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) .
 121                                 '&token=' . urlencode( $wgUser->editToken( array( $wgTitle->getPrefixedText(), $this->mNewUser ) ) ) ) .
 122                                 ']</strong>';
 123                 } else {
 124                         $rollback = '';
 125                 }
 126                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->getID() != 0 &&
 127                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 128                 {
 129                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 130                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 131                 } else {
 132                         $patrol = '';
 133                 }
 134
 135                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'previousdiff' ), 'diff=prev&oldid='.$this->mOldid );
 136                 if ( $this->mNewid == 0 ) {
 137                         $nextlink = '';
 138                 } else {
 139                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 140                 }
 141
 142                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 143                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 144                         '<br />' . $prevlink;
 145                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 146                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 147                         '<br />' . $nextlink . $patrol;
 148
 149                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 150                   $oldHeader, $newHeader );
 151                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 152                 $wgOut->addWikiText( $this->mNewtext );
 153
 154                 wfProfileOut( $fname );
 155         }
 156
 157         # Show the first revision of an article. Uses normal diff headers in contrast to normal
 158         # "old revision" display style.
 159         #
 160         function showFirstRevision()
 161         {
 162                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 163
 164                 $fname = 'DifferenceEngine::showFirstRevision';
 165                 wfProfileIn( $fname );
 166
 167
 168                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 169
 170                 # Get article text from the DB
 171                 #
 172                 if ( ! $this->loadText() ) {
 173                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 174                           "{$this->mNewid})";
 175                         $mtext = wfMsg( 'missingarticle', $t );
 176                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 177                         $wgOut->addHTML( $mtext );
 178                         wfProfileOut( $fname );
 179                         return;
 180                 }
 181
 182                 # Check if user is allowed to look at this page. If not, bail out.
 183                 #
 184                 if ( !( $this->mOldPage->userCanRead() ) ) {
 185                         $wgOut->loginToUse();
 186                         $wgOut->output();
 187                         wfProfileOut( $fname );
 188                         exit;
 189                 }
 190
 191                 # Prepare the header box
 192                 #
 193                 $sk = $wgUser->getSkin();
 194
 195                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 196                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 197                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 198                         'target=' . urlencode($this->mOldUser) );
 199                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 200                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 201                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 202                         '<br />' . $nextlink. "</div>\n";
 203
 204                 $wgOut->addHTML( $header );
 205
 206                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 207                 $wgOut->setRobotpolicy( 'noindex,follow' );
 208
 209
 210                 # Show current revision
 211                 #
 212                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 213                 $wgOut->addWikiText( $this->mNewtext );
 214
 215                 wfProfileOut( $fname );
 216         }
 217
 218         function showDiff( $otext, $ntext, $otitle, $ntitle )
 219         {
 220                 global $wgOut;
 221                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 222         }
 223
 224         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 225                 global $wgUseExternalDiffEngine, $wgContLang;
 226                 $out = "
 227                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 228                         <tr>
 229                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 230                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 231                         </tr>
 232                 ";
 233                 $otext = $wgContLang->segmentForDiff($otext);
 234                 $ntext = $wgContLang->segmentForDiff($ntext);
 235                 $difftext='';
 236                 if ( $wgUseExternalDiffEngine ) {
 237                         # For historical reasons, external diff engine expects
 238                         # input text to be HTML-escaped already
 239                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 240                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 241                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 242                                 dl('php_wikidiff.so');
 243                         }
 244                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 245                 } else {
 246                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 247                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 248                         $diffs =& new Diff( $ota, $nta );
 249                         $formatter =& new TableDiffFormatter();
 250                         $difftext = $formatter->format( $diffs );
 251                 }
 252                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 253                 $out .= $difftext."</table>\n";
 254                 return $out;
 255         }
 256
 257         # Load the text of the articles to compare.  If newid is 0, then compare
 258         # the old article in oldid to the current article; if oldid is 0, then
 259         # compare the current article to the immediately previous one (ignoring
 260         # the value of newid).
 261         #
 262         function loadText()
 263         {
 264                 global $wgTitle, $wgOut, $wgLang;
 265                 $fname = 'DifferenceEngine::loadText';
 266
 267                 $dbr =& wfGetDB( DB_SLAVE );
 268                 if( $this->mNewid ) {
 269                         $this->newRev =& Revision::newFromId( $this->mNewid );
 270                 } else {
 271                         $this->newRev =& Revision::newFromTitle( $wgTitle );
 272                 }
 273
 274                 if( $this->newRev->isCurrent() ) {
 275                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 276                         $this->mNewPage = $wgTitle;
 277                         $newLink = $this->mNewPage->escapeLocalUrl();
 278                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 279                 } else {
 280                         $this->mNewPage = $this->newRev->getTitle();
 281                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 282                         $t = $wgLang->timeanddate( $this->newRev->getTimestamp(), true );
 283                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 284                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 285                 }
 286
 287                 if( $this->mOldid ) {
 288                         $this->oldRev =& Revision::newFromId( $this->mOldid );
 289                 } else {
 290                         $this->oldRev =& $this->newRev->getPrevious();
 291                         $this->mOldid = $this->oldRev->getId();
 292                 }
 293
 294                 $this->mOldPage = $this->oldRev->getTitle();
 295
 296                 $t = $wgLang->timeanddate( $this->oldRev->getTimestamp(), true );
 297                 $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 298                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 299
 300                 $this->mNewUser = $this->newRev->getUserText();
 301                 $this->mNewComment = $this->newRev->getComment();
 302                 $this->mNewtext = $this->newRev->getText();
 303
 304                 $this->mOldUser = $this->oldRev->getUserText();
 305                 $this->mOldComment = $this->oldRev->getComment();
 306                 $this->mOldtext = $this->oldRev->getText();
 307
 308                 return true;
 309         }
 310 }
 311
 312 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 313 //
 314 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 315 // You may copy this code freely under the conditions of the GPL.
 316 //
 317
 318 define('USE_ASSERTS', function_exists('assert'));
 319
 320 /**
 321  * @todo document
 322  * @access private
 323  * @package MediaWiki
 324  * @subpackage DifferenceEngine
 325  */
 326 class _DiffOp {
 327         var $type;
 328         var $orig;
 329         var $closing;
 330
 331         function reverse() {
 332                 trigger_error('pure virtual', E_USER_ERROR);
 333         }
 334
 335         function norig() {
 336                 return $this->orig ? sizeof($this->orig) : 0;
 337         }
 338
 339         function nclosing() {
 340                 return $this->closing ? sizeof($this->closing) : 0;
 341         }
 342 }
 343
 344 /**
 345  * @todo document
 346  * @access private
 347  * @package MediaWiki
 348  * @subpackage DifferenceEngine
 349  */
 350 class _DiffOp_Copy extends _DiffOp {
 351         var $type = 'copy';
 352
 353         function _DiffOp_Copy ($orig, $closing = false) {
 354                 if (!is_array($closing))
 355                         $closing = $orig;
 356                 $this->orig = $orig;
 357                 $this->closing = $closing;
 358         }
 359
 360         function reverse() {
 361                 return new _DiffOp_Copy($this->closing, $this->orig);
 362         }
 363 }
 364
 365 /**
 366  * @todo document
 367  * @access private
 368  * @package MediaWiki
 369  * @subpackage DifferenceEngine
 370  */
 371 class _DiffOp_Delete extends _DiffOp {
 372         var $type = 'delete';
 373
 374         function _DiffOp_Delete ($lines) {
 375                 $this->orig = $lines;
 376                 $this->closing = false;
 377         }
 378
 379         function reverse() {
 380                 return new _DiffOp_Add($this->orig);
 381         }
 382 }
 383
 384 /**
 385  * @todo document
 386  * @access private
 387  * @package MediaWiki
 388  * @subpackage DifferenceEngine
 389  */
 390 class _DiffOp_Add extends _DiffOp {
 391         var $type = 'add';
 392
 393         function _DiffOp_Add ($lines) {
 394                 $this->closing = $lines;
 395                 $this->orig = false;
 396         }
 397
 398         function reverse() {
 399                 return new _DiffOp_Delete($this->closing);
 400         }
 401 }
 402
 403 /**
 404  * @todo document
 405  * @access private
 406  * @package MediaWiki
 407  * @subpackage DifferenceEngine
 408  */
 409 class _DiffOp_Change extends _DiffOp {
 410         var $type = 'change';
 411
 412         function _DiffOp_Change ($orig, $closing) {
 413                 $this->orig = $orig;
 414                 $this->closing = $closing;
 415         }
 416
 417         function reverse() {
 418                 return new _DiffOp_Change($this->closing, $this->orig);
 419         }
 420 }
 421
 422
 423 /**
 424  * Class used internally by Diff to actually compute the diffs.
 425  *
 426  * The algorithm used here is mostly lifted from the perl module
 427  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 428  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 429  *
 430  * More ideas are taken from:
 431  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 432  *
 433  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 434  * diffutils-2.7, which can be found at:
 435  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 436  *
 437  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 438  * are my own.
 439  *
 440  * @author Geoffrey T. Dairiki
 441  * @access private
 442  * @package MediaWiki
 443  * @subpackage DifferenceEngine
 444  */
 445 class _DiffEngine
 446 {
 447         function diff ($from_lines, $to_lines) {
 448                 $fname = '_DiffEngine::diff';
 449                 wfProfileIn( $fname );
 450
 451                 $n_from = sizeof($from_lines);
 452                 $n_to = sizeof($to_lines);
 453
 454                 $this->xchanged = $this->ychanged = array();
 455                 $this->xv = $this->yv = array();
 456                 $this->xind = $this->yind = array();
 457                 unset($this->seq);
 458                 unset($this->in_seq);
 459                 unset($this->lcs);
 460
 461                 // Skip leading common lines.
 462                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 463                         if ($from_lines[$skip] != $to_lines[$skip])
 464                                 break;
 465                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 466                 }
 467                 // Skip trailing common lines.
 468                 $xi = $n_from; $yi = $n_to;
 469                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 470                         if ($from_lines[$xi] != $to_lines[$yi])
 471                                 break;
 472                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 473                 }
 474
 475                 // Ignore lines which do not exist in both files.
 476                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 477                         $xhash[$from_lines[$xi]] = 1;
 478                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 479                         $line = $to_lines[$yi];
 480                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 481                                 continue;
 482                         $yhash[$line] = 1;
 483                         $this->yv[] = $line;
 484                         $this->yind[] = $yi;
 485                 }
 486                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 487                         $line = $from_lines[$xi];
 488                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 489                                 continue;
 490                         $this->xv[] = $line;
 491                         $this->xind[] = $xi;
 492                 }
 493
 494                 // Find the LCS.
 495                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 496
 497                 // Merge edits when possible
 498                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 499                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 500
 501                 // Compute the edit operations.
 502                 $edits = array();
 503                 $xi = $yi = 0;
 504                 while ($xi < $n_from || $yi < $n_to) {
 505                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 506                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 507
 508                         // Skip matching "snake".
 509                         $copy = array();
 510                         while ( $xi < $n_from && $yi < $n_to
 511                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 512                                 $copy[] = $from_lines[$xi++];
 513                                 ++$yi;
 514                         }
 515                         if ($copy)
 516                                 $edits[] = new _DiffOp_Copy($copy);
 517
 518                         // Find deletes & adds.
 519                         $delete = array();
 520                         while ($xi < $n_from && $this->xchanged[$xi])
 521                                 $delete[] = $from_lines[$xi++];
 522
 523                         $add = array();
 524                         while ($yi < $n_to && $this->ychanged[$yi])
 525                                 $add[] = $to_lines[$yi++];
 526
 527                         if ($delete && $add)
 528                                 $edits[] = new _DiffOp_Change($delete, $add);
 529                         elseif ($delete)
 530                                 $edits[] = new _DiffOp_Delete($delete);
 531                         elseif ($add)
 532                                 $edits[] = new _DiffOp_Add($add);
 533                 }
 534                 wfProfileOut( $fname );
 535                 return $edits;
 536         }
 537
 538
 539         /* Divide the Largest Common Subsequence (LCS) of the sequences
 540          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 541          * sized segments.
 542          *
 543          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 544          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 545          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 546          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 547          * that (X0, Y0) == (XOFF, YOFF) and
 548          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 549          *
 550          * This function assumes that the first lines of the specified portions
 551          * of the two files do not match, and likewise that the last lines do not
 552          * match.  The caller must trim matching lines from the beginning and end
 553          * of the portions it is going to specify.
 554          */
 555         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 556                 $fname = '_DiffEngine::_diag';
 557                 wfProfileIn( $fname );
 558                 $flip = false;
 559
 560                 if ($xlim - $xoff > $ylim - $yoff) {
 561                         // Things seems faster (I'm not sure I understand why)
 562                                 // when the shortest sequence in X.
 563                                 $flip = true;
 564                         list ($xoff, $xlim, $yoff, $ylim)
 565                         = array( $yoff, $ylim, $xoff, $xlim);
 566                 }
 567
 568                 if ($flip)
 569                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 570                                 $ymatches[$this->xv[$i]][] = $i;
 571                 else
 572                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 573                                 $ymatches[$this->yv[$i]][] = $i;
 574
 575                 $this->lcs = 0;
 576                 $this->seq[0]= $yoff - 1;
 577                 $this->in_seq = array();
 578                 $ymids[0] = array();
 579
 580                 $numer = $xlim - $xoff + $nchunks - 1;
 581                 $x = $xoff;
 582                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 583                         wfProfileIn( "$fname-chunk" );
 584                         if ($chunk > 0)
 585                                 for ($i = 0; $i <= $this->lcs; $i++)
 586                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 587
 588                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 589                         for ( ; $x < $x1; $x++) {
 590                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 591                                         if (empty($ymatches[$line]))
 592                                                 continue;
 593                                 $matches = $ymatches[$line];
 594                                 reset($matches);
 595                                 while (list ($junk, $y) = each($matches))
 596                                         if (empty($this->in_seq[$y])) {
 597                                                 $k = $this->_lcs_pos($y);
 598                                                 USE_ASSERTS && assert($k > 0);
 599                                                 $ymids[$k] = $ymids[$k-1];
 600                                                 break;
 601                                         }
 602                                 while (list ($junk, $y) = each($matches)) {
 603                                         if ($y > $this->seq[$k-1]) {
 604                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 605                                                 // Optimization: this is a common case:
 606                                                 //      next match is just replacing previous match.
 607                                                 $this->in_seq[$this->seq[$k]] = false;
 608                                                 $this->seq[$k] = $y;
 609                                                 $this->in_seq[$y] = 1;
 610                                         } else if (empty($this->in_seq[$y])) {
 611                                                 $k = $this->_lcs_pos($y);
 612                                                 USE_ASSERTS && assert($k > 0);
 613                                                 $ymids[$k] = $ymids[$k-1];
 614                                         }
 615                                 }
 616                         }
 617                         wfProfileOut( "$fname-chunk" );
 618                 }
 619
 620                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 621                 $ymid = $ymids[$this->lcs];
 622                 for ($n = 0; $n < $nchunks - 1; $n++) {
 623                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 624                         $y1 = $ymid[$n] + 1;
 625                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 626                 }
 627                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 628
 629                 wfProfileOut( $fname );
 630                 return array($this->lcs, $seps);
 631         }
 632
 633         function _lcs_pos ($ypos) {
 634                 $fname = '_DiffEngine::_lcs_pos';
 635                 wfProfileIn( $fname );
 636
 637                 $end = $this->lcs;
 638                 if ($end == 0 || $ypos > $this->seq[$end]) {
 639                         $this->seq[++$this->lcs] = $ypos;
 640                         $this->in_seq[$ypos] = 1;
 641                         wfProfileOut( $fname );
 642                         return $this->lcs;
 643                 }
 644
 645                 $beg = 1;
 646                 while ($beg < $end) {
 647                         $mid = (int)(($beg + $end) / 2);
 648                         if ( $ypos > $this->seq[$mid] )
 649                                 $beg = $mid + 1;
 650                         else
 651                                 $end = $mid;
 652                 }
 653
 654                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 655
 656                 $this->in_seq[$this->seq[$end]] = false;
 657                 $this->seq[$end] = $ypos;
 658                 $this->in_seq[$ypos] = 1;
 659                 wfProfileOut( $fname );
 660                 return $end;
 661         }
 662
 663         /* Find LCS of two sequences.
 664          *
 665          * The results are recorded in the vectors $this->{x,y}changed[], by
 666          * storing a 1 in the element for each line that is an insertion
 667          * or deletion (ie. is not in the LCS).
 668          *
 669          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 670          *
 671          * Note that XLIM, YLIM are exclusive bounds.
 672          * All line numbers are origin-0 and discarded lines are not counted.
 673          */
 674         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 675                 $fname = '_DiffEngine::_compareseq';
 676                 wfProfileIn( $fname );
 677
 678                 // Slide down the bottom initial diagonal.
 679                 while ($xoff < $xlim && $yoff < $ylim
 680                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 681                         ++$xoff;
 682                         ++$yoff;
 683                 }
 684
 685                 // Slide up the top initial diagonal.
 686                 while ($xlim > $xoff && $ylim > $yoff
 687                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 688                         --$xlim;
 689                         --$ylim;
 690                 }
 691
 692                 if ($xoff == $xlim || $yoff == $ylim)
 693                         $lcs = 0;
 694                 else {
 695                         // This is ad hoc but seems to work well.
 696                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 697                         //$nchunks = max(2,min(8,(int)$nchunks));
 698                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 699                         list ($lcs, $seps)
 700                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 701                 }
 702
 703                 if ($lcs == 0) {
 704                         // X and Y sequences have no common subsequence:
 705                         // mark all changed.
 706                         while ($yoff < $ylim)
 707                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 708                         while ($xoff < $xlim)
 709                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 710                 } else {
 711                         // Use the partitions to split this problem into subproblems.
 712                         reset($seps);
 713                         $pt1 = $seps[0];
 714                         while ($pt2 = next($seps)) {
 715                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 716                                 $pt1 = $pt2;
 717                         }
 718                 }
 719                 wfProfileOut( $fname );
 720         }
 721
 722         /* Adjust inserts/deletes of identical lines to join changes
 723          * as much as possible.
 724          *
 725          * We do something when a run of changed lines include a
 726          * line at one end and has an excluded, identical line at the other.
 727          * We are free to choose which identical line is included.
 728          * `compareseq' usually chooses the one at the beginning,
 729          * but usually it is cleaner to consider the following identical line
 730          * to be the "change".
 731          *
 732          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 733          */
 734         function _shift_boundaries ($lines, &$changed, $other_changed) {
 735                 $fname = '_DiffEngine::_shift_boundaries';
 736                 wfProfileIn( $fname );
 737                 $i = 0;
 738                 $j = 0;
 739
 740                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 741                 $len = sizeof($lines);
 742                 $other_len = sizeof($other_changed);
 743
 744                 while (1) {
 745                         /*
 746                          * Scan forwards to find beginning of another run of changes.
 747                          * Also keep track of the corresponding point in the other file.
 748                          *
 749                          * Throughout this code, $i and $j are adjusted together so that
 750                          * the first $i elements of $changed and the first $j elements
 751                          * of $other_changed both contain the same number of zeros
 752                          * (unchanged lines).
 753                          * Furthermore, $j is always kept so that $j == $other_len or
 754                          * $other_changed[$j] == false.
 755                          */
 756                         while ($j < $other_len && $other_changed[$j])
 757                                 $j++;
 758
 759                         while ($i < $len && ! $changed[$i]) {
 760                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 761                                 $i++; $j++;
 762                                 while ($j < $other_len && $other_changed[$j])
 763                                         $j++;
 764                         }
 765
 766                         if ($i == $len)
 767                                 break;
 768
 769                         $start = $i;
 770
 771                         // Find the end of this run of changes.
 772                         while (++$i < $len && $changed[$i])
 773                                 continue;
 774
 775                         do {
 776                                 /*
 777                                  * Record the length of this run of changes, so that
 778                                  * we can later determine whether the run has grown.
 779                                  */
 780                                 $runlength = $i - $start;
 781
 782                                 /*
 783                                  * Move the changed region back, so long as the
 784                                  * previous unchanged line matches the last changed one.
 785                                  * This merges with previous changed regions.
 786                                  */
 787                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 788                                         $changed[--$start] = 1;
 789                                         $changed[--$i] = false;
 790                                         while ($start > 0 && $changed[$start - 1])
 791                                                 $start--;
 792                                         USE_ASSERTS && assert('$j > 0');
 793                                         while ($other_changed[--$j])
 794                                                 continue;
 795                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 796                                 }
 797
 798                                 /*
 799                                  * Set CORRESPONDING to the end of the changed run, at the last
 800                                  * point where it corresponds to a changed run in the other file.
 801                                  * CORRESPONDING == LEN means no such point has been found.
 802                                  */
 803                                 $corresponding = $j < $other_len ? $i : $len;
 804
 805                                 /*
 806                                  * Move the changed region forward, so long as the
 807                                  * first changed line matches the following unchanged one.
 808                                  * This merges with following changed regions.
 809                                  * Do this second, so that if there are no merges,
 810                                  * the changed region is moved forward as far as possible.
 811                                  */
 812                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 813                                         $changed[$start++] = false;
 814                                         $changed[$i++] = 1;
 815                                         while ($i < $len && $changed[$i])
 816                                                 $i++;
 817
 818                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 819                                         $j++;
 820                                         if ($j < $other_len && $other_changed[$j]) {
 821                                                 $corresponding = $i;
 822                                                 while ($j < $other_len && $other_changed[$j])
 823                                                         $j++;
 824                                         }
 825                                 }
 826                         } while ($runlength != $i - $start);
 827
 828                         /*
 829                          * If possible, move the fully-merged run of changes
 830                          * back to a corresponding run in the other file.
 831                          */
 832                         while ($corresponding < $i) {
 833                                 $changed[--$start] = 1;
 834                                 $changed[--$i] = 0;
 835                                 USE_ASSERTS && assert('$j > 0');
 836                                 while ($other_changed[--$j])
 837                                         continue;
 838                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 839                         }
 840                 }
 841                 wfProfileOut( $fname );
 842         }
 843 }
 844
 845 /**
 846  * Class representing a 'diff' between two sequences of strings.
 847  * @todo document
 848  * @access private
 849  * @package MediaWiki
 850  * @subpackage DifferenceEngine
 851  */
 852 class Diff
 853 {
 854         var $edits;
 855
 856         /**
 857          * Constructor.
 858          * Computes diff between sequences of strings.
 859          *
 860          * @param $from_lines array An array of strings.
 861          *                (Typically these are lines from a file.)
 862          * @param $to_lines array An array of strings.
 863          */
 864         function Diff($from_lines, $to_lines) {
 865                 $eng = new _DiffEngine;
 866                 $this->edits = $eng->diff($from_lines, $to_lines);
 867                 //$this->_check($from_lines, $to_lines);
 868         }
 869
 870         /**
 871          * Compute reversed Diff.
 872          *
 873          * SYNOPSIS:
 874          *
 875          *      $diff = new Diff($lines1, $lines2);
 876          *      $rev = $diff->reverse();
 877          * @return object A Diff object representing the inverse of the
 878          *                                original diff.
 879          */
 880         function reverse () {
 881                 $rev = $this;
 882                 $rev->edits = array();
 883                 foreach ($this->edits as $edit) {
 884                         $rev->edits[] = $edit->reverse();
 885                 }
 886                 return $rev;
 887         }
 888
 889         /**
 890          * Check for empty diff.
 891          *
 892          * @return bool True iff two sequences were identical.
 893          */
 894         function isEmpty () {
 895                 foreach ($this->edits as $edit) {
 896                         if ($edit->type != 'copy')
 897                                 return false;
 898                 }
 899                 return true;
 900         }
 901
 902         /**
 903          * Compute the length of the Longest Common Subsequence (LCS).
 904          *
 905          * This is mostly for diagnostic purposed.
 906          *
 907          * @return int The length of the LCS.
 908          */
 909         function lcs () {
 910                 $lcs = 0;
 911                 foreach ($this->edits as $edit) {
 912                         if ($edit->type == 'copy')
 913                                 $lcs += sizeof($edit->orig);
 914                 }
 915                 return $lcs;
 916         }
 917
 918         /**
 919          * Get the original set of lines.
 920          *
 921          * This reconstructs the $from_lines parameter passed to the
 922          * constructor.
 923          *
 924          * @return array The original sequence of strings.
 925          */
 926         function orig() {
 927                 $lines = array();
 928
 929                 foreach ($this->edits as $edit) {
 930                         if ($edit->orig)
 931                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 932                 }
 933                 return $lines;
 934         }
 935
 936         /**
 937          * Get the closing set of lines.
 938          *
 939          * This reconstructs the $to_lines parameter passed to the
 940          * constructor.
 941          *
 942          * @return array The sequence of strings.
 943          */
 944         function closing() {
 945                 $lines = array();
 946
 947                 foreach ($this->edits as $edit) {
 948                         if ($edit->closing)
 949                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 950                 }
 951                 return $lines;
 952         }
 953
 954         /**
 955          * Check a Diff for validity.
 956          *
 957          * This is here only for debugging purposes.
 958          */
 959         function _check ($from_lines, $to_lines) {
 960                 $fname = 'Diff::_check';
 961                 wfProfileIn( $fname );
 962                 if (serialize($from_lines) != serialize($this->orig()))
 963                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 964                 if (serialize($to_lines) != serialize($this->closing()))
 965                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 966
 967                 $rev = $this->reverse();
 968                 if (serialize($to_lines) != serialize($rev->orig()))
 969                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 970                 if (serialize($from_lines) != serialize($rev->closing()))
 971                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 972
 973
 974                 $prevtype = 'none';
 975                 foreach ($this->edits as $edit) {
 976                         if ( $prevtype == $edit->type )
 977                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 978                         $prevtype = $edit->type;
 979                 }
 980
 981                 $lcs = $this->lcs();
 982                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 983                 wfProfileOut( $fname );
 984         }
 985 }
 986
 987 /**
 988  * FIXME: bad name.
 989  * @todo document
 990  * @access private
 991  * @package MediaWiki
 992  * @subpackage DifferenceEngine
 993  */
 994 class MappedDiff extends Diff
 995 {
 996         /**
 997          * Constructor.
 998          *
 999          * Computes diff between sequences of strings.
1000          *
1001          * This can be used to compute things like
1002          * case-insensitve diffs, or diffs which ignore
1003          * changes in white-space.
1004          *
1005          * @param $from_lines array An array of strings.
1006          *      (Typically these are lines from a file.)
1007          *
1008          * @param $to_lines array An array of strings.
1009          *
1010          * @param $mapped_from_lines array This array should
1011          *      have the same size number of elements as $from_lines.
1012          *      The elements in $mapped_from_lines and
1013          *      $mapped_to_lines are what is actually compared
1014          *      when computing the diff.
1015          *
1016          * @param $mapped_to_lines array This array should
1017          *      have the same number of elements as $to_lines.
1018          */
1019         function MappedDiff($from_lines, $to_lines,
1020                                                 $mapped_from_lines, $mapped_to_lines) {
1021                 $fname = 'MappedDiff::MappedDiff';
1022                 wfProfileIn( $fname );
1023
1024                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1025                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1026
1027                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1028
1029                 $xi = $yi = 0;
1030                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1031                         $orig = &$this->edits[$i]->orig;
1032                         if (is_array($orig)) {
1033                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1034                                 $xi += sizeof($orig);
1035                         }
1036
1037                         $closing = &$this->edits[$i]->closing;
1038                         if (is_array($closing)) {
1039                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1040                                 $yi += sizeof($closing);
1041                         }
1042                 }
1043                 wfProfileOut( $fname );
1044         }
1045 }
1046
1047 /**
1048  * A class to format Diffs
1049  *
1050  * This class formats the diff in classic diff format.
1051  * It is intended that this class be customized via inheritance,
1052  * to obtain fancier outputs.
1053  * @todo document
1054  * @access private
1055  * @package MediaWiki
1056  * @subpackage DifferenceEngine
1057  */
1058 class DiffFormatter
1059 {
1060         /**
1061          * Number of leading context "lines" to preserve.
1062          *
1063          * This should be left at zero for this class, but subclasses
1064          * may want to set this to other values.
1065          */
1066         var $leading_context_lines = 0;
1067
1068         /**
1069          * Number of trailing context "lines" to preserve.
1070          *
1071          * This should be left at zero for this class, but subclasses
1072          * may want to set this to other values.
1073          */
1074         var $trailing_context_lines = 0;
1075
1076         /**
1077          * Format a diff.
1078          *
1079          * @param $diff object A Diff object.
1080          * @return string The formatted output.
1081          */
1082         function format($diff) {
1083                 $fname = 'DiffFormatter::format';
1084                 wfProfileIn( $fname );
1085
1086                 $xi = $yi = 1;
1087                 $block = false;
1088                 $context = array();
1089
1090                 $nlead = $this->leading_context_lines;
1091                 $ntrail = $this->trailing_context_lines;
1092
1093                 $this->_start_diff();
1094
1095                 foreach ($diff->edits as $edit) {
1096                         if ($edit->type == 'copy') {
1097                                 if (is_array($block)) {
1098                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1099                                                 $block[] = $edit;
1100                                         }
1101                                         else{
1102                                                 if ($ntrail) {
1103                                                         $context = array_slice($edit->orig, 0, $ntrail);
1104                                                         $block[] = new _DiffOp_Copy($context);
1105                                                 }
1106                                                 $this->_block($x0, $ntrail + $xi - $x0,
1107                                                                           $y0, $ntrail + $yi - $y0,
1108                                                                           $block);
1109                                                 $block = false;
1110                                         }
1111                                 }
1112                                 $context = $edit->orig;
1113                         }
1114                         else {
1115                                 if (! is_array($block)) {
1116                                         $context = array_slice($context, sizeof($context) - $nlead);
1117                                         $x0 = $xi - sizeof($context);
1118                                         $y0 = $yi - sizeof($context);
1119                                         $block = array();
1120                                         if ($context)
1121                                                 $block[] = new _DiffOp_Copy($context);
1122                                 }
1123                                 $block[] = $edit;
1124                         }
1125
1126                         if ($edit->orig)
1127                                 $xi += sizeof($edit->orig);
1128                         if ($edit->closing)
1129                                 $yi += sizeof($edit->closing);
1130                 }
1131
1132                 if (is_array($block))
1133                         $this->_block($x0, $xi - $x0,
1134                                                   $y0, $yi - $y0,
1135                                                   $block);
1136
1137                 $end = $this->_end_diff();
1138                 wfProfileOut( $fname );
1139                 return $end;
1140         }
1141
1142         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1143                 $fname = 'DiffFormatter::_block';
1144                 wfProfileIn( $fname );
1145                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1146                 foreach ($edits as $edit) {
1147                         if ($edit->type == 'copy')
1148                                 $this->_context($edit->orig);
1149                         elseif ($edit->type == 'add')
1150                                 $this->_added($edit->closing);
1151                         elseif ($edit->type == 'delete')
1152                                 $this->_deleted($edit->orig);
1153                         elseif ($edit->type == 'change')
1154                                 $this->_changed($edit->orig, $edit->closing);
1155                         else
1156                                 trigger_error('Unknown edit type', E_USER_ERROR);
1157                 }
1158                 $this->_end_block();
1159                 wfProfileOut( $fname );
1160         }
1161
1162         function _start_diff() {
1163                 ob_start();
1164         }
1165
1166         function _end_diff() {
1167                 $val = ob_get_contents();
1168                 ob_end_clean();
1169                 return $val;
1170         }
1171
1172         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1173                 if ($xlen > 1)
1174                         $xbeg .= "," . ($xbeg + $xlen - 1);
1175                 if ($ylen > 1)
1176                         $ybeg .= "," . ($ybeg + $ylen - 1);
1177
1178                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1179         }
1180
1181         function _start_block($header) {
1182                 echo $header;
1183         }
1184
1185         function _end_block() {
1186         }
1187
1188         function _lines($lines, $prefix = ' ') {
1189                 foreach ($lines as $line)
1190                         echo "$prefix $line\n";
1191         }
1192
1193         function _context($lines) {
1194                 $this->_lines($lines);
1195         }
1196
1197         function _added($lines) {
1198                 $this->_lines($lines, '>');
1199         }
1200         function _deleted($lines) {
1201                 $this->_lines($lines, '<');
1202         }
1203
1204         function _changed($orig, $closing) {
1205                 $this->_deleted($orig);
1206                 echo "---\n";
1207                 $this->_added($closing);
1208         }
1209 }
1210
1211
1212 /**
1213  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1214  *
1215  */
1216
1217 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1218
1219 /**
1220  * @todo document
1221  * @access private
1222  * @package MediaWiki
1223  * @subpackage DifferenceEngine
1224  */
1225 class _HWLDF_WordAccumulator {
1226         function _HWLDF_WordAccumulator () {
1227                 $this->_lines = array();
1228                 $this->_line = '';
1229                 $this->_group = '';
1230                 $this->_tag = '';
1231         }
1232
1233         function _flushGroup ($new_tag) {
1234                 if ($this->_group !== '') {
1235                         if ($this->_tag == 'mark')
1236                                 $this->_line .= '<span class="diffchange">' .
1237                                         htmlspecialchars ( $this->_group ) . '</span>';
1238                         else
1239                                 $this->_line .= htmlspecialchars ( $this->_group );
1240                 }
1241                 $this->_group = '';
1242                 $this->_tag = $new_tag;
1243         }
1244
1245         function _flushLine ($new_tag) {
1246                 $this->_flushGroup($new_tag);
1247                 if ($this->_line != '')
1248                         array_push ( $this->_lines, $this->_line );
1249                 else
1250                         # make empty lines visible by inserting an NBSP
1251                         array_push ( $this->_lines, NBSP );
1252                 $this->_line = '';
1253         }
1254
1255         function addWords ($words, $tag = '') {
1256                 if ($tag != $this->_tag)
1257                         $this->_flushGroup($tag);
1258
1259                 foreach ($words as $word) {
1260                         // new-line should only come as first char of word.
1261                         if ($word == '')
1262                                 continue;
1263                         if ($word[0] == "\n") {
1264                                 $this->_flushLine($tag);
1265                                 $word = substr($word, 1);
1266                         }
1267                         assert(!strstr($word, "\n"));
1268                         $this->_group .= $word;
1269                 }
1270         }
1271
1272         function getLines() {
1273                 $this->_flushLine('~done');
1274                 return $this->_lines;
1275         }
1276 }
1277
1278 /**
1279  * @todo document
1280  * @access private
1281  * @package MediaWiki
1282  * @subpackage DifferenceEngine
1283  */
1284 class WordLevelDiff extends MappedDiff
1285 {
1286         function WordLevelDiff ($orig_lines, $closing_lines) {
1287                 $fname = 'WordLevelDiff::WordLevelDiff';
1288                 wfProfileIn( $fname );
1289
1290                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1291                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1292
1293
1294                 $this->MappedDiff($orig_words, $closing_words,
1295                                                   $orig_stripped, $closing_stripped);
1296                 wfProfileOut( $fname );
1297         }
1298
1299         function _split($lines) {
1300                 $fname = 'WordLevelDiff::_split';
1301                 wfProfileIn( $fname );
1302                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1303                                                         implode("\n", $lines),
1304                                                         $m)) {
1305                         wfProfileOut( $fname );
1306                         return array(array(''), array(''));
1307                 }
1308                 wfProfileOut( $fname );
1309                 return array($m[0], $m[1]);
1310         }
1311
1312         function orig () {
1313                 $fname = 'WordLevelDiff::orig';
1314                 wfProfileIn( $fname );
1315                 $orig = new _HWLDF_WordAccumulator;
1316
1317                 foreach ($this->edits as $edit) {
1318                         if ($edit->type == 'copy')
1319                                 $orig->addWords($edit->orig);
1320                         elseif ($edit->orig)
1321                                 $orig->addWords($edit->orig, 'mark');
1322                 }
1323                 $lines = $orig->getLines();
1324                 wfProfileOut( $fname );
1325                 return $lines;
1326         }
1327
1328         function closing () {
1329                 $fname = 'WordLevelDiff::closing';
1330                 wfProfileIn( $fname );
1331                 $closing = new _HWLDF_WordAccumulator;
1332
1333                 foreach ($this->edits as $edit) {
1334                         if ($edit->type == 'copy')
1335                                 $closing->addWords($edit->closing);
1336                         elseif ($edit->closing)
1337                                 $closing->addWords($edit->closing, 'mark');
1338                 }
1339                 $lines = $closing->getLines();
1340                 wfProfileOut( $fname );
1341                 return $lines;
1342         }
1343 }
1344
1345 /**
1346  *      Wikipedia Table style diff formatter.
1347  * @todo document
1348  * @access private
1349  * @package MediaWiki
1350  * @subpackage DifferenceEngine
1351  */
1352 class TableDiffFormatter extends DiffFormatter
1353 {
1354         function TableDiffFormatter() {
1355                 $this->leading_context_lines = 2;
1356                 $this->trailing_context_lines = 2;
1357         }
1358
1359         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1360                 $l1 = wfMsg( 'lineno', $xbeg );
1361                 $l2 = wfMsg( 'lineno', $ybeg );
1362
1363                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1364                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1365                 return $r;
1366         }
1367
1368         function _start_block( $header ) {
1369                 global $wgOut;
1370                 echo $header;
1371         }
1372
1373         function _end_block() {
1374         }
1375
1376         function _lines( $lines, $prefix=' ', $color='white' ) {
1377         }
1378
1379         # HTML-escape parameter before calling this
1380         function addedLine( $line ) {
1381                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1382         }
1383
1384         # HTML-escape parameter before calling this
1385         function deletedLine( $line ) {
1386                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1387         }
1388
1389         # HTML-escape parameter before calling this
1390         function contextLine( $line ) {
1391                 return "<td> </td><td class='diff-context'>{$line}</td>";
1392         }
1393
1394         function emptyLine() {
1395                 return '<td colspan="2">&nbsp;</td>';
1396         }
1397
1398         function _added( $lines ) {
1399                 foreach ($lines as $line) {
1400                         echo '<tr>' . $this->emptyLine() .
1401                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1402                 }
1403         }
1404
1405         function _deleted($lines) {
1406                 foreach ($lines as $line) {
1407                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1408                           $this->emptyLine() . "</tr>\n";
1409                 }
1410         }
1411
1412         function _context( $lines ) {
1413                 foreach ($lines as $line) {
1414                         echo '<tr>' .
1415                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1416                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1417                 }
1418         }
1419
1420         function _changed( $orig, $closing ) {
1421                 $fname = 'TableDiffFormatter::_changed';
1422                 wfProfileIn( $fname );
1423
1424                 $diff = new WordLevelDiff( $orig, $closing );
1425                 $del = $diff->orig();
1426                 $add = $diff->closing();
1427
1428                 # Notice that WordLevelDiff returns HTML-escaped output.
1429                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1430
1431                 while ( $line = array_shift( $del ) ) {
1432                         $aline = array_shift( $add );
1433                         echo '<tr>' . $this->deletedLine( $line ) .
1434                                 $this->addedLine( $aline ) . "</tr>\n";
1435                 }
1436                 foreach ($add as $line) {       # If any leftovers
1437                         echo '<tr>' . $this->emptyLine() .
1438                                 $this->addedLine( $line ) . "</tr>\n";
1439                 }
1440                 wfProfileOut( $fname );
1441         }
1442 }
1443
1444 ?>