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         /* private */ var $mOldid, $mNewid;
  19         /* private */ var $mOldtitle, $mNewtitle, $mPagetitle;
  20         /* private */ var $mOldtext, $mNewtext;
  21         /* private */ var $mOldUser, $mNewUser;
  22         /* private */ var $mOldComment, $mNewComment;
  23         /* private */ var $mOldPage, $mNewPage;
  24         /* private */ var $mRcidMarkPatrolled;
  25
  26         function DifferenceEngine( $old, $new, $rcid = 0 )
  27         {
  28                 global $wgTitle;
  29                 if ( 'prev' == $new ) {
  30                         # Show diff between revision $old and the previous one.
  31                         # Get previous one from DB.
  32                         #
  33                         $this->mNewid = intval($old);
  34
  35                         $this->mOldid = $wgTitle->getPreviousRevisionID( $this->mNewid );
  36
  37                 } elseif ( 'next' == $new ) {
  38
  39                         # Show diff between revision $old and the previous one.
  40                         # Get previous one from DB.
  41                         #
  42                         $this->mOldid = intval($old);
  43                         $this->mNewid = $wgTitle->getNextRevisionID( $this->mOldid );
  44                         if ( false === $this->mNewid ) {
  45                                 # if no result, NewId points to the newest old revision. The only newer
  46                                 # revision is cur, which is "0".
  47                                 $this->mNewid = 0;
  48                         }
  49
  50                 } else {
  51
  52                         $this->mOldid = intval($old);
  53                         $this->mNewid = intval($new);
  54                 }
  55                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  56         }
  57
  58         function showDiffPage()
  59         {
  60                 global $wgUser, $wgTitle, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol, $wgUseRCPatrol;
  61                 $fname = 'DifferenceEngine::showDiffPage';
  62                 wfProfileIn( $fname );
  63
  64                 # mOldid is false if the difference engine is called with a "vague" query for
  65                 # a diff between a version V and its previous version V' AND the version V
  66                 # is the first version of that article. In that case, V' does not exist.
  67                 if ( $this->mOldid === false ) {
  68                         $this->showFirstRevision();
  69                         wfProfileOut( $fname );
  70                         return;
  71                 }
  72
  73                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  74                   "{$this->mNewid})";
  75                 $mtext = wfMsg( 'missingarticle', $t );
  76
  77                 $wgOut->setArticleFlag( false );
  78                 if ( ! $this->loadText() ) {
  79                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
  80                         $wgOut->addHTML( $mtext );
  81                         wfProfileOut( $fname );
  82                         return;
  83                 }
  84                 $wgOut->suppressQuickbar();
  85
  86                 $oldTitle = $this->mOldPage->getPrefixedText();
  87                 $newTitle = $this->mNewPage->getPrefixedText();
  88                 if( $oldTitle == $newTitle ) {
  89                         $wgOut->setPageTitle( $newTitle );
  90                 } else {
  91                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
  92                 }
  93                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
  94                 $wgOut->setRobotpolicy( 'noindex,follow' );
  95
  96                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  97                         $wgOut->loginToUse();
  98                         $wgOut->output();
  99                         wfProfileOut( $fname );
 100                         exit;
 101                 }
 102
 103                 $sk = $wgUser->getSkin();
 104                 $talk = $wgContLang->getNsText( NS_TALK );
 105                 $contribs = wfMsg( 'contribslink' );
 106
 107                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 108                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 109
 110                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 111                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 112                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 113                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 114                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 115                         'target=' . urlencode($this->mOldUser) );
 116                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 117                         'target=' . urlencode($this->mNewUser) );
 118                 if ( !$this->mNewid && $wgUser->isAllowed('rollback') ) {
 119                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
 120                                 'action=rollback&from=' . urlencode($this->mNewUser) ) . ']</strong>';
 121                 } else {
 122                         $rollback = '';
 123                 }
 124                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->getID() != 0 &&
 125                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 126                 {
 127                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 128                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 129                 } else {
 130                         $patrol = '';
 131                 }
 132
 133                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'previousdiff' ), 'diff=prev&oldid='.$this->mOldid );
 134                 if ( $this->mNewid == 0 ) {
 135                         $nextlink = '';
 136                 } else {
 137                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 138                 }
 139
 140                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 141                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 142                         '<br />' . $prevlink;
 143                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 144                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 145                         '<br />' . $nextlink . $patrol;
 146
 147                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 148                   $oldHeader, $newHeader );
 149                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 150                 $wgOut->addWikiText( $this->mNewtext );
 151
 152                 wfProfileOut( $fname );
 153         }
 154
 155         # Show the first revision of an article. Uses normal diff headers in contrast to normal
 156         # "old revision" display style.
 157         #
 158         function showFirstRevision()
 159         {
 160                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 161
 162                 $fname = 'DifferenceEngine::showFirstRevision';
 163                 wfProfileIn( $fname );
 164
 165
 166                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 167
 168                 # Get article text from the DB
 169                 #
 170                 if ( ! $this->loadText() ) {
 171                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 172                           "{$this->mNewid})";
 173                         $mtext = wfMsg( 'missingarticle', $t );
 174                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 175                         $wgOut->addHTML( $mtext );
 176                         wfProfileOut( $fname );
 177                         return;
 178                 }
 179
 180                 # Check if user is allowed to look at this page. If not, bail out.
 181                 #
 182                 if ( !( $this->mOldPage->userCanRead() ) ) {
 183                         $wgOut->loginToUse();
 184                         $wgOut->output();
 185                         wfProfileOut( $fname );
 186                         exit;
 187                 }
 188
 189                 # Prepare the header box
 190                 #
 191                 $sk = $wgUser->getSkin();
 192
 193                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 194                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 195                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 196                         'target=' . urlencode($this->mOldUser) );
 197                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 198                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 199                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 200                         '<br />' . $nextlink. "</div>\n";
 201
 202                 $wgOut->addHTML( $header );
 203
 204                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 205                 $wgOut->setRobotpolicy( 'noindex,follow' );
 206
 207
 208                 # Show current revision
 209                 #
 210                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 211                 $wgOut->addWikiText( $this->mNewtext );
 212
 213                 wfProfileOut( $fname );
 214         }
 215
 216         function showDiff( $otext, $ntext, $otitle, $ntitle )
 217         {
 218                 global $wgOut;
 219                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 220         }
 221
 222         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 223                 global $wgUseExternalDiffEngine, $wgContLang;
 224                 $out = "
 225                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 226                         <tr>
 227                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 228                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 229                         </tr>
 230                 ";
 231                 $otext = $wgContLang->segmentForDiff($otext);
 232                 $ntext = $wgContLang->segmentForDiff($ntext);
 233                 $difftext='';
 234                 if ( $wgUseExternalDiffEngine ) {
 235                         # For historical reasons, external diff engine expects
 236                         # input text to be HTML-escaped already
 237                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 238                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 239                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 240                                 dl('php_wikidiff.so');
 241                         }
 242                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 243                 } else {
 244                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 245                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 246                         $diffs =& new Diff( $ota, $nta );
 247                         $formatter =& new TableDiffFormatter();
 248                         $difftext = $formatter->format( $diffs );
 249                 }
 250                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 251                 $out .= $difftext."</table>\n";
 252                 return $out;
 253         }
 254
 255         # Load the text of the articles to compare.  If newid is 0, then compare
 256         # the old article in oldid to the current article; if oldid is 0, then
 257         # compare the current article to the immediately previous one (ignoring
 258         # the value of newid).
 259         #
 260         function loadText()
 261         {
 262                 global $wgTitle, $wgOut, $wgLang;
 263                 $fname = 'DifferenceEngine::loadText';
 264
 265                 $dbr =& wfGetDB( DB_SLAVE );
 266                 if( $this->mNewid ) {
 267                         $this->newRev =& Revision::newFromId( $this->mNewid );
 268                 } else {
 269                         $this->newRev =& Revision::newFromTitle( $wgTitle );
 270                 }
 271
 272                 if( $this->newRev->isCurrent() ) {
 273                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 274                         $this->mNewPage = $wgTitle;
 275                         $newLink = $this->mNewPage->escapeLocalUrl();
 276                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 277                 } else {
 278                         $this->mNewPage = $this->newRev->getTitle();
 279                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 280                         $t = $wgLang->timeanddate( $this->newRev->getTimestamp(), true );
 281                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 282                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 283                 }
 284
 285                 if( $this->mOldid ) {
 286                         $this->oldRev =& Revision::newFromId( $this->mOldid );
 287                 } else {
 288                         $this->oldRev =& $this->newRev->getPrevious();
 289                         $this->mOldid = $this->oldRev->getId();
 290                 }
 291
 292                 $this->mOldPage = $this->oldRev->getTitle();
 293
 294                 $t = $wgLang->timeanddate( $this->oldRev->getTimestamp(), true );
 295                 $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 296                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 297
 298                 $this->mNewUser = $this->newRev->getUserText();
 299                 $this->mNewComment = $this->newRev->getComment();
 300                 $this->mNewtext = $this->newRev->getText();
 301
 302                 $this->mOldUser = $this->oldRev->getUserText();
 303                 $this->mOldComment = $this->oldRev->getComment();
 304                 $this->mOldtext = $this->oldRev->getText();
 305
 306                 return true;
 307         }
 308 }
 309
 310 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 311 //
 312 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 313 // You may copy this code freely under the conditions of the GPL.
 314 //
 315
 316 define('USE_ASSERTS', function_exists('assert'));
 317
 318 /**
 319  * @todo document
 320  * @access private
 321  * @package MediaWiki
 322  * @subpackage DifferenceEngine
 323  */
 324 class _DiffOp {
 325         var $type;
 326         var $orig;
 327         var $closing;
 328
 329         function reverse() {
 330                 trigger_error('pure virtual', E_USER_ERROR);
 331         }
 332
 333         function norig() {
 334                 return $this->orig ? sizeof($this->orig) : 0;
 335         }
 336
 337         function nclosing() {
 338                 return $this->closing ? sizeof($this->closing) : 0;
 339         }
 340 }
 341
 342 /**
 343  * @todo document
 344  * @access private
 345  * @package MediaWiki
 346  * @subpackage DifferenceEngine
 347  */
 348 class _DiffOp_Copy extends _DiffOp {
 349         var $type = 'copy';
 350
 351         function _DiffOp_Copy ($orig, $closing = false) {
 352                 if (!is_array($closing))
 353                         $closing = $orig;
 354                 $this->orig = $orig;
 355                 $this->closing = $closing;
 356         }
 357
 358         function reverse() {
 359                 return new _DiffOp_Copy($this->closing, $this->orig);
 360         }
 361 }
 362
 363 /**
 364  * @todo document
 365  * @access private
 366  * @package MediaWiki
 367  * @subpackage DifferenceEngine
 368  */
 369 class _DiffOp_Delete extends _DiffOp {
 370         var $type = 'delete';
 371
 372         function _DiffOp_Delete ($lines) {
 373                 $this->orig = $lines;
 374                 $this->closing = false;
 375         }
 376
 377         function reverse() {
 378                 return new _DiffOp_Add($this->orig);
 379         }
 380 }
 381
 382 /**
 383  * @todo document
 384  * @access private
 385  * @package MediaWiki
 386  * @subpackage DifferenceEngine
 387  */
 388 class _DiffOp_Add extends _DiffOp {
 389         var $type = 'add';
 390
 391         function _DiffOp_Add ($lines) {
 392                 $this->closing = $lines;
 393                 $this->orig = false;
 394         }
 395
 396         function reverse() {
 397                 return new _DiffOp_Delete($this->closing);
 398         }
 399 }
 400
 401 /**
 402  * @todo document
 403  * @access private
 404  * @package MediaWiki
 405  * @subpackage DifferenceEngine
 406  */
 407 class _DiffOp_Change extends _DiffOp {
 408         var $type = 'change';
 409
 410         function _DiffOp_Change ($orig, $closing) {
 411                 $this->orig = $orig;
 412                 $this->closing = $closing;
 413         }
 414
 415         function reverse() {
 416                 return new _DiffOp_Change($this->closing, $this->orig);
 417         }
 418 }
 419
 420
 421 /**
 422  * Class used internally by Diff to actually compute the diffs.
 423  *
 424  * The algorithm used here is mostly lifted from the perl module
 425  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 426  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 427  *
 428  * More ideas are taken from:
 429  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 430  *
 431  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 432  * diffutils-2.7, which can be found at:
 433  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 434  *
 435  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 436  * are my own.
 437  *
 438  * @author Geoffrey T. Dairiki
 439  * @access private
 440  * @package MediaWiki
 441  * @subpackage DifferenceEngine
 442  */
 443 class _DiffEngine
 444 {
 445         function diff ($from_lines, $to_lines) {
 446                 $fname = '_DiffEngine::diff';
 447                 wfProfileIn( $fname );
 448
 449                 $n_from = sizeof($from_lines);
 450                 $n_to = sizeof($to_lines);
 451
 452                 $this->xchanged = $this->ychanged = array();
 453                 $this->xv = $this->yv = array();
 454                 $this->xind = $this->yind = array();
 455                 unset($this->seq);
 456                 unset($this->in_seq);
 457                 unset($this->lcs);
 458
 459                 // Skip leading common lines.
 460                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 461                         if ($from_lines[$skip] != $to_lines[$skip])
 462                                 break;
 463                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 464                 }
 465                 // Skip trailing common lines.
 466                 $xi = $n_from; $yi = $n_to;
 467                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 468                         if ($from_lines[$xi] != $to_lines[$yi])
 469                                 break;
 470                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 471                 }
 472
 473                 // Ignore lines which do not exist in both files.
 474                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 475                         $xhash[$from_lines[$xi]] = 1;
 476                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 477                         $line = $to_lines[$yi];
 478                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 479                                 continue;
 480                         $yhash[$line] = 1;
 481                         $this->yv[] = $line;
 482                         $this->yind[] = $yi;
 483                 }
 484                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 485                         $line = $from_lines[$xi];
 486                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 487                                 continue;
 488                         $this->xv[] = $line;
 489                         $this->xind[] = $xi;
 490                 }
 491
 492                 // Find the LCS.
 493                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 494
 495                 // Merge edits when possible
 496                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 497                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 498
 499                 // Compute the edit operations.
 500                 $edits = array();
 501                 $xi = $yi = 0;
 502                 while ($xi < $n_from || $yi < $n_to) {
 503                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 504                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 505
 506                         // Skip matching "snake".
 507                         $copy = array();
 508                         while ( $xi < $n_from && $yi < $n_to
 509                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 510                                 $copy[] = $from_lines[$xi++];
 511                                 ++$yi;
 512                         }
 513                         if ($copy)
 514                                 $edits[] = new _DiffOp_Copy($copy);
 515
 516                         // Find deletes & adds.
 517                         $delete = array();
 518                         while ($xi < $n_from && $this->xchanged[$xi])
 519                                 $delete[] = $from_lines[$xi++];
 520
 521                         $add = array();
 522                         while ($yi < $n_to && $this->ychanged[$yi])
 523                                 $add[] = $to_lines[$yi++];
 524
 525                         if ($delete && $add)
 526                                 $edits[] = new _DiffOp_Change($delete, $add);
 527                         elseif ($delete)
 528                                 $edits[] = new _DiffOp_Delete($delete);
 529                         elseif ($add)
 530                                 $edits[] = new _DiffOp_Add($add);
 531                 }
 532                 wfProfileOut( $fname );
 533                 return $edits;
 534         }
 535
 536
 537         /* Divide the Largest Common Subsequence (LCS) of the sequences
 538          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 539          * sized segments.
 540          *
 541          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 542          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 543          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 544          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 545          * that (X0, Y0) == (XOFF, YOFF) and
 546          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 547          *
 548          * This function assumes that the first lines of the specified portions
 549          * of the two files do not match, and likewise that the last lines do not
 550          * match.  The caller must trim matching lines from the beginning and end
 551          * of the portions it is going to specify.
 552          */
 553         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 554                 $fname = '_DiffEngine::_diag';
 555                 wfProfileIn( $fname );
 556                 $flip = false;
 557
 558                 if ($xlim - $xoff > $ylim - $yoff) {
 559                         // Things seems faster (I'm not sure I understand why)
 560                                 // when the shortest sequence in X.
 561                                 $flip = true;
 562                         list ($xoff, $xlim, $yoff, $ylim)
 563                         = array( $yoff, $ylim, $xoff, $xlim);
 564                 }
 565
 566                 if ($flip)
 567                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 568                                 $ymatches[$this->xv[$i]][] = $i;
 569                 else
 570                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 571                                 $ymatches[$this->yv[$i]][] = $i;
 572
 573                 $this->lcs = 0;
 574                 $this->seq[0]= $yoff - 1;
 575                 $this->in_seq = array();
 576                 $ymids[0] = array();
 577
 578                 $numer = $xlim - $xoff + $nchunks - 1;
 579                 $x = $xoff;
 580                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 581                         wfProfileIn( "$fname-chunk" );
 582                         if ($chunk > 0)
 583                                 for ($i = 0; $i <= $this->lcs; $i++)
 584                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 585
 586                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 587                         for ( ; $x < $x1; $x++) {
 588                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 589                                         if (empty($ymatches[$line]))
 590                                                 continue;
 591                                 $matches = $ymatches[$line];
 592                                 reset($matches);
 593                                 while (list ($junk, $y) = each($matches))
 594                                         if (empty($this->in_seq[$y])) {
 595                                                 $k = $this->_lcs_pos($y);
 596                                                 USE_ASSERTS && assert($k > 0);
 597                                                 $ymids[$k] = $ymids[$k-1];
 598                                                 break;
 599                                         }
 600                                 while (list ($junk, $y) = each($matches)) {
 601                                         if ($y > $this->seq[$k-1]) {
 602                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 603                                                 // Optimization: this is a common case:
 604                                                 //      next match is just replacing previous match.
 605                                                 $this->in_seq[$this->seq[$k]] = false;
 606                                                 $this->seq[$k] = $y;
 607                                                 $this->in_seq[$y] = 1;
 608                                         } else if (empty($this->in_seq[$y])) {
 609                                                 $k = $this->_lcs_pos($y);
 610                                                 USE_ASSERTS && assert($k > 0);
 611                                                 $ymids[$k] = $ymids[$k-1];
 612                                         }
 613                                 }
 614                         }
 615                         wfProfileOut( "$fname-chunk" );
 616                 }
 617
 618                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 619                 $ymid = $ymids[$this->lcs];
 620                 for ($n = 0; $n < $nchunks - 1; $n++) {
 621                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 622                         $y1 = $ymid[$n] + 1;
 623                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 624                 }
 625                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 626
 627                 wfProfileOut( $fname );
 628                 return array($this->lcs, $seps);
 629         }
 630
 631         function _lcs_pos ($ypos) {
 632                 $fname = '_DiffEngine::_lcs_pos';
 633                 wfProfileIn( $fname );
 634
 635                 $end = $this->lcs;
 636                 if ($end == 0 || $ypos > $this->seq[$end]) {
 637                         $this->seq[++$this->lcs] = $ypos;
 638                         $this->in_seq[$ypos] = 1;
 639                         wfProfileOut( $fname );
 640                         return $this->lcs;
 641                 }
 642
 643                 $beg = 1;
 644                 while ($beg < $end) {
 645                         $mid = (int)(($beg + $end) / 2);
 646                         if ( $ypos > $this->seq[$mid] )
 647                                 $beg = $mid + 1;
 648                         else
 649                                 $end = $mid;
 650                 }
 651
 652                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 653
 654                 $this->in_seq[$this->seq[$end]] = false;
 655                 $this->seq[$end] = $ypos;
 656                 $this->in_seq[$ypos] = 1;
 657                 wfProfileOut( $fname );
 658                 return $end;
 659         }
 660
 661         /* Find LCS of two sequences.
 662          *
 663          * The results are recorded in the vectors $this->{x,y}changed[], by
 664          * storing a 1 in the element for each line that is an insertion
 665          * or deletion (ie. is not in the LCS).
 666          *
 667          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 668          *
 669          * Note that XLIM, YLIM are exclusive bounds.
 670          * All line numbers are origin-0 and discarded lines are not counted.
 671          */
 672         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 673                 $fname = '_DiffEngine::_compareseq';
 674                 wfProfileIn( $fname );
 675
 676                 // Slide down the bottom initial diagonal.
 677                 while ($xoff < $xlim && $yoff < $ylim
 678                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 679                         ++$xoff;
 680                         ++$yoff;
 681                 }
 682
 683                 // Slide up the top initial diagonal.
 684                 while ($xlim > $xoff && $ylim > $yoff
 685                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 686                         --$xlim;
 687                         --$ylim;
 688                 }
 689
 690                 if ($xoff == $xlim || $yoff == $ylim)
 691                         $lcs = 0;
 692                 else {
 693                         // This is ad hoc but seems to work well.
 694                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 695                         //$nchunks = max(2,min(8,(int)$nchunks));
 696                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 697                         list ($lcs, $seps)
 698                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 699                 }
 700
 701                 if ($lcs == 0) {
 702                         // X and Y sequences have no common subsequence:
 703                         // mark all changed.
 704                         while ($yoff < $ylim)
 705                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 706                         while ($xoff < $xlim)
 707                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 708                 } else {
 709                         // Use the partitions to split this problem into subproblems.
 710                         reset($seps);
 711                         $pt1 = $seps[0];
 712                         while ($pt2 = next($seps)) {
 713                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 714                                 $pt1 = $pt2;
 715                         }
 716                 }
 717                 wfProfileOut( $fname );
 718         }
 719
 720         /* Adjust inserts/deletes of identical lines to join changes
 721          * as much as possible.
 722          *
 723          * We do something when a run of changed lines include a
 724          * line at one end and has an excluded, identical line at the other.
 725          * We are free to choose which identical line is included.
 726          * `compareseq' usually chooses the one at the beginning,
 727          * but usually it is cleaner to consider the following identical line
 728          * to be the "change".
 729          *
 730          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 731          */
 732         function _shift_boundaries ($lines, &$changed, $other_changed) {
 733                 $fname = '_DiffEngine::_shift_boundaries';
 734                 wfProfileIn( $fname );
 735                 $i = 0;
 736                 $j = 0;
 737
 738                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 739                 $len = sizeof($lines);
 740                 $other_len = sizeof($other_changed);
 741
 742                 while (1) {
 743                         /*
 744                          * Scan forwards to find beginning of another run of changes.
 745                          * Also keep track of the corresponding point in the other file.
 746                          *
 747                          * Throughout this code, $i and $j are adjusted together so that
 748                          * the first $i elements of $changed and the first $j elements
 749                          * of $other_changed both contain the same number of zeros
 750                          * (unchanged lines).
 751                          * Furthermore, $j is always kept so that $j == $other_len or
 752                          * $other_changed[$j] == false.
 753                          */
 754                         while ($j < $other_len && $other_changed[$j])
 755                                 $j++;
 756
 757                         while ($i < $len && ! $changed[$i]) {
 758                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 759                                 $i++; $j++;
 760                                 while ($j < $other_len && $other_changed[$j])
 761                                         $j++;
 762                         }
 763
 764                         if ($i == $len)
 765                                 break;
 766
 767                         $start = $i;
 768
 769                         // Find the end of this run of changes.
 770                         while (++$i < $len && $changed[$i])
 771                                 continue;
 772
 773                         do {
 774                                 /*
 775                                  * Record the length of this run of changes, so that
 776                                  * we can later determine whether the run has grown.
 777                                  */
 778                                 $runlength = $i - $start;
 779
 780                                 /*
 781                                  * Move the changed region back, so long as the
 782                                  * previous unchanged line matches the last changed one.
 783                                  * This merges with previous changed regions.
 784                                  */
 785                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 786                                         $changed[--$start] = 1;
 787                                         $changed[--$i] = false;
 788                                         while ($start > 0 && $changed[$start - 1])
 789                                                 $start--;
 790                                         USE_ASSERTS && assert('$j > 0');
 791                                         while ($other_changed[--$j])
 792                                                 continue;
 793                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 794                                 }
 795
 796                                 /*
 797                                  * Set CORRESPONDING to the end of the changed run, at the last
 798                                  * point where it corresponds to a changed run in the other file.
 799                                  * CORRESPONDING == LEN means no such point has been found.
 800                                  */
 801                                 $corresponding = $j < $other_len ? $i : $len;
 802
 803                                 /*
 804                                  * Move the changed region forward, so long as the
 805                                  * first changed line matches the following unchanged one.
 806                                  * This merges with following changed regions.
 807                                  * Do this second, so that if there are no merges,
 808                                  * the changed region is moved forward as far as possible.
 809                                  */
 810                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 811                                         $changed[$start++] = false;
 812                                         $changed[$i++] = 1;
 813                                         while ($i < $len && $changed[$i])
 814                                                 $i++;
 815
 816                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 817                                         $j++;
 818                                         if ($j < $other_len && $other_changed[$j]) {
 819                                                 $corresponding = $i;
 820                                                 while ($j < $other_len && $other_changed[$j])
 821                                                         $j++;
 822                                         }
 823                                 }
 824                         } while ($runlength != $i - $start);
 825
 826                         /*
 827                          * If possible, move the fully-merged run of changes
 828                          * back to a corresponding run in the other file.
 829                          */
 830                         while ($corresponding < $i) {
 831                                 $changed[--$start] = 1;
 832                                 $changed[--$i] = 0;
 833                                 USE_ASSERTS && assert('$j > 0');
 834                                 while ($other_changed[--$j])
 835                                         continue;
 836                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 837                         }
 838                 }
 839                 wfProfileOut( $fname );
 840         }
 841 }
 842
 843 /**
 844  * Class representing a 'diff' between two sequences of strings.
 845  * @todo document
 846  * @access private
 847  * @package MediaWiki
 848  * @subpackage DifferenceEngine
 849  */
 850 class Diff
 851 {
 852         var $edits;
 853
 854         /**
 855          * Constructor.
 856          * Computes diff between sequences of strings.
 857          *
 858          * @param $from_lines array An array of strings.
 859          *                (Typically these are lines from a file.)
 860          * @param $to_lines array An array of strings.
 861          */
 862         function Diff($from_lines, $to_lines) {
 863                 $eng = new _DiffEngine;
 864                 $this->edits = $eng->diff($from_lines, $to_lines);
 865                 //$this->_check($from_lines, $to_lines);
 866         }
 867
 868         /**
 869          * Compute reversed Diff.
 870          *
 871          * SYNOPSIS:
 872          *
 873          *      $diff = new Diff($lines1, $lines2);
 874          *      $rev = $diff->reverse();
 875          * @return object A Diff object representing the inverse of the
 876          *                                original diff.
 877          */
 878         function reverse () {
 879                 $rev = $this;
 880                 $rev->edits = array();
 881                 foreach ($this->edits as $edit) {
 882                         $rev->edits[] = $edit->reverse();
 883                 }
 884                 return $rev;
 885         }
 886
 887         /**
 888          * Check for empty diff.
 889          *
 890          * @return bool True iff two sequences were identical.
 891          */
 892         function isEmpty () {
 893                 foreach ($this->edits as $edit) {
 894                         if ($edit->type != 'copy')
 895                                 return false;
 896                 }
 897                 return true;
 898         }
 899
 900         /**
 901          * Compute the length of the Longest Common Subsequence (LCS).
 902          *
 903          * This is mostly for diagnostic purposed.
 904          *
 905          * @return int The length of the LCS.
 906          */
 907         function lcs () {
 908                 $lcs = 0;
 909                 foreach ($this->edits as $edit) {
 910                         if ($edit->type == 'copy')
 911                                 $lcs += sizeof($edit->orig);
 912                 }
 913                 return $lcs;
 914         }
 915
 916         /**
 917          * Get the original set of lines.
 918          *
 919          * This reconstructs the $from_lines parameter passed to the
 920          * constructor.
 921          *
 922          * @return array The original sequence of strings.
 923          */
 924         function orig() {
 925                 $lines = array();
 926
 927                 foreach ($this->edits as $edit) {
 928                         if ($edit->orig)
 929                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 930                 }
 931                 return $lines;
 932         }
 933
 934         /**
 935          * Get the closing set of lines.
 936          *
 937          * This reconstructs the $to_lines parameter passed to the
 938          * constructor.
 939          *
 940          * @return array The sequence of strings.
 941          */
 942         function closing() {
 943                 $lines = array();
 944
 945                 foreach ($this->edits as $edit) {
 946                         if ($edit->closing)
 947                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 948                 }
 949                 return $lines;
 950         }
 951
 952         /**
 953          * Check a Diff for validity.
 954          *
 955          * This is here only for debugging purposes.
 956          */
 957         function _check ($from_lines, $to_lines) {
 958                 $fname = 'Diff::_check';
 959                 wfProfileIn( $fname );
 960                 if (serialize($from_lines) != serialize($this->orig()))
 961                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 962                 if (serialize($to_lines) != serialize($this->closing()))
 963                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 964
 965                 $rev = $this->reverse();
 966                 if (serialize($to_lines) != serialize($rev->orig()))
 967                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 968                 if (serialize($from_lines) != serialize($rev->closing()))
 969                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 970
 971
 972                 $prevtype = 'none';
 973                 foreach ($this->edits as $edit) {
 974                         if ( $prevtype == $edit->type )
 975                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 976                         $prevtype = $edit->type;
 977                 }
 978
 979                 $lcs = $this->lcs();
 980                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 981                 wfProfileOut( $fname );
 982         }
 983 }
 984
 985 /**
 986  * FIXME: bad name.
 987  * @todo document
 988  * @access private
 989  * @package MediaWiki
 990  * @subpackage DifferenceEngine
 991  */
 992 class MappedDiff extends Diff
 993 {
 994         /**
 995          * Constructor.
 996          *
 997          * Computes diff between sequences of strings.
 998          *
 999          * This can be used to compute things like
1000          * case-insensitve diffs, or diffs which ignore
1001          * changes in white-space.
1002          *
1003          * @param $from_lines array An array of strings.
1004          *      (Typically these are lines from a file.)
1005          *
1006          * @param $to_lines array An array of strings.
1007          *
1008          * @param $mapped_from_lines array This array should
1009          *      have the same size number of elements as $from_lines.
1010          *      The elements in $mapped_from_lines and
1011          *      $mapped_to_lines are what is actually compared
1012          *      when computing the diff.
1013          *
1014          * @param $mapped_to_lines array This array should
1015          *      have the same number of elements as $to_lines.
1016          */
1017         function MappedDiff($from_lines, $to_lines,
1018                                                 $mapped_from_lines, $mapped_to_lines) {
1019                 $fname = 'MappedDiff::MappedDiff';
1020                 wfProfileIn( $fname );
1021
1022                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1023                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1024
1025                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1026
1027                 $xi = $yi = 0;
1028                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1029                         $orig = &$this->edits[$i]->orig;
1030                         if (is_array($orig)) {
1031                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1032                                 $xi += sizeof($orig);
1033                         }
1034
1035                         $closing = &$this->edits[$i]->closing;
1036                         if (is_array($closing)) {
1037                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1038                                 $yi += sizeof($closing);
1039                         }
1040                 }
1041                 wfProfileOut( $fname );
1042         }
1043 }
1044
1045 /**
1046  * A class to format Diffs
1047  *
1048  * This class formats the diff in classic diff format.
1049  * It is intended that this class be customized via inheritance,
1050  * to obtain fancier outputs.
1051  * @todo document
1052  * @access private
1053  * @package MediaWiki
1054  * @subpackage DifferenceEngine
1055  */
1056 class DiffFormatter
1057 {
1058         /**
1059          * Number of leading context "lines" to preserve.
1060          *
1061          * This should be left at zero for this class, but subclasses
1062          * may want to set this to other values.
1063          */
1064         var $leading_context_lines = 0;
1065
1066         /**
1067          * Number of trailing context "lines" to preserve.
1068          *
1069          * This should be left at zero for this class, but subclasses
1070          * may want to set this to other values.
1071          */
1072         var $trailing_context_lines = 0;
1073
1074         /**
1075          * Format a diff.
1076          *
1077          * @param $diff object A Diff object.
1078          * @return string The formatted output.
1079          */
1080         function format($diff) {
1081                 $fname = 'DiffFormatter::format';
1082                 wfProfileIn( $fname );
1083
1084                 $xi = $yi = 1;
1085                 $block = false;
1086                 $context = array();
1087
1088                 $nlead = $this->leading_context_lines;
1089                 $ntrail = $this->trailing_context_lines;
1090
1091                 $this->_start_diff();
1092
1093                 foreach ($diff->edits as $edit) {
1094                         if ($edit->type == 'copy') {
1095                                 if (is_array($block)) {
1096                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1097                                                 $block[] = $edit;
1098                                         }
1099                                         else{
1100                                                 if ($ntrail) {
1101                                                         $context = array_slice($edit->orig, 0, $ntrail);
1102                                                         $block[] = new _DiffOp_Copy($context);
1103                                                 }
1104                                                 $this->_block($x0, $ntrail + $xi - $x0,
1105                                                                           $y0, $ntrail + $yi - $y0,
1106                                                                           $block);
1107                                                 $block = false;
1108                                         }
1109                                 }
1110                                 $context = $edit->orig;
1111                         }
1112                         else {
1113                                 if (! is_array($block)) {
1114                                         $context = array_slice($context, sizeof($context) - $nlead);
1115                                         $x0 = $xi - sizeof($context);
1116                                         $y0 = $yi - sizeof($context);
1117                                         $block = array();
1118                                         if ($context)
1119                                                 $block[] = new _DiffOp_Copy($context);
1120                                 }
1121                                 $block[] = $edit;
1122                         }
1123
1124                         if ($edit->orig)
1125                                 $xi += sizeof($edit->orig);
1126                         if ($edit->closing)
1127                                 $yi += sizeof($edit->closing);
1128                 }
1129
1130                 if (is_array($block))
1131                         $this->_block($x0, $xi - $x0,
1132                                                   $y0, $yi - $y0,
1133                                                   $block);
1134
1135                 $end = $this->_end_diff();
1136                 wfProfileOut( $fname );
1137                 return $end;
1138         }
1139
1140         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1141                 $fname = 'DiffFormatter::_block';
1142                 wfProfileIn( $fname );
1143                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1144                 foreach ($edits as $edit) {
1145                         if ($edit->type == 'copy')
1146                                 $this->_context($edit->orig);
1147                         elseif ($edit->type == 'add')
1148                                 $this->_added($edit->closing);
1149                         elseif ($edit->type == 'delete')
1150                                 $this->_deleted($edit->orig);
1151                         elseif ($edit->type == 'change')
1152                                 $this->_changed($edit->orig, $edit->closing);
1153                         else
1154                                 trigger_error('Unknown edit type', E_USER_ERROR);
1155                 }
1156                 $this->_end_block();
1157                 wfProfileOut( $fname );
1158         }
1159
1160         function _start_diff() {
1161                 ob_start();
1162         }
1163
1164         function _end_diff() {
1165                 $val = ob_get_contents();
1166                 ob_end_clean();
1167                 return $val;
1168         }
1169
1170         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1171                 if ($xlen > 1)
1172                         $xbeg .= "," . ($xbeg + $xlen - 1);
1173                 if ($ylen > 1)
1174                         $ybeg .= "," . ($ybeg + $ylen - 1);
1175
1176                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1177         }
1178
1179         function _start_block($header) {
1180                 echo $header;
1181         }
1182
1183         function _end_block() {
1184         }
1185
1186         function _lines($lines, $prefix = ' ') {
1187                 foreach ($lines as $line)
1188                         echo "$prefix $line\n";
1189         }
1190
1191         function _context($lines) {
1192                 $this->_lines($lines);
1193         }
1194
1195         function _added($lines) {
1196                 $this->_lines($lines, '>');
1197         }
1198         function _deleted($lines) {
1199                 $this->_lines($lines, '<');
1200         }
1201
1202         function _changed($orig, $closing) {
1203                 $this->_deleted($orig);
1204                 echo "---\n";
1205                 $this->_added($closing);
1206         }
1207 }
1208
1209
1210 /**
1211  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1212  *
1213  */
1214
1215 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1216
1217 /**
1218  * @todo document
1219  * @access private
1220  * @package MediaWiki
1221  * @subpackage DifferenceEngine
1222  */
1223 class _HWLDF_WordAccumulator {
1224         function _HWLDF_WordAccumulator () {
1225                 $this->_lines = array();
1226                 $this->_line = '';
1227                 $this->_group = '';
1228                 $this->_tag = '';
1229         }
1230
1231         function _flushGroup ($new_tag) {
1232                 if ($this->_group !== '') {
1233                         if ($this->_tag == 'mark')
1234                                 $this->_line .= '<span class="diffchange">' .
1235                                         htmlspecialchars ( $this->_group ) . '</span>';
1236                         else
1237                                 $this->_line .= htmlspecialchars ( $this->_group );
1238                 }
1239                 $this->_group = '';
1240                 $this->_tag = $new_tag;
1241         }
1242
1243         function _flushLine ($new_tag) {
1244                 $this->_flushGroup($new_tag);
1245                 if ($this->_line != '')
1246                         array_push ( $this->_lines, $this->_line );
1247                 else
1248                         # make empty lines visible by inserting an NBSP
1249                         array_push ( $this->_lines, NBSP );
1250                 $this->_line = '';
1251         }
1252
1253         function addWords ($words, $tag = '') {
1254                 if ($tag != $this->_tag)
1255                         $this->_flushGroup($tag);
1256
1257                 foreach ($words as $word) {
1258                         // new-line should only come as first char of word.
1259                         if ($word == '')
1260                                 continue;
1261                         if ($word[0] == "\n") {
1262                                 $this->_flushLine($tag);
1263                                 $word = substr($word, 1);
1264                         }
1265                         assert(!strstr($word, "\n"));
1266                         $this->_group .= $word;
1267                 }
1268         }
1269
1270         function getLines() {
1271                 $this->_flushLine('~done');
1272                 return $this->_lines;
1273         }
1274 }
1275
1276 /**
1277  * @todo document
1278  * @access private
1279  * @package MediaWiki
1280  * @subpackage DifferenceEngine
1281  */
1282 class WordLevelDiff extends MappedDiff
1283 {
1284         function WordLevelDiff ($orig_lines, $closing_lines) {
1285                 $fname = 'WordLevelDiff::WordLevelDiff';
1286                 wfProfileIn( $fname );
1287
1288                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1289                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1290
1291
1292                 $this->MappedDiff($orig_words, $closing_words,
1293                                                   $orig_stripped, $closing_stripped);
1294                 wfProfileOut( $fname );
1295         }
1296
1297         function _split($lines) {
1298                 $fname = 'WordLevelDiff::_split';
1299                 wfProfileIn( $fname );
1300                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1301                                                         implode("\n", $lines),
1302                                                         $m)) {
1303                         wfProfileOut( $fname );
1304                         return array(array(''), array(''));
1305                 }
1306                 wfProfileOut( $fname );
1307                 return array($m[0], $m[1]);
1308         }
1309
1310         function orig () {
1311                 $fname = 'WordLevelDiff::orig';
1312                 wfProfileIn( $fname );
1313                 $orig = new _HWLDF_WordAccumulator;
1314
1315                 foreach ($this->edits as $edit) {
1316                         if ($edit->type == 'copy')
1317                                 $orig->addWords($edit->orig);
1318                         elseif ($edit->orig)
1319                                 $orig->addWords($edit->orig, 'mark');
1320                 }
1321                 $lines = $orig->getLines();
1322                 wfProfileOut( $fname );
1323                 return $lines;
1324         }
1325
1326         function closing () {
1327                 $fname = 'WordLevelDiff::closing';
1328                 wfProfileIn( $fname );
1329                 $closing = new _HWLDF_WordAccumulator;
1330
1331                 foreach ($this->edits as $edit) {
1332                         if ($edit->type == 'copy')
1333                                 $closing->addWords($edit->closing);
1334                         elseif ($edit->closing)
1335                                 $closing->addWords($edit->closing, 'mark');
1336                 }
1337                 $lines = $closing->getLines();
1338                 wfProfileOut( $fname );
1339                 return $lines;
1340         }
1341 }
1342
1343 /**
1344  *      Wikipedia Table style diff formatter.
1345  * @todo document
1346  * @access private
1347  * @package MediaWiki
1348  * @subpackage DifferenceEngine
1349  */
1350 class TableDiffFormatter extends DiffFormatter
1351 {
1352         function TableDiffFormatter() {
1353                 $this->leading_context_lines = 2;
1354                 $this->trailing_context_lines = 2;
1355         }
1356
1357         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1358                 $l1 = wfMsg( 'lineno', $xbeg );
1359                 $l2 = wfMsg( 'lineno', $ybeg );
1360
1361                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1362                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1363                 return $r;
1364         }
1365
1366         function _start_block( $header ) {
1367                 global $wgOut;
1368                 echo $header;
1369         }
1370
1371         function _end_block() {
1372         }
1373
1374         function _lines( $lines, $prefix=' ', $color='white' ) {
1375         }
1376
1377         # HTML-escape parameter before calling this
1378         function addedLine( $line ) {
1379                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1380         }
1381
1382         # HTML-escape parameter before calling this
1383         function deletedLine( $line ) {
1384                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1385         }
1386
1387         # HTML-escape parameter before calling this
1388         function contextLine( $line ) {
1389                 return "<td> </td><td class='diff-context'>{$line}</td>";
1390         }
1391
1392         function emptyLine() {
1393                 return '<td colspan="2">&nbsp;</td>';
1394         }
1395
1396         function _added( $lines ) {
1397                 foreach ($lines as $line) {
1398                         echo '<tr>' . $this->emptyLine() .
1399                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1400                 }
1401         }
1402
1403         function _deleted($lines) {
1404                 foreach ($lines as $line) {
1405                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1406                           $this->emptyLine() . "</tr>\n";
1407                 }
1408         }
1409
1410         function _context( $lines ) {
1411                 foreach ($lines as $line) {
1412                         echo '<tr>' .
1413                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1414                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1415                 }
1416         }
1417
1418         function _changed( $orig, $closing ) {
1419                 $fname = 'TableDiffFormatter::_changed';
1420                 wfProfileIn( $fname );
1421
1422                 $diff = new WordLevelDiff( $orig, $closing );
1423                 $del = $diff->orig();
1424                 $add = $diff->closing();
1425
1426                 # Notice that WordLevelDiff returns HTML-escaped output.
1427                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1428
1429                 while ( $line = array_shift( $del ) ) {
1430                         $aline = array_shift( $add );
1431                         echo '<tr>' . $this->deletedLine( $line ) .
1432                                 $this->addedLine( $aline ) . "</tr>\n";
1433                 }
1434                 foreach ($add as $line) {       # If any leftovers
1435                         echo '<tr>' . $this->emptyLine() .
1436                                 $this->addedLine( $line ) . "</tr>\n";
1437                 }
1438                 wfProfileOut( $fname );
1439         }
1440 }
1441
1442 ?>