
 291                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 292                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 293                         $diffs =& new Diff( $ota, $nta );
 294                         $formatter =& new TableDiffFormatter();
 295                         $difftext = $formatter->format( $diffs );
 296                 }
 297                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 298                 $out .= $difftext."</table>\n";
 299                 return $out;
 300         }
 301
 302         /**
 303          * Load the text of the articles to compare.  If newid is 0, then compare
 304          * the old article in oldid to the current article; if oldid is 0, then
 305          * compare the current article to the immediately previous one (ignoring the
 306          * value of newid).
 307          */
 308         function loadText() {
 309                 global $wgTitle, $wgOut, $wgLang;
 310                 $fname = 'DifferenceEngine::loadText';
 311
 312                 $dbr =& wfGetDB( DB_SLAVE );
 313                 if( $this->mNewid ) {
 314                         $this->newRev = Revision::newFromId( $this->mNewid );
 315                 } else {
 316                         $this->newRev = Revision::newFromTitle( $wgTitle );
 317                 }
 318                 if( is_null( $this->newRev ) ) {
 319                         return false;
 320                 }
 321
 322                 if( $this->newRev->isCurrent() ) {
 323                         $wgOut->setArticleFlag( true );
 324                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 325                         $this->mNewPage = $wgTitle;
 326                         $newLink = $this->mNewPage->escapeLocalUrl();
 327                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 328                 } else {
 329                         $this->mNewPage = $this->newRev->getTitle();
 330                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 331                         $t = $wgLang->timeanddate( $this->newRev->getTimestamp(), true );
 332                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 333                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 334                 }
 335
 336                 if( $this->mOldid ) {
 337                         $this->oldRev = Revision::newFromId( $this->mOldid );
 338                 } else {
 339                         $this->oldRev = $this->newRev->getPrevious();
 340                         $this->mOldid = $this->oldRev->getId();
 341                 }
 342                 if( is_null( $this->oldRev ) ) {
 343                         return false;
 344                 }
 345
 346                 $this->mOldPage = $this->oldRev->getTitle();
 347
 348                 $t = $wgLang->timeanddate( $this->oldRev->getTimestamp(), true );
 349                 $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 350                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 351
 352                 $this->mNewUser = $this->newRev->getUserText();
 353                 $this->mNewComment = $this->newRev->getComment();
 354                 $this->mNewtext = $this->newRev->getText();
 355
 356                 $this->mOldUser = $this->oldRev->getUserText();
 357                 $this->mOldComment = $this->oldRev->getComment();
 358                 $this->mOldtext = $this->oldRev->getText();
 359
 360                 return true;
 361         }
 362 }
 363
 364 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 365 //
 366 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 367 // You may copy this code freely under the conditions of the GPL.
 368 //
 369
 370 define('USE_ASSERTS', function_exists('assert'));
 371
 372 /**
 373  * @todo document
 374  * @access private
 375  * @package MediaWiki
 376  * @subpackage DifferenceEngine
 377  */
 378 class _DiffOp {
 379         var $type;
 380         var $orig;
 381         var $closing;
 382
 383         function reverse() {
 384                 trigger_error('pure virtual', E_USER_ERROR);
 385         }
 386
 387         function norig() {
 388                 return $this->orig ? sizeof($this->orig) : 0;
 389         }
 390
 391         function nclosing() {
 392                 return $this->closing ? sizeof($this->closing) : 0;
 393         }
 394 }
 395
 396 /**
 397  * @todo document
 398  * @access private
 399  * @package MediaWiki
 400  * @subpackage DifferenceEngine
 401  */
 402 class _DiffOp_Copy extends _DiffOp {
 403         var $type = 'copy';
 404
 405         function _DiffOp_Copy ($orig, $closing = false) {
 406                 if (!is_array($closing))
 407                         $closing = $orig;
 408                 $this->orig = $orig;
 409                 $this->closing = $closing;
 410         }
 411
 412         function reverse() {
 413                 return new _DiffOp_Copy($this->closing, $this->orig);
 414         }
 415 }
 416
 417 /**
 418  * @todo document
 419  * @access private
 420  * @package MediaWiki
 421  * @subpackage DifferenceEngine
 422  */
 423 class _DiffOp_Delete extends _DiffOp {
 424         var $type = 'delete';
 425
 426         function _DiffOp_Delete ($lines) {
 427                 $this->orig = $lines;
 428                 $this->closing = false;
 429         }
 430
 431         function reverse() {
 432                 return new _DiffOp_Add($this->orig);
 433         }
 434 }
 435
 436 /**
 437  * @todo document
 438  * @access private
 439  * @package MediaWiki
 440  * @subpackage DifferenceEngine
 441  */
 442 class _DiffOp_Add extends _DiffOp {
 443         var $type = 'add';
 444
 445         function _DiffOp_Add ($lines) {
 446                 $this->closing = $lines;
 447                 $this->orig = false;
 448         }
 449
 450         function reverse() {
 451                 return new _DiffOp_Delete($this->closing);
 452         }
 453 }
 454
 455 /**
 456  * @todo document
 457  * @access private
 458  * @package MediaWiki
 459  * @subpackage DifferenceEngine
 460  */
 461 class _DiffOp_Change extends _DiffOp {
 462         var $type = 'change';
 463
 464         function _DiffOp_Change ($orig, $closing) {
 465                 $this->orig = $orig;
 466                 $this->closing = $closing;
 467         }
 468
 469         function reverse() {
 470                 return new _DiffOp_Change($this->closing, $this->orig);
 471         }
 472 }
 473
 474
 475 /**
 476  * Class used internally by Diff to actually compute the diffs.
 477  *
 478  * The algorithm used here is mostly lifted from the perl module
 479  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 480  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 481  *
 482  * More ideas are taken from:
 483  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 484  *
 485  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 486  * diffutils-2.7, which can be found at:
 487  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 488  *
 489  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 490  * are my own.
 491  *
 492  * @author Geoffrey T. Dairiki
 493  * @access private
 494  * @package MediaWiki
 495  * @subpackage DifferenceEngine
 496  */
 497 class _DiffEngine
 498 {
 499         function diff ($from_lines, $to_lines) {
 500                 $fname = '_DiffEngine::diff';
 501                 wfProfileIn( $fname );
 502
 503                 $n_from = sizeof($from_lines);
 504                 $n_to = sizeof($to_lines);
 505
 506                 $this->xchanged = $this->ychanged = array();
 507                 $this->xv = $this->yv = array();
 508                 $this->xind = $this->yind = array();
 509                 unset($this->seq);
 510                 unset($this->in_seq);
 511                 unset($this->lcs);
 512
 513                 // Skip leading common lines.
 514                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 515                         if ($from_lines[$skip] != $to_lines[$skip])
 516                                 break;
 517                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 518                 }
 519                 // Skip trailing common lines.
 520                 $xi = $n_from; $yi = $n_to;
 521                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 522                         if ($from_lines[$xi] != $to_lines[$yi])
 523                                 break;
 524                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 525                 }
 526
 527                 // Ignore lines which do not exist in both files.
 528                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 529                         $xhash[$from_lines[$xi]] = 1;
 530                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 531                         $line = $to_lines[$yi];
 532                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 533                                 continue;
 534                         $yhash[$line] = 1;
 535                         $this->yv[] = $line;
 536                         $this->yind[] = $yi;
 537                 }
 538                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 539                         $line = $from_lines[$xi];
 540                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 541                                 continue;
 542                         $this->xv[] = $line;
 543                         $this->xind[] = $xi;
 544                 }
 545
 546                 // Find the LCS.
 547                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 548
 549                 // Merge edits when possible
 550                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 551                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 552
 553                 // Compute the edit operations.
 554                 $edits = array();
 555                 $xi = $yi = 0;
 556                 while ($xi < $n_from || $yi < $n_to) {
 557                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 558                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 559
 560                         // Skip matching "snake".
 561                         $copy = array();
 562                         while ( $xi < $n_from && $yi < $n_to
 563                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 564                                 $copy[] = $from_lines[$xi++];
 565                                 ++$yi;
 566                         }
 567                         if ($copy)
 568                                 $edits[] = new _DiffOp_Copy($copy);
 569
 570                         // Find deletes & adds.
 571                         $delete = array();
 572                         while ($xi < $n_from && $this->xchanged[$xi])
 573                                 $delete[] = $from_lines[$xi++];
 574
 575                         $add = array();
 576                         while ($yi < $n_to && $this->ychanged[$yi])
 577                                 $add[] = $to_lines[$yi++];
 578
 579                         if ($delete && $add)
 580                                 $edits[] = new _DiffOp_Change($delete, $add);
 581                         elseif ($delete)
 582                                 $edits[] = new _DiffOp_Delete($delete);
 583                         elseif ($add)
 584                                 $edits[] = new _DiffOp_Add($add);
 585                 }
 586                 wfProfileOut( $fname );
 587                 return $edits;
 588         }
 589
 590
 591         /* Divide the Largest Common Subsequence (LCS) of the sequences
 592          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 593          * sized segments.
 594          *
 595          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 596          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 597          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 598          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 599          * that (X0, Y0) == (XOFF, YOFF) and
 600          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 601          *
 602          * This function assumes that the first lines of the specified portions
 603          * of the two files do not match, and likewise that the last lines do not
 604          * match.  The caller must trim matching lines from the beginning and end
 605          * of the portions it is going to specify.
 606          */
 607         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 608                 $fname = '_DiffEngine::_diag';
 609                 wfProfileIn( $fname );
 610                 $flip = false;
 611
 612                 if ($xlim - $xoff > $ylim - $yoff) {
 613                         // Things seems faster (I'm not sure I understand why)
 614                                 // when the shortest sequence in X.
 615                                 $flip = true;
 616                         list ($xoff, $xlim, $yoff, $ylim)
 617                         = array( $yoff, $ylim, $xoff, $xlim);
 618                 }
 619
 620                 if ($flip)
 621                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 622                                 $ymatches[$this->xv[$i]][] = $i;
 623                 else
 624                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 625                                 $ymatches[$this->yv[$i]][] = $i;
 626
 627                 $this->lcs = 0;
 628                 $this->seq[0]= $yoff - 1;
 629                 $this->in_seq = array();
 630                 $ymids[0] = array();
 631
 632                 $numer = $xlim - $xoff + $nchunks - 1;
 633                 $x = $xoff;
 634                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 635                         wfProfileIn( "$fname-chunk" );
 636                         if ($chunk > 0)
 637                                 for ($i = 0; $i <= $this->lcs; $i++)
 638                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 639
 640                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 641                         for ( ; $x < $x1; $x++) {
 642                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 643                                         if (empty($ymatches[$line]))
 644                                                 continue;
 645                                 $matches = $ymatches[$line];
 646                                 reset($matches);
 647                                 while (list ($junk, $y) = each($matches))
 648                                         if (empty($this->in_seq[$y])) {
 649                                                 $k = $this->_lcs_pos($y);
 650                                                 USE_ASSERTS && assert($k > 0);
 651                                                 $ymids[$k] = $ymids[$k-1];
 652                                                 break;
 653                                         }
 654                                 while (list ($junk, $y) = each($matches)) {
 655                                         if ($y > $this->seq[$k-1]) {
 656                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 657                                                 // Optimization: this is a common case:
 658                                                 //      next match is just replacing previous match.
 659                                                 $this->in_seq[$this->seq[$k]] = false;
 660                                                 $this->seq[$k] = $y;
 661                                                 $this->in_seq[$y] = 1;
 662                                         } else if (empty($this->in_seq[$y])) {
 663                                                 $k = $this->_lcs_pos($y);
 664                                                 USE_ASSERTS && assert($k > 0);
 665                                                 $ymids[$k] = $ymids[$k-1];
 666                                         }
 667                                 }
 668                         }
 669                         wfProfileOut( "$fname-chunk" );
 670                 }
 671
 672                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 673                 $ymid = $ymids[$this->lcs];
 674                 for ($n = 0; $n < $nchunks - 1; $n++) {
 675                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 676                         $y1 = $ymid[$n] + 1;
 677                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 678                 }
 679                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 680
 681                 wfProfileOut( $fname );
 682                 return array($this->lcs, $seps);
 683         }
 684
 685         function _lcs_pos ($ypos) {
 686                 $fname = '_DiffEngine::_lcs_pos';
 687                 wfProfileIn( $fname );
 688
 689                 $end = $this->lcs;
 690                 if ($end == 0 || $ypos > $this->seq[$end]) {
 691                         $this->seq[++$this->lcs] = $ypos;
 692                         $this->in_seq[$ypos] = 1;
 693                         wfProfileOut( $fname );
 694                         return $this->lcs;
 695                 }
 696
 697                 $beg = 1;
 698                 while ($beg < $end) {
 699                         $mid = (int)(($beg + $end) / 2);
 700                         if ( $ypos > $this->seq[$mid] )
 701                                 $beg = $mid + 1;
 702                         else
 703                                 $end = $mid;
 704                 }
 705
 706                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 707
 708                 $this->in_seq[$this->seq[$end]] = false;
 709                 $this->seq[$end] = $ypos;
 710                 $this->in_seq[$ypos] = 1;
 711                 wfProfileOut( $fname );
 712                 return $end;
 713         }
 714
 715         /* Find LCS of two sequences.
 716          *
 717          * The results are recorded in the vectors $this->{x,y}changed[], by
 718          * storing a 1 in the element for each line that is an insertion
 719          * or deletion (ie. is not in the LCS).
 720          *
 721          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 722          *
 723          * Note that XLIM, YLIM are exclusive bounds.
 724          * All line numbers are origin-0 and discarded lines are not counted.
 725          */
 726         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 727                 $fname = '_DiffEngine::_compareseq';
 728                 wfProfileIn( $fname );
 729
 730                 // Slide down the bottom initial diagonal.
 731                 while ($xoff < $xlim && $yoff < $ylim
 732                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 733                         ++$xoff;
 734                         ++$yoff;
 735                 }
 736
 737                 // Slide up the top initial diagonal.
 738                 while ($xlim > $xoff && $ylim > $yoff
 739                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 740                         --$xlim;
 741                         --$ylim;
 742                 }
 743
 744                 if ($xoff == $xlim || $yoff == $ylim)
 745                         $lcs = 0;
 746                 else {
 747                         // This is ad hoc but seems to work well.
 748                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 749                         //$nchunks = max(2,min(8,(int)$nchunks));
 750                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 751                         list ($lcs, $seps)
 752                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 753                 }
 754
 755                 if ($lcs == 0) {
 756                         // X and Y sequences have no common subsequence:
 757                         // mark all changed.
 758                         while ($yoff < $ylim)
 759                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 760                         while ($xoff < $xlim)
 761                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 762                 } else {
 763                         // Use the partitions to split this problem into subproblems.
 764                         reset($seps);
 765                         $pt1 = $seps[0];
 766                         while ($pt2 = next($seps)) {
 767                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 768                                 $pt1 = $pt2;
 769                         }
 770                 }
 771                 wfProfileOut( $fname );
 772         }
 773
 774         /* Adjust inserts/deletes of identical lines to join changes
 775          * as much as possible.
 776          *
 777          * We do something when a run of changed lines include a
 778          * line at one end and has an excluded, identical line at the other.
 779          * We are free to choose which identical line is included.
 780          * `compareseq' usually chooses the one at the beginning,
 781          * but usually it is cleaner to consider the following identical line
 782          * to be the "change".
 783          *
 784          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 785          */
 786         function _shift_boundaries ($lines, &$changed, $other_changed) {
 787                 $fname = '_DiffEngine::_shift_boundaries';
 788                 wfProfileIn( $fname );
 789                 $i = 0;
 790                 $j = 0;
 791
 792                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 793                 $len = sizeof($lines);
 794                 $other_len = sizeof($other_changed);
 795
 796                 while (1) {
 797                         /*
 798                          * Scan forwards to find beginning of another run of changes.
 799                          * Also keep track of the corresponding point in the other file.
 800                          *
 801                          * Throughout this code, $i and $j are adjusted together so that
 802                          * the first $i elements of $changed and the first $j elements
 803                          * of $other_changed both contain the same number of zeros
 804                          * (unchanged lines).
 805                          * Furthermore, $j is always kept so that $j == $other_len or
 806                          * $other_changed[$j] == false.
 807                          */
 808                         while ($j < $other_len && $other_changed[$j])
 809                                 $j++;
 810
 811                         while ($i < $len && ! $changed[$i]) {
 812                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 813                                 $i++; $j++;
 814                                 while ($j < $other_len && $other_changed[$j])
 815                                         $j++;
 816                         }
 817
 818                         if ($i == $len)
 819                                 break;
 820
 821                         $start = $i;
 822
 823                         // Find the end of this run of changes.
 824                         while (++$i < $len && $changed[$i])
 825                                 continue;
 826
 827                         do {
 828                                 /*
 829                                  * Record the length of this run of changes, so that
 830                                  * we can later determine whether the run has grown.
 831                                  */
 832                                 $runlength = $i - $start;
 833
 834                                 /*
 835                                  * Move the changed region back, so long as the
 836                                  * previous unchanged line matches the last changed one.
 837                                  * This merges with previous changed regions.
 838                                  */
 839                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 840                                         $changed[--$start] = 1;
 841                                         $changed[--$i] = false;
 842                                         while ($start > 0 && $changed[$start - 1])
 843                                                 $start--;
 844                                         USE_ASSERTS && assert('$j > 0');
 845                                         while ($other_changed[--$j])
 846                                                 continue;
 847                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 848                                 }
 849
 850                                 /*
 851                                  * Set CORRESPONDING to the end of the changed run, at the last
 852                                  * point where it corresponds to a changed run in the other file.
 853                                  * CORRESPONDING == LEN means no such point has been found.
 854                                  */
 855                                 $corresponding = $j < $other_len ? $i : $len;
 856
 857                                 /*
 858                                  * Move the changed region forward, so long as the
 859                                  * first changed line matches the following unchanged one.
 860                                  * This merges with following changed regions.
 861                                  * Do this second, so that if there are no merges,
 862                                  * the changed region is moved forward as far as possible.
 863                                  */
 864                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 865                                         $changed[$start++] = false;
 866                                         $changed[$i++] = 1;
 867                                         while ($i < $len && $changed[$i])
 868                                                 $i++;
 869
 870                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 871                                         $j++;
 872                                         if ($j < $other_len && $other_changed[$j]) {
 873                                                 $corresponding = $i;
 874                                                 while ($j < $other_len && $other_changed[$j])
 875                                                         $j++;
 876                                         }
 877                                 }
 878                         } while ($runlength != $i - $start);
 879
 880                         /*
 881                          * If possible, move the fully-merged run of changes
 882                          * back to a corresponding run in the other file.
 883                          */
 884                         while ($corresponding < $i) {
 885                                 $changed[--$start] = 1;
 886                                 $changed[--$i] = 0;
 887                                 USE_ASSERTS && assert('$j > 0');
 888                                 while ($other_changed[--$j])
 889                                         continue;
 890                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 891                         }
 892                 }
 893                 wfProfileOut( $fname );
 894         }
 895 }
 896
 897 /**
 898  * Class representing a 'diff' between two sequences of strings.
 899  * @todo document
 900  * @access private
 901  * @package MediaWiki
 902  * @subpackage DifferenceEngine
 903  */
 904 class Diff
 905 {
 906         var $edits;
 907
 908         /**
 909          * Constructor.
 910          * Computes diff between sequences of strings.
 911          *
 912          * @param $from_lines array An array of strings.
 913          *                (Typically these are lines from a file.)
 914          * @param $to_lines array An array of strings.
 915          */
 916         function Diff($from_lines, $to_lines) {
 917                 $eng = new _DiffEngine;
 918                 $this->edits = $eng->diff($from_lines, $to_lines);
 919                 //$this->_check($from_lines, $to_lines);
 920         }
 921
 922         /**
 923          * Compute reversed Diff.
 924          *
 925          * SYNOPSIS:
 926          *
 927          *      $diff = new Diff($lines1, $lines2);
 928          *      $rev = $diff->reverse();
 929          * @return object A Diff object representing the inverse of the
 930          *                                original diff.
 931          */
 932         function reverse () {
 933                 $rev = $this;
 934                 $rev->edits = array();
 935                 foreach ($this->edits as $edit) {
 936                         $rev->edits[] = $edit->reverse();
 937                 }
 938                 return $rev;
 939         }
 940
 941         /**
 942          * Check for empty diff.
 943          *
 944          * @return bool True iff two sequences were identical.
 945          */
 946         function isEmpty () {
 947                 foreach ($this->edits as $edit) {
 948                         if ($edit->type != 'copy')
 949                                 return false;
 950                 }
 951                 return true;
 952         }
 953
 954         /**
 955          * Compute the length of the Longest Common Subsequence (LCS).
 956          *
 957          * This is mostly for diagnostic purposed.
 958          *
 959          * @return int The length of the LCS.
 960          */
 961         function lcs () {
 962                 $lcs = 0;
 963                 foreach ($this->edits as $edit) {
 964                         if ($edit->type == 'copy')
 965                                 $lcs += sizeof($edit->orig);
 966                 }
 967                 return $lcs;
 968         }
 969
 970         /**
 971          * Get the original set of lines.
 972          *
 973          * This reconstructs the $from_lines parameter passed to the
 974          * constructor.
 975          *
 976          * @return array The original sequence of strings.
 977          */
 978         function orig() {
 979                 $lines = array();
 980
 981                 foreach ($this->edits as $edit) {
 982                         if ($edit->orig)
 983                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 984                 }
 985                 return $lines;
 986         }
 987
 988         /**
 989          * Get the closing set of lines.
 990          *
 991          * This reconstructs the $to_lines parameter passed to the
 992          * constructor.
 993          *
 994          * @return array The sequence of strings.
 995          */
 996         function closing() {
 997                 $lines = array();
 998
 999                 foreach ($this->edits as $edit) {
1000                         if ($edit->closing)
1001                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1002                 }
1003                 return $lines;
1004         }
1005
1006         /**
1007          * Check a Diff for validity.
1008          *
1009          * This is here only for debugging purposes.
1010          */
1011         function _check ($from_lines, $to_lines) {
1012                 $fname = 'Diff::_check';
1013                 wfProfileIn( $fname );
1014                 if (serialize($from_lines) != serialize($this->orig()))
1015                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1016                 if (serialize($to_lines) != serialize($this->closing()))
1017                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1018
1019                 $rev = $this->reverse();
1020                 if (serialize($to_lines) != serialize($rev->orig()))
1021                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1022                 if (serialize($from_lines) != serialize($rev->closing()))
1023                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1024
1025
1026                 $prevtype = 'none';
1027                 foreach ($this->edits as $edit) {
1028                         if ( $prevtype == $edit->type )
1029                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1030                         $prevtype = $edit->type;
1031                 }
1032
1033                 $lcs = $this->lcs();
1034                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1035                 wfProfileOut( $fname );
1036         }
1037 }
1038
1039 /**
1040  * FIXME: bad name.
1041  * @todo document
1042  * @access private
1043  * @package MediaWiki
1044  * @subpackage DifferenceEngine
1045  */
1046 class MappedDiff extends Diff
1047 {
1048         /**
1049          * Constructor.
1050          *
1051          * Computes diff between sequences of strings.
1052          *
1053          * This can be used to compute things like
1054          * case-insensitve diffs, or diffs which ignore
1055          * changes in white-space.
1056          *
1057          * @param $from_lines array An array of strings.
1058          *      (Typically these are lines from a file.)
1059          *
1060          * @param $to_lines array An array of strings.
1061          *
1062          * @param $mapped_from_lines array This array should
1063          *      have the same size number of elements as $from_lines.
1064          *      The elements in $mapped_from_lines and
1065          *      $mapped_to_lines are what is actually compared
1066          *      when computing the diff.
1067          *
1068          * @param $mapped_to_lines array This array should
1069          *      have the same number of elements as $to_lines.
1070          */
1071         function MappedDiff($from_lines, $to_lines,
1072                                                 $mapped_from_lines, $mapped_to_lines) {
1073                 $fname = 'MappedDiff::MappedDiff';
1074                 wfProfileIn( $fname );
1075
1076                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1077                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1078
1079                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1080
1081                 $xi = $yi = 0;
1082                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1083                         $orig = &$this->edits[$i]->orig;
1084                         if (is_array($orig)) {
1085                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1086                                 $xi += sizeof($orig);
1087                         }
1088
1089                         $closing = &$this->edits[$i]->closing;
1090                         if (is_array($closing)) {
1091                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1092                                 $yi += sizeof($closing);
1093                         }
1094                 }
1095                 wfProfileOut( $fname );
1096         }
1097 }
1098
1099 /**
1100  * A class to format Diffs
1101  *
1102  * This class formats the diff in classic diff format.
1103  * It is intended that this class be customized via inheritance,
1104  * to obtain fancier outputs.
1105  * @todo document
1106  * @access private
1107  * @package MediaWiki
1108  * @subpackage DifferenceEngine
1109  */
1110 class DiffFormatter
1111 {
1112         /**
1113          * Number of leading context "lines" to preserve.
1114          *
1115          * This should be left at zero for this class, but subclasses
1116          * may want to set this to other values.
1117          */
1118         var $leading_context_lines = 0;
1119
1120         /**
1121          * Number of trailing context "lines" to preserve.
1122          *
1123          * This should be left at zero for this class, but subclasses
1124          * may want to set this to other values.
1125          */
1126         var $trailing_context_lines = 0;
1127
1128         /**
1129          * Format a diff.
1130          *
1131          * @param $diff object A Diff object.
1132          * @return string The formatted output.
1133          */
1134         function format($diff) {
1135                 $fname = 'DiffFormatter::format';
1136                 wfProfileIn( $fname );
1137
1138                 $xi = $yi = 1;
1139                 $block = false;
1140                 $context = array();
1141
1142                 $nlead = $this->leading_context_lines;
1143                 $ntrail = $this->trailing_context_lines;
1144
1145                 $this->_start_diff();
1146
1147                 foreach ($diff->edits as $edit) {
1148                         if ($edit->type == 'copy') {
1149                                 if (is_array($block)) {
1150                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1151                                                 $block[] = $edit;
1152                                         }
1153                                         else{
1154                                                 if ($ntrail) {
1155                                                         $context = array_slice($edit->orig, 0, $ntrail);
1156                                                         $block[] = new _DiffOp_Copy($context);
1157                                                 }
1158                                                 $this->_block($x0, $ntrail + $xi - $x0,
1159                                                                           $y0, $ntrail + $yi - $y0,
1160                                                                           $block);
1161                                                 $block = false;
1162                                         }
1163                                 }
1164                                 $context = $edit->orig;
1165                         }
1166                         else {
1167                                 if (! is_array($block)) {
1168                                         $context = array_slice($context, sizeof($context) - $nlead);
1169                                         $x0 = $xi - sizeof($context);
1170                                         $y0 = $yi - sizeof($context);
1171                                         $block = array();
1172                                         if ($context)
1173                                                 $block[] = new _DiffOp_Copy($context);
1174                                 }
1175                                 $block[] = $edit;
1176                         }
1177
1178                         if ($edit->orig)
1179                                 $xi += sizeof($edit->orig);
1180                         if ($edit->closing)
1181                                 $yi += sizeof($edit->closing);
1182                 }
1183
1184                 if (is_array($block))
1185                         $this->_block($x0, $xi - $x0,
1186                                                   $y0, $yi - $y0,
1187                                                   $block);
1188
1189                 $end = $this->_end_diff();
1190                 wfProfileOut( $fname );
1191                 return $end;
1192         }
1193
1194         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1195                 $fname = 'DiffFormatter::_block';
1196                 wfProfileIn( $fname );
1197                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1198                 foreach ($edits as $edit) {
1199                         if ($edit->type == 'copy')
1200                                 $this->_context($edit->orig);
1201                         elseif ($edit->type == 'add')
1202                                 $this->_added($edit->closing);
1203                         elseif ($edit->type == 'delete')
1204                                 $this->_deleted($edit->orig);
1205                         elseif ($edit->type == 'change')
1206                                 $this->_changed($edit->orig, $edit->closing);
1207                         else
1208                                 trigger_error('Unknown edit type', E_USER_ERROR);
1209                 }
1210                 $this->_end_block();
1211                 wfProfileOut( $fname );
1212         }
1213
1214         function _start_diff() {
1215                 ob_start();
1216         }
1217
1218         function _end_diff() {
1219                 $val = ob_get_contents();
1220                 ob_end_clean();
1221                 return $val;
1222         }
1223
1224         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1225                 if ($xlen > 1)
1226                         $xbeg .= "," . ($xbeg + $xlen - 1);
1227                 if ($ylen > 1)
1228                         $ybeg .= "," . ($ybeg + $ylen - 1);
1229
1230                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1231         }
1232
1233         function _start_block($header) {
1234                 echo $header;
1235         }
1236
1237         function _end_block() {
1238         }
1239
1240         function _lines($lines, $prefix = ' ') {
1241                 foreach ($lines as $line)
1242                         echo "$prefix $line\n";
1243         }
1244
1245         function _context($lines) {
1246                 $this->_lines($lines);
1247         }
1248
1249         function _added($lines) {
1250                 $this->_lines($lines, '>');
1251         }
1252         function _deleted($lines) {
1253                 $this->_lines($lines, '<');
1254         }
1255
1256         function _changed($orig, $closing) {
1257                 $this->_deleted($orig);
1258                 echo "---\n";
1259                 $this->_added($closing);
1260         }
1261 }
1262
1263
1264 /**
1265  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1266  *
1267  */
1268
1269 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1270
1271 /**
1272  * @todo document
1273  * @access private
1274  * @package MediaWiki
1275  * @subpackage DifferenceEngine
1276  */
1277 class _HWLDF_WordAccumulator {
1278         function _HWLDF_WordAccumulator () {
1279                 $this->_lines = array();
1280                 $this->_line = '';
1281                 $this->_group = '';
1282                 $this->_tag = '';
1283         }
1284
1285         function _flushGroup ($new_tag) {
1286                 if ($this->_group !== '') {
1287                         if ($this->_tag == 'mark')
1288                                 $this->_line .= '<span class="diffchange">' .
1289                                         htmlspecialchars ( $this->_group ) . '</span>';
1290                         else
1291                                 $this->_line .= htmlspecialchars ( $this->_group );
1292                 }
1293                 $this->_group = '';
1294                 $this->_tag = $new_tag;
1295         }
1296
1297         function _flushLine ($new_tag) {
1298                 $this->_flushGroup($new_tag);
1299                 if ($this->_line != '')
1300                         array_push ( $this->_lines, $this->_line );
1301                 else
1302                         # make empty lines visible by inserting an NBSP
1303                         array_push ( $this->_lines, NBSP );
1304                 $this->_line = '';
1305         }
1306
1307         function addWords ($words, $tag = '') {
1308                 if ($tag != $this->_tag)
1309                         $this->_flushGroup($tag);
1310
1311                 foreach ($words as $word) {
1312                         // new-line should only come as first char of word.
1313                         if ($word == '')
1314                                 continue;
1315                         if ($word[0] == "\n") {
1316                                 $this->_flushLine($tag);
1317                                 $word = substr($word, 1);
1318                         }
1319                         assert(!strstr($word, "\n"));
1320                         $this->_group .= $word;
1321                 }
1322         }
1323
1324         function getLines() {
1325                 $this->_flushLine('~done');
1326                 return $this->_lines;
1327         }
1328 }
1329
1330 /**
1331  * @todo document
1332  * @access private
1333  * @package MediaWiki
1334  * @subpackage DifferenceEngine
1335  */
1336 class WordLevelDiff extends MappedDiff
1337 {
1338         function WordLevelDiff ($orig_lines, $closing_lines) {
1339                 $fname = 'WordLevelDiff::WordLevelDiff';
1340                 wfProfileIn( $fname );
1341
1342                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1343                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1344
1345
1346                 $this->MappedDiff($orig_words, $closing_words,
1347                                                   $orig_stripped, $closing_stripped);
1348                 wfProfileOut( $fname );
1349         }
1350
1351         function _split($lines) {
1352                 $fname = 'WordLevelDiff::_split';
1353                 wfProfileIn( $fname );
1354                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1355                                                         implode("\n", $lines),
1356                                                         $m)) {
1357                         wfProfileOut( $fname );
1358                         return array(array(''), array(''));
1359                 }
1360                 wfProfileOut( $fname );
1361                 return array($m[0], $m[1]);
1362         }
1363
1364         function orig () {
1365                 $fname = 'WordLevelDiff::orig';
1366                 wfProfileIn( $fname );
1367                 $orig = new _HWLDF_WordAccumulator;
1368
1369                 foreach ($this->edits as $edit) {
1370                         if ($edit->type == 'copy')
1371                                 $orig->addWords($edit->orig);
1372                         elseif ($edit->orig)
1373                                 $orig->addWords($edit->orig, 'mark');
1374                 }
1375                 $lines = $orig->getLines();
1376                 wfProfileOut( $fname );
1377                 return $lines;
1378         }
1379
1380         function closing () {
1381                 $fname = 'WordLevelDiff::closing';
1382                 wfProfileIn( $fname );
1383                 $closing = new _HWLDF_WordAccumulator;
1384
1385                 foreach ($this->edits as $edit) {
1386                         if ($edit->type == 'copy')
1387                                 $closing->addWords($edit->closing);
1388                         elseif ($edit->closing)
1389                                 $closing->addWords($edit->closing, 'mark');
1390                 }
1391                 $lines = $closing->getLines();
1392                 wfProfileOut( $fname );
1393                 return $lines;
1394         }
1395 }
1396
1397 /**
1398  *      Wikipedia Table style diff formatter.
1399  * @todo document
1400  * @access private
1401  * @package MediaWiki
1402  * @subpackage DifferenceEngine
1403  */
1404 class TableDiffFormatter extends DiffFormatter
1405 {
1406         function TableDiffFormatter() {
1407                 $this->leading_context_lines = 2;
1408                 $this->trailing_context_lines = 2;
1409         }
1410
1411         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1412                 $l1 = wfMsg( 'lineno', $xbeg );
1413                 $l2 = wfMsg( 'lineno', $ybeg );
1414
1415                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1416                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1417                 return $r;
1418         }
1419
1420         function _start_block( $header ) {
1421                 global $wgOut;
1422                 echo $header;
1423         }
1424
1425         function _end_block() {
1426         }
1427
1428         function _lines( $lines, $prefix=' ', $color='white' ) {
1429         }
1430
1431         # HTML-escape parameter before calling this
1432         function addedLine( $line ) {
1433                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1434         }
1435
1436         # HTML-escape parameter before calling this
1437         function deletedLine( $line ) {
1438                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1439         }
1440
1441         # HTML-escape parameter before calling this
1442         function contextLine( $line ) {
1443                 return "<td> </td><td class='diff-context'>{$line}</td>";
1444         }
1445
1446         function emptyLine() {
1447                 return '<td colspan="2">&nbsp;</td>';
1448         }
1449
1450         function _added( $lines ) {
1451                 foreach ($lines as $line) {
1452                         echo '<tr>' . $this->emptyLine() .
1453                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1454                 }
1455         }
1456
1457         function _deleted($lines) {
1458                 foreach ($lines as $line) {
1459                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1460                           $this->emptyLine() . "</tr>\n";
1461                 }
1462         }
1463
1464         function _context( $lines ) {
1465                 foreach ($lines as $line) {
1466                         echo '<tr>' .
1467                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1468                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1469                 }
1470         }
1471
1472         function _changed( $orig, $closing ) {
1473                 $fname = 'TableDiffFormatter::_changed';
1474                 wfProfileIn( $fname );
1475
1476                 $diff = new WordLevelDiff( $orig, $closing );
1477                 $del = $diff->orig();
1478                 $add = $diff->closing();
1479
1480                 # Notice that WordLevelDiff returns HTML-escaped output.
1481                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1482
1483                 while ( $line = array_shift( $del ) ) {
1484                         $aline = array_shift( $add );
1485                         echo '<tr>' . $this->deletedLine( $line ) .
1486                                 $this->addedLine( $aline ) . "</tr>\n";
1487                 }
1488                 foreach ($add as $line) {       # If any leftovers
1489                         echo '<tr>' . $this->emptyLine() .
1490                                 $this->addedLine( $line ) . "</tr>\n";
1491                 }
1492                 wfProfileOut( $fname );
1493         }
1494 }
1495
1496 ?>