includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @package MediaWiki
   5  * @subpackage DifferenceEngine
   6  */
   7
   8 require_once( 'Revision.php' );
   9
  10 /**
  11  * @todo document
  12  * @access public
  13  */
  14 class DifferenceEngine {
  15         /* private */ var $mOldid, $mNewid;
  16         /* private */ var $mOldtitle, $mNewtitle, $mPagetitle;
  17         /* private */ var $mOldtext, $mNewtext;
  18         /* private */ var $mOldUser, $mNewUser;
  19         /* private */ var $mOldComment, $mNewComment;
  20         /* private */ var $mOldPage, $mNewPage;
  21         /* private */ var $mRcidMarkPatrolled;
  22
  23         function DifferenceEngine( $old, $new, $rcid = 0 )
  24         {
  25                 global $wgTitle;
  26                 if ( 'prev' == $new ) {
  27                         # Show diff between revision $old and the previous one.
  28                         # Get previous one from DB.
  29                         #
  30                         $this->mNewid = intval($old);
  31
  32                         $this->mOldid = $wgTitle->getPreviousRevisionID( $this->mNewid );
  33
  34                 } elseif ( 'next' == $new ) {
  35
  36                         # Show diff between revision $old and the previous one.
  37                         # Get previous one from DB.
  38                         #
  39                         $this->mOldid = intval($old);
  40                         $this->mNewid = $wgTitle->getNextRevisionID( $this->mOldid );
  41                         if ( false === $this->mNewid ) {
  42                                 # if no result, NewId points to the newest old revision. The only newer
  43                                 # revision is cur, which is "0".
  44                                 $this->mNewid = 0;
  45                         }
  46
  47                 } else {
  48
  49                         $this->mOldid = intval($old);
  50                         $this->mNewid = intval($new);
  51                 }
  52                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  53         }
  54
  55         function showDiffPage()
  56         {
  57                 global $wgUser, $wgTitle, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol, $wgUseRCPatrol;
  58                 $fname = 'DifferenceEngine::showDiffPage';
  59                 wfProfileIn( $fname );
  60
  61                 # mOldid is false if the difference engine is called with a "vague" query for
  62                 # a diff between a version V and its previous version V' AND the version V
  63                 # is the first version of that article. In that case, V' does not exist.
  64                 if ( $this->mOldid === false ) {
  65                         $this->showFirstRevision();
  66                         wfProfileOut( $fname );
  67                         return;
  68                 }
  69
  70                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  71                   "{$this->mNewid})";
  72                 $mtext = wfMsg( 'missingarticle', $t );
  73
  74                 $wgOut->setArticleFlag( false );
  75                 if ( ! $this->loadText() ) {
  76                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
  77                         $wgOut->addHTML( $mtext );
  78                         wfProfileOut( $fname );
  79                         return;
  80                 }
  81                 $wgOut->suppressQuickbar();
  82
  83                 $oldTitle = $this->mOldPage->getPrefixedText();
  84                 $newTitle = $this->mNewPage->getPrefixedText();
  85                 if( $oldTitle == $newTitle ) {
  86                         $wgOut->setPageTitle( $newTitle );
  87                 } else {
  88                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
  89                 }
  90                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
  91                 $wgOut->setRobotpolicy( 'noindex,follow' );
  92
  93                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  94                         $wgOut->loginToUse();
  95                         $wgOut->output();
  96                         wfProfileOut( $fname );
  97                         exit;
  98                 }
  99
 100                 $sk = $wgUser->getSkin();
 101                 $talk = $wgContLang->getNsText( NS_TALK );
 102                 $contribs = wfMsg( 'contribslink' );
 103
 104                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 105                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 106
 107                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 108                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 109                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 110                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 111                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 112                         'target=' . urlencode($this->mOldUser) );
 113                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 114                         'target=' . urlencode($this->mNewUser) );
 115                 if ( !$this->mNewid && $wgUser->isAllowed('rollback') ) {
 116                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
 117                                 'action=rollback&from=' . urlencode($this->mNewUser) ) . ']</strong>';
 118                 } else {
 119                         $rollback = '';
 120                 }
 121                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->getID() != 0 &&
 122                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 123                 {
 124                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 125                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 126                 } else {
 127                         $patrol = '';
 128                 }
 129
 130                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'previousdiff' ), 'diff=prev&oldid='.$this->mOldid );
 131                 if ( $this->mNewid == 0 ) {
 132                         $nextlink = '';
 133                 } else {
 134                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 135                 }
 136
 137                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 138                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 139                         '<br />' . $prevlink;
 140                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 141                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 142                         '<br />' . $nextlink . $patrol;
 143
 144                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 145                   $oldHeader, $newHeader );
 146                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 147                 $wgOut->addWikiText( $this->mNewtext );
 148
 149                 wfProfileOut( $fname );
 150         }
 151
 152         # Show the first revision of an article. Uses normal diff headers in contrast to normal
 153         # "old revision" display style.
 154         #
 155         function showFirstRevision()
 156         {
 157                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 158
 159                 $fname = 'DifferenceEngine::showFirstRevision';
 160                 wfProfileIn( $fname );
 161
 162
 163                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 164
 165                 # Get article text from the DB
 166                 #
 167                 if ( ! $this->loadText() ) {
 168                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 169                           "{$this->mNewid})";
 170                         $mtext = wfMsg( 'missingarticle', $t );
 171                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 172                         $wgOut->addHTML( $mtext );
 173                         wfProfileOut( $fname );
 174                         return;
 175                 }
 176
 177                 # Check if user is allowed to look at this page. If not, bail out.
 178                 #
 179                 if ( !( $this->mOldPage->userCanRead() ) ) {
 180                         $wgOut->loginToUse();
 181                         $wgOut->output();
 182                         wfProfileOut( $fname );
 183                         exit;
 184                 }
 185
 186                 # Prepare the header box
 187                 #
 188                 $sk = $wgUser->getSkin();
 189
 190                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 191                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 192                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 193                         'target=' . urlencode($this->mOldUser) );
 194                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 195                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 196                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 197                         '<br />' . $nextlink. "</div>\n";
 198
 199                 $wgOut->addHTML( $header );
 200
 201                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 202                 $wgOut->setRobotpolicy( 'noindex,follow' );
 203
 204
 205                 # Show current revision
 206                 #
 207                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 208                 $wgOut->addWikiText( $this->mNewtext );
 209
 210                 wfProfileOut( $fname );
 211         }
 212
 213         function showDiff( $otext, $ntext, $otitle, $ntitle )
 214         {
 215                 global $wgOut;
 216                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 217         }
 218
 219         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 220                 global $wgUseExternalDiffEngine, $wgContLang;
 221                 $out = "
 222                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 223                         <tr>
 224                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 225                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 226                         </tr>
 227                 ";
 228                 $otext = $wgContLang->segmentForDiff($otext);
 229                 $ntext = $wgContLang->segmentForDiff($ntext);
 230                 $difftext='';
 231                 if ( $wgUseExternalDiffEngine ) {
 232                         # For historical reasons, external diff engine expects
 233                         # input text to be HTML-escaped already
 234                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 235                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 236                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 237                                 dl('php_wikidiff.so');
 238                         }
 239                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 240                 } else {
 241                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 242                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 243                         $diffs =& new Diff( $ota, $nta );
 244                         $formatter =& new TableDiffFormatter();
 245                         $difftext = $formatter->format( $diffs );
 246                 }
 247                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 248                 $out .= $difftext."</table>\n";
 249                 return $out;
 250         }
 251
 252         # Load the text of the articles to compare.  If newid is 0, then compare
 253         # the old article in oldid to the current article; if oldid is 0, then
 254         # compare the current article to the immediately previous one (ignoring
 255         # the value of newid).
 256         #
 257         function loadText()
 258         {
 259                 global $wgTitle, $wgOut, $wgLang;
 260                 $fname = 'DifferenceEngine::loadText';
 261
 262                 $dbr =& wfGetDB( DB_SLAVE );
 263                 if( $this->mNewid ) {
 264                         $this->newRev =& Revision::newFromId( $this->mNewid );
 265                 } else {
 266                         $this->newRev =& Revision::newFromTitle( $wgTitle );
 267                 }
 268
 269                 if( $this->newRev->isCurrent() ) {
 270                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 271                         $this->mNewPage = $wgTitle;
 272                         $newLink = $this->mNewPage->escapeLocalUrl();
 273                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 274                 } else {
 275                         $this->mNewPage = $this->newRev->getTitle();
 276                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 277                         $t = $wgLang->timeanddate( $this->newRev->getTimestamp(), true );
 278                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 279                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 280                 }
 281
 282                 if( $this->mOldid ) {
 283                         $this->oldRev =& Revision::newFromId( $this->mOldid );
 284                 } else {
 285                         $this->oldRev =& $this->newRev->getPrevious();
 286                 }
 287
 288                 $this->mOldPage = $this->oldRev->getTitle();
 289
 290                 $t = $wgLang->timeanddate( $this->oldRev->getTimestamp(), true );
 291                 $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 292                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 293
 294                 $this->mNewUser = $this->newRev->getUserText();
 295                 $this->mNewComment = $this->newRev->getComment();
 296                 $this->mNewtext = $this->newRev->getText();
 297
 298                 $this->mOldUser = $this->oldRev->getUserText();
 299                 $this->mOldComment = $this->oldRev->getComment();
 300                 $this->mOldtext = $this->oldRev->getText();
 301
 302                 return true;
 303         }
 304 }
 305
 306 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 307 //
 308 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 309 // You may copy this code freely under the conditions of the GPL.
 310 //
 311
 312 define('USE_ASSERTS', function_exists('assert'));
 313
 314 /**
 315  * @todo document
 316  * @access private
 317  */
 318 class _DiffOp {
 319         var $type;
 320         var $orig;
 321         var $closing;
 322
 323         function reverse() {
 324                 trigger_error('pure virtual', E_USER_ERROR);
 325         }
 326
 327         function norig() {
 328                 return $this->orig ? sizeof($this->orig) : 0;
 329         }
 330
 331         function nclosing() {
 332                 return $this->closing ? sizeof($this->closing) : 0;
 333         }
 334 }
 335
 336 /**
 337  * @todo document
 338  * @access private
 339  */
 340 class _DiffOp_Copy extends _DiffOp {
 341         var $type = 'copy';
 342
 343         function _DiffOp_Copy ($orig, $closing = false) {
 344                 if (!is_array($closing))
 345                         $closing = $orig;
 346                 $this->orig = $orig;
 347                 $this->closing = $closing;
 348         }
 349
 350         function reverse() {
 351                 return new _DiffOp_Copy($this->closing, $this->orig);
 352         }
 353 }
 354
 355 /**
 356  * @todo document
 357  * @access private
 358  */
 359 class _DiffOp_Delete extends _DiffOp {
 360         var $type = 'delete';
 361
 362         function _DiffOp_Delete ($lines) {
 363                 $this->orig = $lines;
 364                 $this->closing = false;
 365         }
 366
 367         function reverse() {
 368                 return new _DiffOp_Add($this->orig);
 369         }
 370 }
 371
 372 /**
 373  * @todo document
 374  * @access private
 375  */
 376 class _DiffOp_Add extends _DiffOp {
 377         var $type = 'add';
 378
 379         function _DiffOp_Add ($lines) {
 380                 $this->closing = $lines;
 381                 $this->orig = false;
 382         }
 383
 384         function reverse() {
 385                 return new _DiffOp_Delete($this->closing);
 386         }
 387 }
 388
 389 /**
 390  * @todo document
 391  * @access private
 392  */
 393 class _DiffOp_Change extends _DiffOp {
 394         var $type = 'change';
 395
 396         function _DiffOp_Change ($orig, $closing) {
 397                 $this->orig = $orig;
 398                 $this->closing = $closing;
 399         }
 400
 401         function reverse() {
 402                 return new _DiffOp_Change($this->closing, $this->orig);
 403         }
 404 }
 405
 406
 407 /**
 408  * Class used internally by Diff to actually compute the diffs.
 409  *
 410  * The algorithm used here is mostly lifted from the perl module
 411  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 412  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 413  *
 414  * More ideas are taken from:
 415  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 416  *
 417  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 418  * diffutils-2.7, which can be found at:
 419  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 420  *
 421  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 422  * are my own.
 423  *
 424  * @author Geoffrey T. Dairiki
 425  * @access private
 426  */
 427 class _DiffEngine
 428 {
 429         function diff ($from_lines, $to_lines) {
 430                 $fname = '_DiffEngine::diff';
 431                 wfProfileIn( $fname );
 432
 433                 $n_from = sizeof($from_lines);
 434                 $n_to = sizeof($to_lines);
 435
 436                 $this->xchanged = $this->ychanged = array();
 437                 $this->xv = $this->yv = array();
 438                 $this->xind = $this->yind = array();
 439                 unset($this->seq);
 440                 unset($this->in_seq);
 441                 unset($this->lcs);
 442
 443                 // Skip leading common lines.
 444                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 445                         if ($from_lines[$skip] != $to_lines[$skip])
 446                                 break;
 447                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 448                 }
 449                 // Skip trailing common lines.
 450                 $xi = $n_from; $yi = $n_to;
 451                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 452                         if ($from_lines[$xi] != $to_lines[$yi])
 453                                 break;
 454                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 455                 }
 456
 457                 // Ignore lines which do not exist in both files.
 458                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 459                         $xhash[$from_lines[$xi]] = 1;
 460                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 461                         $line = $to_lines[$yi];
 462                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 463                                 continue;
 464                         $yhash[$line] = 1;
 465                         $this->yv[] = $line;
 466                         $this->yind[] = $yi;
 467                 }
 468                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 469                         $line = $from_lines[$xi];
 470                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 471                                 continue;
 472                         $this->xv[] = $line;
 473                         $this->xind[] = $xi;
 474                 }
 475
 476                 // Find the LCS.
 477                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 478
 479                 // Merge edits when possible
 480                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 481                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 482
 483                 // Compute the edit operations.
 484                 $edits = array();
 485                 $xi = $yi = 0;
 486                 while ($xi < $n_from || $yi < $n_to) {
 487                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 488                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 489
 490                         // Skip matching "snake".
 491                         $copy = array();
 492                         while ( $xi < $n_from && $yi < $n_to
 493                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 494                                 $copy[] = $from_lines[$xi++];
 495                                 ++$yi;
 496                         }
 497                         if ($copy)
 498                                 $edits[] = new _DiffOp_Copy($copy);
 499
 500                         // Find deletes & adds.
 501                         $delete = array();
 502                         while ($xi < $n_from && $this->xchanged[$xi])
 503                                 $delete[] = $from_lines[$xi++];
 504
 505                         $add = array();
 506                         while ($yi < $n_to && $this->ychanged[$yi])
 507                                 $add[] = $to_lines[$yi++];
 508
 509                         if ($delete && $add)
 510                                 $edits[] = new _DiffOp_Change($delete, $add);
 511                         elseif ($delete)
 512                                 $edits[] = new _DiffOp_Delete($delete);
 513                         elseif ($add)
 514                                 $edits[] = new _DiffOp_Add($add);
 515                 }
 516                 wfProfileOut( $fname );
 517                 return $edits;
 518         }
 519
 520
 521         /* Divide the Largest Common Subsequence (LCS) of the sequences
 522          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 523          * sized segments.
 524          *
 525          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 526          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 527          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 528          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 529          * that (X0, Y0) == (XOFF, YOFF) and
 530          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 531          *
 532          * This function assumes that the first lines of the specified portions
 533          * of the two files do not match, and likewise that the last lines do not
 534          * match.  The caller must trim matching lines from the beginning and end
 535          * of the portions it is going to specify.
 536          */
 537         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 538                 $fname = '_DiffEngine::_diag';
 539                 wfProfileIn( $fname );
 540                 $flip = false;
 541
 542                 if ($xlim - $xoff > $ylim - $yoff) {
 543                         // Things seems faster (I'm not sure I understand why)
 544                                 // when the shortest sequence in X.
 545                                 $flip = true;
 546                         list ($xoff, $xlim, $yoff, $ylim)
 547                         = array( $yoff, $ylim, $xoff, $xlim);
 548                 }
 549
 550                 if ($flip)
 551                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 552                                 $ymatches[$this->xv[$i]][] = $i;
 553                 else
 554                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 555                                 $ymatches[$this->yv[$i]][] = $i;
 556
 557                 $this->lcs = 0;
 558                 $this->seq[0]= $yoff - 1;
 559                 $this->in_seq = array();
 560                 $ymids[0] = array();
 561
 562                 $numer = $xlim - $xoff + $nchunks - 1;
 563                 $x = $xoff;
 564                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 565                         wfProfileIn( "$fname-chunk" );
 566                         if ($chunk > 0)
 567                                 for ($i = 0; $i <= $this->lcs; $i++)
 568                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 569
 570                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 571                         for ( ; $x < $x1; $x++) {
 572                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 573                                         if (empty($ymatches[$line]))
 574                                                 continue;
 575                                 $matches = $ymatches[$line];
 576                                 reset($matches);
 577                                 while (list ($junk, $y) = each($matches))
 578                                         if (empty($this->in_seq[$y])) {
 579                                                 $k = $this->_lcs_pos($y);
 580                                                 USE_ASSERTS && assert($k > 0);
 581                                                 $ymids[$k] = $ymids[$k-1];
 582                                                 break;
 583                                         }
 584                                 while (list ($junk, $y) = each($matches)) {
 585                                         if ($y > $this->seq[$k-1]) {
 586                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 587                                                 // Optimization: this is a common case:
 588                                                 //      next match is just replacing previous match.
 589                                                 $this->in_seq[$this->seq[$k]] = false;
 590                                                 $this->seq[$k] = $y;
 591                                                 $this->in_seq[$y] = 1;
 592                                         } else if (empty($this->in_seq[$y])) {
 593                                                 $k = $this->_lcs_pos($y);
 594                                                 USE_ASSERTS && assert($k > 0);
 595                                                 $ymids[$k] = $ymids[$k-1];
 596                                         }
 597                                 }
 598                         }
 599                         wfProfileOut( "$fname-chunk" );
 600                 }
 601
 602                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 603                 $ymid = $ymids[$this->lcs];
 604                 for ($n = 0; $n < $nchunks - 1; $n++) {
 605                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 606                         $y1 = $ymid[$n] + 1;
 607                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 608                 }
 609                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 610
 611                 wfProfileOut( $fname );
 612                 return array($this->lcs, $seps);
 613         }
 614
 615         function _lcs_pos ($ypos) {
 616                 $fname = '_DiffEngine::_lcs_pos';
 617                 wfProfileIn( $fname );
 618
 619                 $end = $this->lcs;
 620                 if ($end == 0 || $ypos > $this->seq[$end]) {
 621                         $this->seq[++$this->lcs] = $ypos;
 622                         $this->in_seq[$ypos] = 1;
 623                         wfProfileOut( $fname );
 624                         return $this->lcs;
 625                 }
 626
 627                 $beg = 1;
 628                 while ($beg < $end) {
 629                         $mid = (int)(($beg + $end) / 2);
 630                         if ( $ypos > $this->seq[$mid] )
 631                                 $beg = $mid + 1;
 632                         else
 633                                 $end = $mid;
 634                 }
 635
 636                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 637
 638                 $this->in_seq[$this->seq[$end]] = false;
 639                 $this->seq[$end] = $ypos;
 640                 $this->in_seq[$ypos] = 1;
 641                 wfProfileOut( $fname );
 642                 return $end;
 643         }
 644
 645         /* Find LCS of two sequences.
 646          *
 647          * The results are recorded in the vectors $this->{x,y}changed[], by
 648          * storing a 1 in the element for each line that is an insertion
 649          * or deletion (ie. is not in the LCS).
 650          *
 651          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 652          *
 653          * Note that XLIM, YLIM are exclusive bounds.
 654          * All line numbers are origin-0 and discarded lines are not counted.
 655          */
 656         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 657                 $fname = '_DiffEngine::_compareseq';
 658                 wfProfileIn( $fname );
 659
 660                 // Slide down the bottom initial diagonal.
 661                 while ($xoff < $xlim && $yoff < $ylim
 662                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 663                         ++$xoff;
 664                         ++$yoff;
 665                 }
 666
 667                 // Slide up the top initial diagonal.
 668                 while ($xlim > $xoff && $ylim > $yoff
 669                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 670                         --$xlim;
 671                         --$ylim;
 672                 }
 673
 674                 if ($xoff == $xlim || $yoff == $ylim)
 675                         $lcs = 0;
 676                 else {
 677                         // This is ad hoc but seems to work well.
 678                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 679                         //$nchunks = max(2,min(8,(int)$nchunks));
 680                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 681                         list ($lcs, $seps)
 682                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 683                 }
 684
 685                 if ($lcs == 0) {
 686                         // X and Y sequences have no common subsequence:
 687                         // mark all changed.
 688                         while ($yoff < $ylim)
 689                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 690                         while ($xoff < $xlim)
 691                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 692                 } else {
 693                         // Use the partitions to split this problem into subproblems.
 694                         reset($seps);
 695                         $pt1 = $seps[0];
 696                         while ($pt2 = next($seps)) {
 697                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 698                                 $pt1 = $pt2;
 699                         }
 700                 }
 701                 wfProfileOut( $fname );
 702         }
 703
 704         /* Adjust inserts/deletes of identical lines to join changes
 705          * as much as possible.
 706          *
 707          * We do something when a run of changed lines include a
 708          * line at one end and has an excluded, identical line at the other.
 709          * We are free to choose which identical line is included.
 710          * `compareseq' usually chooses the one at the beginning,
 711          * but usually it is cleaner to consider the following identical line
 712          * to be the "change".
 713          *
 714          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 715          */
 716         function _shift_boundaries ($lines, &$changed, $other_changed) {
 717                 $fname = '_DiffEngine::_shift_boundaries';
 718                 wfProfileIn( $fname );
 719                 $i = 0;
 720                 $j = 0;
 721
 722                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 723                 $len = sizeof($lines);
 724                 $other_len = sizeof($other_changed);
 725
 726                 while (1) {
 727                         /*
 728                          * Scan forwards to find beginning of another run of changes.
 729                          * Also keep track of the corresponding point in the other file.
 730                          *
 731                          * Throughout this code, $i and $j are adjusted together so that
 732                          * the first $i elements of $changed and the first $j elements
 733                          * of $other_changed both contain the same number of zeros
 734                          * (unchanged lines).
 735                          * Furthermore, $j is always kept so that $j == $other_len or
 736                          * $other_changed[$j] == false.
 737                          */
 738                         while ($j < $other_len && $other_changed[$j])
 739                                 $j++;
 740
 741                         while ($i < $len && ! $changed[$i]) {
 742                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 743                                 $i++; $j++;
 744                                 while ($j < $other_len && $other_changed[$j])
 745                                         $j++;
 746                         }
 747
 748                         if ($i == $len)
 749                                 break;
 750
 751                         $start = $i;
 752
 753                         // Find the end of this run of changes.
 754                         while (++$i < $len && $changed[$i])
 755                                 continue;
 756
 757                         do {
 758                                 /*
 759                                  * Record the length of this run of changes, so that
 760                                  * we can later determine whether the run has grown.
 761                                  */
 762                                 $runlength = $i - $start;
 763
 764                                 /*
 765                                  * Move the changed region back, so long as the
 766                                  * previous unchanged line matches the last changed one.
 767                                  * This merges with previous changed regions.
 768                                  */
 769                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 770                                         $changed[--$start] = 1;
 771                                         $changed[--$i] = false;
 772                                         while ($start > 0 && $changed[$start - 1])
 773                                                 $start--;
 774                                         USE_ASSERTS && assert('$j > 0');
 775                                         while ($other_changed[--$j])
 776                                                 continue;
 777                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 778                                 }
 779
 780                                 /*
 781                                  * Set CORRESPONDING to the end of the changed run, at the last
 782                                  * point where it corresponds to a changed run in the other file.
 783                                  * CORRESPONDING == LEN means no such point has been found.
 784                                  */
 785                                 $corresponding = $j < $other_len ? $i : $len;
 786
 787                                 /*
 788                                  * Move the changed region forward, so long as the
 789                                  * first changed line matches the following unchanged one.
 790                                  * This merges with following changed regions.
 791                                  * Do this second, so that if there are no merges,
 792                                  * the changed region is moved forward as far as possible.
 793                                  */
 794                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 795                                         $changed[$start++] = false;
 796                                         $changed[$i++] = 1;
 797                                         while ($i < $len && $changed[$i])
 798                                                 $i++;
 799
 800                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 801                                         $j++;
 802                                         if ($j < $other_len && $other_changed[$j]) {
 803                                                 $corresponding = $i;
 804                                                 while ($j < $other_len && $other_changed[$j])
 805                                                         $j++;
 806                                         }
 807                                 }
 808                         } while ($runlength != $i - $start);
 809
 810                         /*
 811                          * If possible, move the fully-merged run of changes
 812                          * back to a corresponding run in the other file.
 813                          */
 814                         while ($corresponding < $i) {
 815                                 $changed[--$start] = 1;
 816                                 $changed[--$i] = 0;
 817                                 USE_ASSERTS && assert('$j > 0');
 818                                 while ($other_changed[--$j])
 819                                         continue;
 820                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 821                         }
 822                 }
 823                 wfProfileOut( $fname );
 824         }
 825 }
 826
 827 /**
 828  * Class representing a 'diff' between two sequences of strings.
 829  * @todo document
 830  * @access private
 831  */
 832 class Diff
 833 {
 834         var $edits;
 835
 836         /**
 837          * Constructor.
 838          * Computes diff between sequences of strings.
 839          *
 840          * @param $from_lines array An array of strings.
 841          *                (Typically these are lines from a file.)
 842          * @param $to_lines array An array of strings.
 843          */
 844         function Diff($from_lines, $to_lines) {
 845                 $eng = new _DiffEngine;
 846                 $this->edits = $eng->diff($from_lines, $to_lines);
 847                 //$this->_check($from_lines, $to_lines);
 848         }
 849
 850         /**
 851          * Compute reversed Diff.
 852          *
 853          * SYNOPSIS:
 854          *
 855          *      $diff = new Diff($lines1, $lines2);
 856          *      $rev = $diff->reverse();
 857          * @return object A Diff object representing the inverse of the
 858          *                                original diff.
 859          */
 860         function reverse () {
 861                 $rev = $this;
 862                 $rev->edits = array();
 863                 foreach ($this->edits as $edit) {
 864                         $rev->edits[] = $edit->reverse();
 865                 }
 866                 return $rev;
 867         }
 868
 869         /**
 870          * Check for empty diff.
 871          *
 872          * @return bool True iff two sequences were identical.
 873          */
 874         function isEmpty () {
 875                 foreach ($this->edits as $edit) {
 876                         if ($edit->type != 'copy')
 877                                 return false;
 878                 }
 879                 return true;
 880         }
 881
 882         /**
 883          * Compute the length of the Longest Common Subsequence (LCS).
 884          *
 885          * This is mostly for diagnostic purposed.
 886          *
 887          * @return int The length of the LCS.
 888          */
 889         function lcs () {
 890                 $lcs = 0;
 891                 foreach ($this->edits as $edit) {
 892                         if ($edit->type == 'copy')
 893                                 $lcs += sizeof($edit->orig);
 894                 }
 895                 return $lcs;
 896         }
 897
 898         /**
 899          * Get the original set of lines.
 900          *
 901          * This reconstructs the $from_lines parameter passed to the
 902          * constructor.
 903          *
 904          * @return array The original sequence of strings.
 905          */
 906         function orig() {
 907                 $lines = array();
 908
 909                 foreach ($this->edits as $edit) {
 910                         if ($edit->orig)
 911                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 912                 }
 913                 return $lines;
 914         }
 915
 916         /**
 917          * Get the closing set of lines.
 918          *
 919          * This reconstructs the $to_lines parameter passed to the
 920          * constructor.
 921          *
 922          * @return array The sequence of strings.
 923          */
 924         function closing() {
 925                 $lines = array();
 926
 927                 foreach ($this->edits as $edit) {
 928                         if ($edit->closing)
 929                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 930                 }
 931                 return $lines;
 932         }
 933
 934         /**
 935          * Check a Diff for validity.
 936          *
 937          * This is here only for debugging purposes.
 938          */
 939         function _check ($from_lines, $to_lines) {
 940                 $fname = 'Diff::_check';
 941                 wfProfileIn( $fname );
 942                 if (serialize($from_lines) != serialize($this->orig()))
 943                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 944                 if (serialize($to_lines) != serialize($this->closing()))
 945                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 946
 947                 $rev = $this->reverse();
 948                 if (serialize($to_lines) != serialize($rev->orig()))
 949                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 950                 if (serialize($from_lines) != serialize($rev->closing()))
 951                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 952
 953
 954                 $prevtype = 'none';
 955                 foreach ($this->edits as $edit) {
 956                         if ( $prevtype == $edit->type )
 957                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 958                         $prevtype = $edit->type;
 959                 }
 960
 961                 $lcs = $this->lcs();
 962                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 963                 wfProfileOut( $fname );
 964         }
 965 }
 966
 967 /**
 968  * FIXME: bad name.
 969  * @todo document
 970  * @access private
 971  */
 972 class MappedDiff extends Diff
 973 {
 974         /**
 975          * Constructor.
 976          *
 977          * Computes diff between sequences of strings.
 978          *
 979          * This can be used to compute things like
 980          * case-insensitve diffs, or diffs which ignore
 981          * changes in white-space.
 982          *
 983          * @param $from_lines array An array of strings.
 984          *      (Typically these are lines from a file.)
 985          *
 986          * @param $to_lines array An array of strings.
 987          *
 988          * @param $mapped_from_lines array This array should
 989          *      have the same size number of elements as $from_lines.
 990          *      The elements in $mapped_from_lines and
 991          *      $mapped_to_lines are what is actually compared
 992          *      when computing the diff.
 993          *
 994          * @param $mapped_to_lines array This array should
 995          *      have the same number of elements as $to_lines.
 996          */
 997         function MappedDiff($from_lines, $to_lines,
 998                                                 $mapped_from_lines, $mapped_to_lines) {
 999                 $fname = 'MappedDiff::MappedDiff';
1000                 wfProfileIn( $fname );
1001
1002                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1003                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1004
1005                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1006
1007                 $xi = $yi = 0;
1008                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1009                         $orig = &$this->edits[$i]->orig;
1010                         if (is_array($orig)) {
1011                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1012                                 $xi += sizeof($orig);
1013                         }
1014
1015                         $closing = &$this->edits[$i]->closing;
1016                         if (is_array($closing)) {
1017                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1018                                 $yi += sizeof($closing);
1019                         }
1020                 }
1021                 wfProfileOut( $fname );
1022         }
1023 }
1024
1025 /**
1026  * A class to format Diffs
1027  *
1028  * This class formats the diff in classic diff format.
1029  * It is intended that this class be customized via inheritance,
1030  * to obtain fancier outputs.
1031  * @todo document
1032  * @access private
1033  */
1034 class DiffFormatter
1035 {
1036         /**
1037          * Number of leading context "lines" to preserve.
1038          *
1039          * This should be left at zero for this class, but subclasses
1040          * may want to set this to other values.
1041          */
1042         var $leading_context_lines = 0;
1043
1044         /**
1045          * Number of trailing context "lines" to preserve.
1046          *
1047          * This should be left at zero for this class, but subclasses
1048          * may want to set this to other values.
1049          */
1050         var $trailing_context_lines = 0;
1051
1052         /**
1053          * Format a diff.
1054          *
1055          * @param $diff object A Diff object.
1056          * @return string The formatted output.
1057          */
1058         function format($diff) {
1059                 $fname = 'DiffFormatter::format';
1060                 wfProfileIn( $fname );
1061
1062                 $xi = $yi = 1;
1063                 $block = false;
1064                 $context = array();
1065
1066                 $nlead = $this->leading_context_lines;
1067                 $ntrail = $this->trailing_context_lines;
1068
1069                 $this->_start_diff();
1070
1071                 foreach ($diff->edits as $edit) {
1072                         if ($edit->type == 'copy') {
1073                                 if (is_array($block)) {
1074                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1075                                                 $block[] = $edit;
1076                                         }
1077                                         else{
1078                                                 if ($ntrail) {
1079                                                         $context = array_slice($edit->orig, 0, $ntrail);
1080                                                         $block[] = new _DiffOp_Copy($context);
1081                                                 }
1082                                                 $this->_block($x0, $ntrail + $xi - $x0,
1083                                                                           $y0, $ntrail + $yi - $y0,
1084                                                                           $block);
1085                                                 $block = false;
1086                                         }
1087                                 }
1088                                 $context = $edit->orig;
1089                         }
1090                         else {
1091                                 if (! is_array($block)) {
1092                                         $context = array_slice($context, sizeof($context) - $nlead);
1093                                         $x0 = $xi - sizeof($context);
1094                                         $y0 = $yi - sizeof($context);
1095                                         $block = array();
1096                                         if ($context)
1097                                                 $block[] = new _DiffOp_Copy($context);
1098                                 }
1099                                 $block[] = $edit;
1100                         }
1101
1102                         if ($edit->orig)
1103                                 $xi += sizeof($edit->orig);
1104                         if ($edit->closing)
1105                                 $yi += sizeof($edit->closing);
1106                 }
1107
1108                 if (is_array($block))
1109                         $this->_block($x0, $xi - $x0,
1110                                                   $y0, $yi - $y0,
1111                                                   $block);
1112
1113                 $end = $this->_end_diff();
1114                 wfProfileOut( $fname );
1115                 return $end;
1116         }
1117
1118         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1119                 $fname = 'DiffFormatter::_block';
1120                 wfProfileIn( $fname );
1121                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1122                 foreach ($edits as $edit) {
1123                         if ($edit->type == 'copy')
1124                                 $this->_context($edit->orig);
1125                         elseif ($edit->type == 'add')
1126                                 $this->_added($edit->closing);
1127                         elseif ($edit->type == 'delete')
1128                                 $this->_deleted($edit->orig);
1129                         elseif ($edit->type == 'change')
1130                                 $this->_changed($edit->orig, $edit->closing);
1131                         else
1132                                 trigger_error('Unknown edit type', E_USER_ERROR);
1133                 }
1134                 $this->_end_block();
1135                 wfProfileOut( $fname );
1136         }
1137
1138         function _start_diff() {
1139                 ob_start();
1140         }
1141
1142         function _end_diff() {
1143                 $val = ob_get_contents();
1144                 ob_end_clean();
1145                 return $val;
1146         }
1147
1148         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1149                 if ($xlen > 1)
1150                         $xbeg .= "," . ($xbeg + $xlen - 1);
1151                 if ($ylen > 1)
1152                         $ybeg .= "," . ($ybeg + $ylen - 1);
1153
1154                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1155         }
1156
1157         function _start_block($header) {
1158                 echo $header;
1159         }
1160
1161         function _end_block() {
1162         }
1163
1164         function _lines($lines, $prefix = ' ') {
1165                 foreach ($lines as $line)
1166                         echo "$prefix $line\n";
1167         }
1168
1169         function _context($lines) {
1170                 $this->_lines($lines);
1171         }
1172
1173         function _added($lines) {
1174                 $this->_lines($lines, '>');
1175         }
1176         function _deleted($lines) {
1177                 $this->_lines($lines, '<');
1178         }
1179
1180         function _changed($orig, $closing) {
1181                 $this->_deleted($orig);
1182                 echo "---\n";
1183                 $this->_added($closing);
1184         }
1185 }
1186
1187
1188 /**
1189  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1190  *
1191  */
1192
1193 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1194
1195 /**
1196  * @todo document
1197  * @access private
1198  */
1199 class _HWLDF_WordAccumulator {
1200         function _HWLDF_WordAccumulator () {
1201                 $this->_lines = array();
1202                 $this->_line = '';
1203                 $this->_group = '';
1204                 $this->_tag = '';
1205         }
1206
1207         function _flushGroup ($new_tag) {
1208                 if ($this->_group !== '') {
1209                         if ($this->_tag == 'mark')
1210                                 $this->_line .= '<span class="diffchange">' .
1211                                         htmlspecialchars ( $this->_group ) . '</span>';
1212                         else
1213                                 $this->_line .= htmlspecialchars ( $this->_group );
1214                 }
1215                 $this->_group = '';
1216                 $this->_tag = $new_tag;
1217         }
1218
1219         function _flushLine ($new_tag) {
1220                 $this->_flushGroup($new_tag);
1221                 if ($this->_line != '')
1222                         array_push ( $this->_lines, $this->_line );
1223                 else
1224                         # make empty lines visible by inserting an NBSP
1225                         array_push ( $this->_lines, NBSP );
1226                 $this->_line = '';
1227         }
1228
1229         function addWords ($words, $tag = '') {
1230                 if ($tag != $this->_tag)
1231                         $this->_flushGroup($tag);
1232
1233                 foreach ($words as $word) {
1234                         // new-line should only come as first char of word.
1235                         if ($word == '')
1236                                 continue;
1237                         if ($word[0] == "\n") {
1238                                 $this->_flushLine($tag);
1239                                 $word = substr($word, 1);
1240                         }
1241                         assert(!strstr($word, "\n"));
1242                         $this->_group .= $word;
1243                 }
1244         }
1245
1246         function getLines() {
1247                 $this->_flushLine('~done');
1248                 return $this->_lines;
1249         }
1250 }
1251
1252 /**
1253  * @todo document
1254  * @access private
1255  */
1256 class WordLevelDiff extends MappedDiff
1257 {
1258         function WordLevelDiff ($orig_lines, $closing_lines) {
1259                 $fname = 'WordLevelDiff::WordLevelDiff';
1260                 wfProfileIn( $fname );
1261
1262                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1263                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1264
1265
1266                 $this->MappedDiff($orig_words, $closing_words,
1267                                                   $orig_stripped, $closing_stripped);
1268                 wfProfileOut( $fname );
1269         }
1270
1271         function _split($lines) {
1272                 $fname = 'WordLevelDiff::_split';
1273                 wfProfileIn( $fname );
1274                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1275                                                         implode("\n", $lines),
1276                                                         $m)) {
1277                         wfProfileOut( $fname );
1278                         return array(array(''), array(''));
1279                 }
1280                 wfProfileOut( $fname );
1281                 return array($m[0], $m[1]);
1282         }
1283
1284         function orig () {
1285                 $fname = 'WordLevelDiff::orig';
1286                 wfProfileIn( $fname );
1287                 $orig = new _HWLDF_WordAccumulator;
1288
1289                 foreach ($this->edits as $edit) {
1290                         if ($edit->type == 'copy')
1291                                 $orig->addWords($edit->orig);
1292                         elseif ($edit->orig)
1293                                 $orig->addWords($edit->orig, 'mark');
1294                 }
1295                 $lines = $orig->getLines();
1296                 wfProfileOut( $fname );
1297                 return $lines;
1298         }
1299
1300         function closing () {
1301                 $fname = 'WordLevelDiff::closing';
1302                 wfProfileIn( $fname );
1303                 $closing = new _HWLDF_WordAccumulator;
1304
1305                 foreach ($this->edits as $edit) {
1306                         if ($edit->type == 'copy')
1307                                 $closing->addWords($edit->closing);
1308                         elseif ($edit->closing)
1309                                 $closing->addWords($edit->closing, 'mark');
1310                 }
1311                 $lines = $closing->getLines();
1312                 wfProfileOut( $fname );
1313                 return $lines;
1314         }
1315 }
1316
1317 /**
1318  *      Wikipedia Table style diff formatter.
1319  * @todo document
1320  * @access private
1321  */
1322 class TableDiffFormatter extends DiffFormatter
1323 {
1324         function TableDiffFormatter() {
1325                 $this->leading_context_lines = 2;
1326                 $this->trailing_context_lines = 2;
1327         }
1328
1329         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1330                 $l1 = wfMsg( 'lineno', $xbeg );
1331                 $l2 = wfMsg( 'lineno', $ybeg );
1332
1333                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1334                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1335                 return $r;
1336         }
1337
1338         function _start_block( $header ) {
1339                 global $wgOut;
1340                 echo $header;
1341         }
1342
1343         function _end_block() {
1344         }
1345
1346         function _lines( $lines, $prefix=' ', $color='white' ) {
1347         }
1348
1349         # HTML-escape parameter before calling this
1350         function addedLine( $line ) {
1351                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1352         }
1353
1354         # HTML-escape parameter before calling this
1355         function deletedLine( $line ) {
1356                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1357         }
1358
1359         # HTML-escape parameter before calling this
1360         function contextLine( $line ) {
1361                 return "<td> </td><td class='diff-context'>{$line}</td>";
1362         }
1363
1364         function emptyLine() {
1365                 return '<td colspan="2">&nbsp;</td>';
1366         }
1367
1368         function _added( $lines ) {
1369                 foreach ($lines as $line) {
1370                         echo '<tr>' . $this->emptyLine() .
1371                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1372                 }
1373         }
1374
1375         function _deleted($lines) {
1376                 foreach ($lines as $line) {
1377                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1378                           $this->emptyLine() . "</tr>\n";
1379                 }
1380         }
1381
1382         function _context( $lines ) {
1383                 foreach ($lines as $line) {
1384                         echo '<tr>' .
1385                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1386                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1387                 }
1388         }
1389
1390         function _changed( $orig, $closing ) {
1391                 $fname = 'TableDiffFormatter::_changed';
1392                 wfProfileIn( $fname );
1393
1394                 $diff = new WordLevelDiff( $orig, $closing );
1395                 $del = $diff->orig();
1396                 $add = $diff->closing();
1397
1398                 # Notice that WordLevelDiff returns HTML-escaped output.
1399                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1400
1401                 while ( $line = array_shift( $del ) ) {
1402                         $aline = array_shift( $add );
1403                         echo '<tr>' . $this->deletedLine( $line ) .
1404                                 $this->addedLine( $aline ) . "</tr>\n";
1405                 }
1406                 foreach ($add as $line) {       # If any leftovers
1407                         echo '<tr>' . $this->emptyLine() .
1408                                 $this->addedLine( $line ) . "</tr>\n";
1409                 }
1410                 wfProfileOut( $fname );
1411         }
1412 }
1413
1414 ?>