includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @package MediaWiki
   5  * @subpackage DifferenceEngine
   6  */
   7
   8 /** */
   9 require_once( 'Revision.php' );
  10
  11 /**
  12  * @todo document
  13  * @access public
  14  * @package MediaWiki
  15  * @subpackage DifferenceEngine
  16  */
  17 class DifferenceEngine {
  18         /**#@+
  19          * @access private
  20          */
  21         var $mOldid, $mNewid;
  22         var $mOldtitle, $mNewtitle, $mPagetitle;
  23         var $mOldtext, $mNewtext;
  24         var $mOldUser, $mNewUser;
  25         var $mOldComment, $mNewComment;
  26         var $mOldPage, $mNewPage;
  27         var $mRcidMarkPatrolled;
  28         /**#@-*/
  29
  30         /**
  31          * Constructor
  32          * @param integer $old Old ID we want to show and diff with.
  33          * @param string $new Either 'prev' or 'next'.
  34          * @param integer $rcid ??? (default 0)
  35          */
  36         function DifferenceEngine( $old, $new, $rcid = 0 ) {
  37                 global $wgTitle;
  38
  39                 if ( 'prev' == $new ) {
  40                         # Show diff between revision $old and the previous one.
  41                         # Get previous one from DB.
  42                         #
  43                         $this->mNewid = intval($old);
  44
  45                         $this->mOldid = $wgTitle->getPreviousRevisionID( $this->mNewid );
  46
  47                 } elseif ( 'next' == $new ) {
  48                         # Show diff between revision $old and the previous one.
  49                         # Get previous one from DB.
  50                         #
  51                         $this->mOldid = intval($old);
  52                         $this->mNewid = $wgTitle->getNextRevisionID( $this->mOldid );
  53                         if ( false === $this->mNewid ) {
  54                                 # if no result, NewId points to the newest old revision. The only newer
  55                                 # revision is cur, which is "0".
  56                                 $this->mNewid = 0;
  57                         }
  58
  59                 } else {
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = intval($new);
  62                 }
  63                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  64         }
  65
  66         function showDiffPage() {
  67                 global $wgUser, $wgTitle, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol, $wgUseRCPatrol;
  68                 $fname = 'DifferenceEngine::showDiffPage';
  69                 wfProfileIn( $fname );
  70
  71                 # mOldid is false if the difference engine is called with a "vague" query for
  72                 # a diff between a version V and its previous version V' AND the version V
  73                 # is the first version of that article. In that case, V' does not exist.
  74                 if ( $this->mOldid === false ) {
  75                         $this->showFirstRevision();
  76                         wfProfileOut( $fname );
  77                         return;
  78                 }
  79
  80                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  81                   "{$this->mNewid})";
  82                 $mtext = wfMsg( 'missingarticle', $t );
  83
  84                 $wgOut->setArticleFlag( false );
  85                 if ( ! $this->loadText() ) {
  86                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
  87                         $wgOut->addHTML( $mtext );
  88                         wfProfileOut( $fname );
  89                         return;
  90                 }
  91                 $wgOut->suppressQuickbar();
  92
  93                 $oldTitle = $this->mOldPage->getPrefixedText();
  94                 $newTitle = $this->mNewPage->getPrefixedText();
  95                 if( $oldTitle == $newTitle ) {
  96                         $wgOut->setPageTitle( $newTitle );
  97                 } else {
  98                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
  99                 }
 100                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 101                 $wgOut->setRobotpolicy( 'noindex,follow' );
 102
 103                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 104                         $wgOut->loginToUse();
 105                         $wgOut->output();
 106                         wfProfileOut( $fname );
 107                         exit;
 108                 }
 109
 110                 $sk = $wgUser->getSkin();
 111                 $talk = $wgContLang->getNsText( NS_TALK );
 112                 $contribs = wfMsg( 'contribslink' );
 113
 114                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 115                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 116
 117                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 118                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 119                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 120                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 121                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 122                         'target=' . urlencode($this->mOldUser) );
 123                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 124                         'target=' . urlencode($this->mNewUser) );
 125                 if ( !$this->mNewid && $wgUser->isAllowed('rollback') ) {
 126                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
 127                                 'action=rollback&from=' . urlencode($this->mNewUser) .
 128                                 '&token=' . urlencode( $wgUser->editToken( array( $wgTitle->getPrefixedText(), $this->mNewUser ) ) ) ) .
 129                                 ']</strong>';
 130                 } else {
 131                         $rollback = '';
 132                 }
 133                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isLoggedIn() &&
 134                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 135                 {
 136                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 137                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 138                 } else {
 139                         $patrol = '';
 140                 }
 141
 142                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'previousdiff' ), 'diff=prev&oldid='.$this->mOldid );
 143                 if ( $this->mNewid == 0 ) {
 144                         $nextlink = '';
 145                 } else {
 146                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 147                 }
 148
 149                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 150                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 151                         '<br />' . $prevlink;
 152                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 153                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 154                         '<br />' . $nextlink . $patrol;
 155
 156                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 157                   $oldHeader, $newHeader );
 158                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 159                 $wgOut->addWikiText( $this->mNewtext );
 160
 161                 wfProfileOut( $fname );
 162         }
 163
 164         /**
 165          * Show the first revision of an article. Uses normal diff headers in
 166          * contrast to normal "old revision" display style.
 167          */
 168         function showFirstRevision() {
 169                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 170
 171                 $fname = 'DifferenceEngine::showFirstRevision';
 172                 wfProfileIn( $fname );
 173
 174
 175                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 176
 177                 # Get article text from the DB
 178                 #
 179                 if ( ! $this->loadText() ) {
 180                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 181                           "{$this->mNewid})";
 182                         $mtext = wfMsg( 'missingarticle', $t );
 183                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 184                         $wgOut->addHTML( $mtext );
 185                         wfProfileOut( $fname );
 186                         return;
 187                 }
 188
 189                 # Check if user is allowed to look at this page. If not, bail out.
 190                 #
 191                 if ( !( $this->mOldPage->userCanRead() ) ) {
 192                         $wgOut->loginToUse();
 193                         $wgOut->output();
 194                         wfProfileOut( $fname );
 195                         exit;
 196                 }
 197
 198                 # Prepare the header box
 199                 #
 200                 $sk = $wgUser->getSkin();
 201
 202                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 203                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 204                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 205                         'target=' . urlencode($this->mOldUser) );
 206                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 207                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 208                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 209                         '<br />' . $nextlink. "</div>\n";
 210
 211                 $wgOut->addHTML( $header );
 212
 213                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 214                 $wgOut->setRobotpolicy( 'noindex,follow' );
 215
 216
 217                 # Show current revision
 218                 #
 219                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 220                 $wgOut->addWikiText( $this->mNewtext );
 221
 222                 wfProfileOut( $fname );
 223         }
 224
 225         function showDiff( $otext, $ntext, $otitle, $ntitle ) {
 226                 global $wgOut;
 227                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 228         }
 229
 230         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 231                 global $wgUseExternalDiffEngine, $wgContLang;
 232                 $out = "
 233                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 234                         <tr>
 235                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 236                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 237                         </tr>
 238                 ";
 239                 $otext = $wgContLang->segmentForDiff($otext);
 240                 $ntext = $wgContLang->segmentForDiff($ntext);
 241                 $difftext='';
 242                 if ( $wgUseExternalDiffEngine ) {
 243                         # For historical reasons, external diff engine expects
 244                         # input text to be HTML-escaped already
 245                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 246                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 247                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 248                                 dl('php_wikidiff.so');
 249                         }
 250                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 251                 } else {
 252                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 253                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 254                         $diffs =& new Diff( $ota, $nta );
 255                         $formatter =& new TableDiffFormatter();
 256                         $difftext = $formatter->format( $diffs );
 257                 }
 258                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 259                 $out .= $difftext."</table>\n";
 260                 return $out;
 261         }
 262
 263         /**
 264          * Load the text of the articles to compare.  If newid is 0, then compare
 265          * the old article in oldid to the current article; if oldid is 0, then
 266          * compare the current article to the immediately previous one (ignoring the
 267          * value of newid).
 268          */
 269         function loadText() {
 270                 global $wgTitle, $wgOut, $wgLang;
 271                 $fname = 'DifferenceEngine::loadText';
 272
 273                 $dbr =& wfGetDB( DB_SLAVE );
 274                 if( $this->mNewid ) {
 275                         $this->newRev =& Revision::newFromId( $this->mNewid );
 276                 } else {
 277                         $this->newRev =& Revision::newFromTitle( $wgTitle );
 278                 }
 279                 if( is_null( $this->newRev ) ) {
 280                         return false;
 281                 }
 282
 283                 if( $this->newRev->isCurrent() ) {
 284                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 285                         $this->mNewPage = $wgTitle;
 286                         $newLink = $this->mNewPage->escapeLocalUrl();
 287                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 288                 } else {
 289                         $this->mNewPage = $this->newRev->getTitle();
 290                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 291                         $t = $wgLang->timeanddate( $this->newRev->getTimestamp(), true );
 292                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 293                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 294                 }
 295
 296                 if( $this->mOldid ) {
 297                         $this->oldRev =& Revision::newFromId( $this->mOldid );
 298                 } else {
 299                         $this->oldRev =& $this->newRev->getPrevious();
 300                         $this->mOldid = $this->oldRev->getId();
 301                 }
 302                 if( is_null( $this->oldRev ) ) {
 303                         return false;
 304                 }
 305
 306                 $this->mOldPage = $this->oldRev->getTitle();
 307
 308                 $t = $wgLang->timeanddate( $this->oldRev->getTimestamp(), true );
 309                 $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 310                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 311
 312                 $this->mNewUser = $this->newRev->getUserText();
 313                 $this->mNewComment = $this->newRev->getComment();
 314                 $this->mNewtext = $this->newRev->getText();
 315
 316                 $this->mOldUser = $this->oldRev->getUserText();
 317                 $this->mOldComment = $this->oldRev->getComment();
 318                 $this->mOldtext = $this->oldRev->getText();
 319
 320                 return true;
 321         }
 322 }
 323
 324 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 325 //
 326 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 327 // You may copy this code freely under the conditions of the GPL.
 328 //
 329
 330 define('USE_ASSERTS', function_exists('assert'));
 331
 332 /**
 333  * @todo document
 334  * @access private
 335  * @package MediaWiki
 336  * @subpackage DifferenceEngine
 337  */
 338 class _DiffOp {
 339         var $type;
 340         var $orig;
 341         var $closing;
 342
 343         function reverse() {
 344                 trigger_error('pure virtual', E_USER_ERROR);
 345         }
 346
 347         function norig() {
 348                 return $this->orig ? sizeof($this->orig) : 0;
 349         }
 350
 351         function nclosing() {
 352                 return $this->closing ? sizeof($this->closing) : 0;
 353         }
 354 }
 355
 356 /**
 357  * @todo document
 358  * @access private
 359  * @package MediaWiki
 360  * @subpackage DifferenceEngine
 361  */
 362 class _DiffOp_Copy extends _DiffOp {
 363         var $type = 'copy';
 364
 365         function _DiffOp_Copy ($orig, $closing = false) {
 366                 if (!is_array($closing))
 367                         $closing = $orig;
 368                 $this->orig = $orig;
 369                 $this->closing = $closing;
 370         }
 371
 372         function reverse() {
 373                 return new _DiffOp_Copy($this->closing, $this->orig);
 374         }
 375 }
 376
 377 /**
 378  * @todo document
 379  * @access private
 380  * @package MediaWiki
 381  * @subpackage DifferenceEngine
 382  */
 383 class _DiffOp_Delete extends _DiffOp {
 384         var $type = 'delete';
 385
 386         function _DiffOp_Delete ($lines) {
 387                 $this->orig = $lines;
 388                 $this->closing = false;
 389         }
 390
 391         function reverse() {
 392                 return new _DiffOp_Add($this->orig);
 393         }
 394 }
 395
 396 /**
 397  * @todo document
 398  * @access private
 399  * @package MediaWiki
 400  * @subpackage DifferenceEngine
 401  */
 402 class _DiffOp_Add extends _DiffOp {
 403         var $type = 'add';
 404
 405         function _DiffOp_Add ($lines) {
 406                 $this->closing = $lines;
 407                 $this->orig = false;
 408         }
 409
 410         function reverse() {
 411                 return new _DiffOp_Delete($this->closing);
 412         }
 413 }
 414
 415 /**
 416  * @todo document
 417  * @access private
 418  * @package MediaWiki
 419  * @subpackage DifferenceEngine
 420  */
 421 class _DiffOp_Change extends _DiffOp {
 422         var $type = 'change';
 423
 424         function _DiffOp_Change ($orig, $closing) {
 425                 $this->orig = $orig;
 426                 $this->closing = $closing;
 427         }
 428
 429         function reverse() {
 430                 return new _DiffOp_Change($this->closing, $this->orig);
 431         }
 432 }
 433
 434
 435 /**
 436  * Class used internally by Diff to actually compute the diffs.
 437  *
 438  * The algorithm used here is mostly lifted from the perl module
 439  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 440  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 441  *
 442  * More ideas are taken from:
 443  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 444  *
 445  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 446  * diffutils-2.7, which can be found at:
 447  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 448  *
 449  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 450  * are my own.
 451  *
 452  * @author Geoffrey T. Dairiki
 453  * @access private
 454  * @package MediaWiki
 455  * @subpackage DifferenceEngine
 456  */
 457 class _DiffEngine
 458 {
 459         function diff ($from_lines, $to_lines) {
 460                 $fname = '_DiffEngine::diff';
 461                 wfProfileIn( $fname );
 462
 463                 $n_from = sizeof($from_lines);
 464                 $n_to = sizeof($to_lines);
 465
 466                 $this->xchanged = $this->ychanged = array();
 467                 $this->xv = $this->yv = array();
 468                 $this->xind = $this->yind = array();
 469                 unset($this->seq);
 470                 unset($this->in_seq);
 471                 unset($this->lcs);
 472
 473                 // Skip leading common lines.
 474                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 475                         if ($from_lines[$skip] != $to_lines[$skip])
 476                                 break;
 477                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 478                 }
 479                 // Skip trailing common lines.
 480                 $xi = $n_from; $yi = $n_to;
 481                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 482                         if ($from_lines[$xi] != $to_lines[$yi])
 483                                 break;
 484                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 485                 }
 486
 487                 // Ignore lines which do not exist in both files.
 488                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 489                         $xhash[$from_lines[$xi]] = 1;
 490                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 491                         $line = $to_lines[$yi];
 492                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 493                                 continue;
 494                         $yhash[$line] = 1;
 495                         $this->yv[] = $line;
 496                         $this->yind[] = $yi;
 497                 }
 498                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 499                         $line = $from_lines[$xi];
 500                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 501                                 continue;
 502                         $this->xv[] = $line;
 503                         $this->xind[] = $xi;
 504                 }
 505
 506                 // Find the LCS.
 507                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 508
 509                 // Merge edits when possible
 510                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 511                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 512
 513                 // Compute the edit operations.
 514                 $edits = array();
 515                 $xi = $yi = 0;
 516                 while ($xi < $n_from || $yi < $n_to) {
 517                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 518                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 519
 520                         // Skip matching "snake".
 521                         $copy = array();
 522                         while ( $xi < $n_from && $yi < $n_to
 523                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 524                                 $copy[] = $from_lines[$xi++];
 525                                 ++$yi;
 526                         }
 527                         if ($copy)
 528                                 $edits[] = new _DiffOp_Copy($copy);
 529
 530                         // Find deletes & adds.
 531                         $delete = array();
 532                         while ($xi < $n_from && $this->xchanged[$xi])
 533                                 $delete[] = $from_lines[$xi++];
 534
 535                         $add = array();
 536                         while ($yi < $n_to && $this->ychanged[$yi])
 537                                 $add[] = $to_lines[$yi++];
 538
 539                         if ($delete && $add)
 540                                 $edits[] = new _DiffOp_Change($delete, $add);
 541                         elseif ($delete)
 542                                 $edits[] = new _DiffOp_Delete($delete);
 543                         elseif ($add)
 544                                 $edits[] = new _DiffOp_Add($add);
 545                 }
 546                 wfProfileOut( $fname );
 547                 return $edits;
 548         }
 549
 550
 551         /* Divide the Largest Common Subsequence (LCS) of the sequences
 552          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 553          * sized segments.
 554          *
 555          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 556          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 557          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 558          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 559          * that (X0, Y0) == (XOFF, YOFF) and
 560          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 561          *
 562          * This function assumes that the first lines of the specified portions
 563          * of the two files do not match, and likewise that the last lines do not
 564          * match.  The caller must trim matching lines from the beginning and end
 565          * of the portions it is going to specify.
 566          */
 567         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 568                 $fname = '_DiffEngine::_diag';
 569                 wfProfileIn( $fname );
 570                 $flip = false;
 571
 572                 if ($xlim - $xoff > $ylim - $yoff) {
 573                         // Things seems faster (I'm not sure I understand why)
 574                                 // when the shortest sequence in X.
 575                                 $flip = true;
 576                         list ($xoff, $xlim, $yoff, $ylim)
 577                         = array( $yoff, $ylim, $xoff, $xlim);
 578                 }
 579
 580                 if ($flip)
 581                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 582                                 $ymatches[$this->xv[$i]][] = $i;
 583                 else
 584                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 585                                 $ymatches[$this->yv[$i]][] = $i;
 586
 587                 $this->lcs = 0;
 588                 $this->seq[0]= $yoff - 1;
 589                 $this->in_seq = array();
 590                 $ymids[0] = array();
 591
 592                 $numer = $xlim - $xoff + $nchunks - 1;
 593                 $x = $xoff;
 594                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 595                         wfProfileIn( "$fname-chunk" );
 596                         if ($chunk > 0)
 597                                 for ($i = 0; $i <= $this->lcs; $i++)
 598                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 599
 600                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 601                         for ( ; $x < $x1; $x++) {
 602                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 603                                         if (empty($ymatches[$line]))
 604                                                 continue;
 605                                 $matches = $ymatches[$line];
 606                                 reset($matches);
 607                                 while (list ($junk, $y) = each($matches))
 608                                         if (empty($this->in_seq[$y])) {
 609                                                 $k = $this->_lcs_pos($y);
 610                                                 USE_ASSERTS && assert($k > 0);
 611                                                 $ymids[$k] = $ymids[$k-1];
 612                                                 break;
 613                                         }
 614                                 while (list ($junk, $y) = each($matches)) {
 615                                         if ($y > $this->seq[$k-1]) {
 616                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 617                                                 // Optimization: this is a common case:
 618                                                 //      next match is just replacing previous match.
 619                                                 $this->in_seq[$this->seq[$k]] = false;
 620                                                 $this->seq[$k] = $y;
 621                                                 $this->in_seq[$y] = 1;
 622                                         } else if (empty($this->in_seq[$y])) {
 623                                                 $k = $this->_lcs_pos($y);
 624                                                 USE_ASSERTS && assert($k > 0);
 625                                                 $ymids[$k] = $ymids[$k-1];
 626                                         }
 627                                 }
 628                         }
 629                         wfProfileOut( "$fname-chunk" );
 630                 }
 631
 632                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 633                 $ymid = $ymids[$this->lcs];
 634                 for ($n = 0; $n < $nchunks - 1; $n++) {
 635                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 636                         $y1 = $ymid[$n] + 1;
 637                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 638                 }
 639                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 640
 641                 wfProfileOut( $fname );
 642                 return array($this->lcs, $seps);
 643         }
 644
 645         function _lcs_pos ($ypos) {
 646                 $fname = '_DiffEngine::_lcs_pos';
 647                 wfProfileIn( $fname );
 648
 649                 $end = $this->lcs;
 650                 if ($end == 0 || $ypos > $this->seq[$end]) {
 651                         $this->seq[++$this->lcs] = $ypos;
 652                         $this->in_seq[$ypos] = 1;
 653                         wfProfileOut( $fname );
 654                         return $this->lcs;
 655                 }
 656
 657                 $beg = 1;
 658                 while ($beg < $end) {
 659                         $mid = (int)(($beg + $end) / 2);
 660                         if ( $ypos > $this->seq[$mid] )
 661                                 $beg = $mid + 1;
 662                         else
 663                                 $end = $mid;
 664                 }
 665
 666                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 667
 668                 $this->in_seq[$this->seq[$end]] = false;
 669                 $this->seq[$end] = $ypos;
 670                 $this->in_seq[$ypos] = 1;
 671                 wfProfileOut( $fname );
 672                 return $end;
 673         }
 674
 675         /* Find LCS of two sequences.
 676          *
 677          * The results are recorded in the vectors $this->{x,y}changed[], by
 678          * storing a 1 in the element for each line that is an insertion
 679          * or deletion (ie. is not in the LCS).
 680          *
 681          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 682          *
 683          * Note that XLIM, YLIM are exclusive bounds.
 684          * All line numbers are origin-0 and discarded lines are not counted.
 685          */
 686         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 687                 $fname = '_DiffEngine::_compareseq';
 688                 wfProfileIn( $fname );
 689
 690                 // Slide down the bottom initial diagonal.
 691                 while ($xoff < $xlim && $yoff < $ylim
 692                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 693                         ++$xoff;
 694                         ++$yoff;
 695                 }
 696
 697                 // Slide up the top initial diagonal.
 698                 while ($xlim > $xoff && $ylim > $yoff
 699                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 700                         --$xlim;
 701                         --$ylim;
 702                 }
 703
 704                 if ($xoff == $xlim || $yoff == $ylim)
 705                         $lcs = 0;
 706                 else {
 707                         // This is ad hoc but seems to work well.
 708                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 709                         //$nchunks = max(2,min(8,(int)$nchunks));
 710                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 711                         list ($lcs, $seps)
 712                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 713                 }
 714
 715                 if ($lcs == 0) {
 716                         // X and Y sequences have no common subsequence:
 717                         // mark all changed.
 718                         while ($yoff < $ylim)
 719                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 720                         while ($xoff < $xlim)
 721                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 722                 } else {
 723                         // Use the partitions to split this problem into subproblems.
 724                         reset($seps);
 725                         $pt1 = $seps[0];
 726                         while ($pt2 = next($seps)) {
 727                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 728                                 $pt1 = $pt2;
 729                         }
 730                 }
 731                 wfProfileOut( $fname );
 732         }
 733
 734         /* Adjust inserts/deletes of identical lines to join changes
 735          * as much as possible.
 736          *
 737          * We do something when a run of changed lines include a
 738          * line at one end and has an excluded, identical line at the other.
 739          * We are free to choose which identical line is included.
 740          * `compareseq' usually chooses the one at the beginning,
 741          * but usually it is cleaner to consider the following identical line
 742          * to be the "change".
 743          *
 744          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 745          */
 746         function _shift_boundaries ($lines, &$changed, $other_changed) {
 747                 $fname = '_DiffEngine::_shift_boundaries';
 748                 wfProfileIn( $fname );
 749                 $i = 0;
 750                 $j = 0;
 751
 752                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 753                 $len = sizeof($lines);
 754                 $other_len = sizeof($other_changed);
 755
 756                 while (1) {
 757                         /*
 758                          * Scan forwards to find beginning of another run of changes.
 759                          * Also keep track of the corresponding point in the other file.
 760                          *
 761                          * Throughout this code, $i and $j are adjusted together so that
 762                          * the first $i elements of $changed and the first $j elements
 763                          * of $other_changed both contain the same number of zeros
 764                          * (unchanged lines).
 765                          * Furthermore, $j is always kept so that $j == $other_len or
 766                          * $other_changed[$j] == false.
 767                          */
 768                         while ($j < $other_len && $other_changed[$j])
 769                                 $j++;
 770
 771                         while ($i < $len && ! $changed[$i]) {
 772                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 773                                 $i++; $j++;
 774                                 while ($j < $other_len && $other_changed[$j])
 775                                         $j++;
 776                         }
 777
 778                         if ($i == $len)
 779                                 break;
 780
 781                         $start = $i;
 782
 783                         // Find the end of this run of changes.
 784                         while (++$i < $len && $changed[$i])
 785                                 continue;
 786
 787                         do {
 788                                 /*
 789                                  * Record the length of this run of changes, so that
 790                                  * we can later determine whether the run has grown.
 791                                  */
 792                                 $runlength = $i - $start;
 793
 794                                 /*
 795                                  * Move the changed region back, so long as the
 796                                  * previous unchanged line matches the last changed one.
 797                                  * This merges with previous changed regions.
 798                                  */
 799                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 800                                         $changed[--$start] = 1;
 801                                         $changed[--$i] = false;
 802                                         while ($start > 0 && $changed[$start - 1])
 803                                                 $start--;
 804                                         USE_ASSERTS && assert('$j > 0');
 805                                         while ($other_changed[--$j])
 806                                                 continue;
 807                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 808                                 }
 809
 810                                 /*
 811                                  * Set CORRESPONDING to the end of the changed run, at the last
 812                                  * point where it corresponds to a changed run in the other file.
 813                                  * CORRESPONDING == LEN means no such point has been found.
 814                                  */
 815                                 $corresponding = $j < $other_len ? $i : $len;
 816
 817                                 /*
 818                                  * Move the changed region forward, so long as the
 819                                  * first changed line matches the following unchanged one.
 820                                  * This merges with following changed regions.
 821                                  * Do this second, so that if there are no merges,
 822                                  * the changed region is moved forward as far as possible.
 823                                  */
 824                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 825                                         $changed[$start++] = false;
 826                                         $changed[$i++] = 1;
 827                                         while ($i < $len && $changed[$i])
 828                                                 $i++;
 829
 830                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 831                                         $j++;
 832                                         if ($j < $other_len && $other_changed[$j]) {
 833                                                 $corresponding = $i;
 834                                                 while ($j < $other_len && $other_changed[$j])
 835                                                         $j++;
 836                                         }
 837                                 }
 838                         } while ($runlength != $i - $start);
 839
 840                         /*
 841                          * If possible, move the fully-merged run of changes
 842                          * back to a corresponding run in the other file.
 843                          */
 844                         while ($corresponding < $i) {
 845                                 $changed[--$start] = 1;
 846                                 $changed[--$i] = 0;
 847                                 USE_ASSERTS && assert('$j > 0');
 848                                 while ($other_changed[--$j])
 849                                         continue;
 850                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 851                         }
 852                 }
 853                 wfProfileOut( $fname );
 854         }
 855 }
 856
 857 /**
 858  * Class representing a 'diff' between two sequences of strings.
 859  * @todo document
 860  * @access private
 861  * @package MediaWiki
 862  * @subpackage DifferenceEngine
 863  */
 864 class Diff
 865 {
 866         var $edits;
 867
 868         /**
 869          * Constructor.
 870          * Computes diff between sequences of strings.
 871          *
 872          * @param $from_lines array An array of strings.
 873          *                (Typically these are lines from a file.)
 874          * @param $to_lines array An array of strings.
 875          */
 876         function Diff($from_lines, $to_lines) {
 877                 $eng = new _DiffEngine;
 878                 $this->edits = $eng->diff($from_lines, $to_lines);
 879                 //$this->_check($from_lines, $to_lines);
 880         }
 881
 882         /**
 883          * Compute reversed Diff.
 884          *
 885          * SYNOPSIS:
 886          *
 887          *      $diff = new Diff($lines1, $lines2);
 888          *      $rev = $diff->reverse();
 889          * @return object A Diff object representing the inverse of the
 890          *                                original diff.
 891          */
 892         function reverse () {
 893                 $rev = $this;
 894                 $rev->edits = array();
 895                 foreach ($this->edits as $edit) {
 896                         $rev->edits[] = $edit->reverse();
 897                 }
 898                 return $rev;
 899         }
 900
 901         /**
 902          * Check for empty diff.
 903          *
 904          * @return bool True iff two sequences were identical.
 905          */
 906         function isEmpty () {
 907                 foreach ($this->edits as $edit) {
 908                         if ($edit->type != 'copy')
 909                                 return false;
 910                 }
 911                 return true;
 912         }
 913
 914         /**
 915          * Compute the length of the Longest Common Subsequence (LCS).
 916          *
 917          * This is mostly for diagnostic purposed.
 918          *
 919          * @return int The length of the LCS.
 920          */
 921         function lcs () {
 922                 $lcs = 0;
 923                 foreach ($this->edits as $edit) {
 924                         if ($edit->type == 'copy')
 925                                 $lcs += sizeof($edit->orig);
 926                 }
 927                 return $lcs;
 928         }
 929
 930         /**
 931          * Get the original set of lines.
 932          *
 933          * This reconstructs the $from_lines parameter passed to the
 934          * constructor.
 935          *
 936          * @return array The original sequence of strings.
 937          */
 938         function orig() {
 939                 $lines = array();
 940
 941                 foreach ($this->edits as $edit) {
 942                         if ($edit->orig)
 943                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 944                 }
 945                 return $lines;
 946         }
 947
 948         /**
 949          * Get the closing set of lines.
 950          *
 951          * This reconstructs the $to_lines parameter passed to the
 952          * constructor.
 953          *
 954          * @return array The sequence of strings.
 955          */
 956         function closing() {
 957                 $lines = array();
 958
 959                 foreach ($this->edits as $edit) {
 960                         if ($edit->closing)
 961                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 962                 }
 963                 return $lines;
 964         }
 965
 966         /**
 967          * Check a Diff for validity.
 968          *
 969          * This is here only for debugging purposes.
 970          */
 971         function _check ($from_lines, $to_lines) {
 972                 $fname = 'Diff::_check';
 973                 wfProfileIn( $fname );
 974                 if (serialize($from_lines) != serialize($this->orig()))
 975                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 976                 if (serialize($to_lines) != serialize($this->closing()))
 977                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 978
 979                 $rev = $this->reverse();
 980                 if (serialize($to_lines) != serialize($rev->orig()))
 981                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 982                 if (serialize($from_lines) != serialize($rev->closing()))
 983                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 984
 985
 986                 $prevtype = 'none';
 987                 foreach ($this->edits as $edit) {
 988                         if ( $prevtype == $edit->type )
 989                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 990                         $prevtype = $edit->type;
 991                 }
 992
 993                 $lcs = $this->lcs();
 994                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 995                 wfProfileOut( $fname );
 996         }
 997 }
 998
 999 /**
1000  * FIXME: bad name.
1001  * @todo document
1002  * @access private
1003  * @package MediaWiki
1004  * @subpackage DifferenceEngine
1005  */
1006 class MappedDiff extends Diff
1007 {
1008         /**
1009          * Constructor.
1010          *
1011          * Computes diff between sequences of strings.
1012          *
1013          * This can be used to compute things like
1014          * case-insensitve diffs, or diffs which ignore
1015          * changes in white-space.
1016          *
1017          * @param $from_lines array An array of strings.
1018          *      (Typically these are lines from a file.)
1019          *
1020          * @param $to_lines array An array of strings.
1021          *
1022          * @param $mapped_from_lines array This array should
1023          *      have the same size number of elements as $from_lines.
1024          *      The elements in $mapped_from_lines and
1025          *      $mapped_to_lines are what is actually compared
1026          *      when computing the diff.
1027          *
1028          * @param $mapped_to_lines array This array should
1029          *      have the same number of elements as $to_lines.
1030          */
1031         function MappedDiff($from_lines, $to_lines,
1032                                                 $mapped_from_lines, $mapped_to_lines) {
1033                 $fname = 'MappedDiff::MappedDiff';
1034                 wfProfileIn( $fname );
1035
1036                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1037                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1038
1039                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1040
1041                 $xi = $yi = 0;
1042                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1043                         $orig = &$this->edits[$i]->orig;
1044                         if (is_array($orig)) {
1045                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1046                                 $xi += sizeof($orig);
1047                         }
1048
1049                         $closing = &$this->edits[$i]->closing;
1050                         if (is_array($closing)) {
1051                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1052                                 $yi += sizeof($closing);
1053                         }
1054                 }
1055                 wfProfileOut( $fname );
1056         }
1057 }
1058
1059 /**
1060  * A class to format Diffs
1061  *
1062  * This class formats the diff in classic diff format.
1063  * It is intended that this class be customized via inheritance,
1064  * to obtain fancier outputs.
1065  * @todo document
1066  * @access private
1067  * @package MediaWiki
1068  * @subpackage DifferenceEngine
1069  */
1070 class DiffFormatter
1071 {
1072         /**
1073          * Number of leading context "lines" to preserve.
1074          *
1075          * This should be left at zero for this class, but subclasses
1076          * may want to set this to other values.
1077          */
1078         var $leading_context_lines = 0;
1079
1080         /**
1081          * Number of trailing context "lines" to preserve.
1082          *
1083          * This should be left at zero for this class, but subclasses
1084          * may want to set this to other values.
1085          */
1086         var $trailing_context_lines = 0;
1087
1088         /**
1089          * Format a diff.
1090          *
1091          * @param $diff object A Diff object.
1092          * @return string The formatted output.
1093          */
1094         function format($diff) {
1095                 $fname = 'DiffFormatter::format';
1096                 wfProfileIn( $fname );
1097
1098                 $xi = $yi = 1;
1099                 $block = false;
1100                 $context = array();
1101
1102                 $nlead = $this->leading_context_lines;
1103                 $ntrail = $this->trailing_context_lines;
1104
1105                 $this->_start_diff();
1106
1107                 foreach ($diff->edits as $edit) {
1108                         if ($edit->type == 'copy') {
1109                                 if (is_array($block)) {
1110                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1111                                                 $block[] = $edit;
1112                                         }
1113                                         else{
1114                                                 if ($ntrail) {
1115                                                         $context = array_slice($edit->orig, 0, $ntrail);
1116                                                         $block[] = new _DiffOp_Copy($context);
1117                                                 }
1118                                                 $this->_block($x0, $ntrail + $xi - $x0,
1119                                                                           $y0, $ntrail + $yi - $y0,
1120                                                                           $block);
1121                                                 $block = false;
1122                                         }
1123                                 }
1124                                 $context = $edit->orig;
1125                         }
1126                         else {
1127                                 if (! is_array($block)) {
1128                                         $context = array_slice($context, sizeof($context) - $nlead);
1129                                         $x0 = $xi - sizeof($context);
1130                                         $y0 = $yi - sizeof($context);
1131                                         $block = array();
1132                                         if ($context)
1133                                                 $block[] = new _DiffOp_Copy($context);
1134                                 }
1135                                 $block[] = $edit;
1136                         }
1137
1138                         if ($edit->orig)
1139                                 $xi += sizeof($edit->orig);
1140                         if ($edit->closing)
1141                                 $yi += sizeof($edit->closing);
1142                 }
1143
1144                 if (is_array($block))
1145                         $this->_block($x0, $xi - $x0,
1146                                                   $y0, $yi - $y0,
1147                                                   $block);
1148
1149                 $end = $this->_end_diff();
1150                 wfProfileOut( $fname );
1151                 return $end;
1152         }
1153
1154         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1155                 $fname = 'DiffFormatter::_block';
1156                 wfProfileIn( $fname );
1157                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1158                 foreach ($edits as $edit) {
1159                         if ($edit->type == 'copy')
1160                                 $this->_context($edit->orig);
1161                         elseif ($edit->type == 'add')
1162                                 $this->_added($edit->closing);
1163                         elseif ($edit->type == 'delete')
1164                                 $this->_deleted($edit->orig);
1165                         elseif ($edit->type == 'change')
1166                                 $this->_changed($edit->orig, $edit->closing);
1167                         else
1168                                 trigger_error('Unknown edit type', E_USER_ERROR);
1169                 }
1170                 $this->_end_block();
1171                 wfProfileOut( $fname );
1172         }
1173
1174         function _start_diff() {
1175                 ob_start();
1176         }
1177
1178         function _end_diff() {
1179                 $val = ob_get_contents();
1180                 ob_end_clean();
1181                 return $val;
1182         }
1183
1184         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1185                 if ($xlen > 1)
1186                         $xbeg .= "," . ($xbeg + $xlen - 1);
1187                 if ($ylen > 1)
1188                         $ybeg .= "," . ($ybeg + $ylen - 1);
1189
1190                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1191         }
1192
1193         function _start_block($header) {
1194                 echo $header;
1195         }
1196
1197         function _end_block() {
1198         }
1199
1200         function _lines($lines, $prefix = ' ') {
1201                 foreach ($lines as $line)
1202                         echo "$prefix $line\n";
1203         }
1204
1205         function _context($lines) {
1206                 $this->_lines($lines);
1207         }
1208
1209         function _added($lines) {
1210                 $this->_lines($lines, '>');
1211         }
1212         function _deleted($lines) {
1213                 $this->_lines($lines, '<');
1214         }
1215
1216         function _changed($orig, $closing) {
1217                 $this->_deleted($orig);
1218                 echo "---\n";
1219                 $this->_added($closing);
1220         }
1221 }
1222
1223
1224 /**
1225  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1226  *
1227  */
1228
1229 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1230
1231 /**
1232  * @todo document
1233  * @access private
1234  * @package MediaWiki
1235  * @subpackage DifferenceEngine
1236  */
1237 class _HWLDF_WordAccumulator {
1238         function _HWLDF_WordAccumulator () {
1239                 $this->_lines = array();
1240                 $this->_line = '';
1241                 $this->_group = '';
1242                 $this->_tag = '';
1243         }
1244
1245         function _flushGroup ($new_tag) {
1246                 if ($this->_group !== '') {
1247                         if ($this->_tag == 'mark')
1248                                 $this->_line .= '<span class="diffchange">' .
1249                                         htmlspecialchars ( $this->_group ) . '</span>';
1250                         else
1251                                 $this->_line .= htmlspecialchars ( $this->_group );
1252                 }
1253                 $this->_group = '';
1254                 $this->_tag = $new_tag;
1255         }
1256
1257         function _flushLine ($new_tag) {
1258                 $this->_flushGroup($new_tag);
1259                 if ($this->_line != '')
1260                         array_push ( $this->_lines, $this->_line );
1261                 else
1262                         # make empty lines visible by inserting an NBSP
1263                         array_push ( $this->_lines, NBSP );
1264                 $this->_line = '';
1265         }
1266
1267         function addWords ($words, $tag = '') {
1268                 if ($tag != $this->_tag)
1269                         $this->_flushGroup($tag);
1270
1271                 foreach ($words as $word) {
1272                         // new-line should only come as first char of word.
1273                         if ($word == '')
1274                                 continue;
1275                         if ($word[0] == "\n") {
1276                                 $this->_flushLine($tag);
1277                                 $word = substr($word, 1);
1278                         }
1279                         assert(!strstr($word, "\n"));
1280                         $this->_group .= $word;
1281                 }
1282         }
1283
1284         function getLines() {
1285                 $this->_flushLine('~done');
1286                 return $this->_lines;
1287         }
1288 }
1289
1290 /**
1291  * @todo document
1292  * @access private
1293  * @package MediaWiki
1294  * @subpackage DifferenceEngine
1295  */
1296 class WordLevelDiff extends MappedDiff
1297 {
1298         function WordLevelDiff ($orig_lines, $closing_lines) {
1299                 $fname = 'WordLevelDiff::WordLevelDiff';
1300                 wfProfileIn( $fname );
1301
1302                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1303                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1304
1305
1306                 $this->MappedDiff($orig_words, $closing_words,
1307                                                   $orig_stripped, $closing_stripped);
1308                 wfProfileOut( $fname );
1309         }
1310
1311         function _split($lines) {
1312                 $fname = 'WordLevelDiff::_split';
1313                 wfProfileIn( $fname );
1314                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1315                                                         implode("\n", $lines),
1316                                                         $m)) {
1317                         wfProfileOut( $fname );
1318                         return array(array(''), array(''));
1319                 }
1320                 wfProfileOut( $fname );
1321                 return array($m[0], $m[1]);
1322         }
1323
1324         function orig () {
1325                 $fname = 'WordLevelDiff::orig';
1326                 wfProfileIn( $fname );
1327                 $orig = new _HWLDF_WordAccumulator;
1328
1329                 foreach ($this->edits as $edit) {
1330                         if ($edit->type == 'copy')
1331                                 $orig->addWords($edit->orig);
1332                         elseif ($edit->orig)
1333                                 $orig->addWords($edit->orig, 'mark');
1334                 }
1335                 $lines = $orig->getLines();
1336                 wfProfileOut( $fname );
1337                 return $lines;
1338         }
1339
1340         function closing () {
1341                 $fname = 'WordLevelDiff::closing';
1342                 wfProfileIn( $fname );
1343                 $closing = new _HWLDF_WordAccumulator;
1344
1345                 foreach ($this->edits as $edit) {
1346                         if ($edit->type == 'copy')
1347                                 $closing->addWords($edit->closing);
1348                         elseif ($edit->closing)
1349                                 $closing->addWords($edit->closing, 'mark');
1350                 }
1351                 $lines = $closing->getLines();
1352                 wfProfileOut( $fname );
1353                 return $lines;
1354         }
1355 }
1356
1357 /**
1358  *      Wikipedia Table style diff formatter.
1359  * @todo document
1360  * @access private
1361  * @package MediaWiki
1362  * @subpackage DifferenceEngine
1363  */
1364 class TableDiffFormatter extends DiffFormatter
1365 {
1366         function TableDiffFormatter() {
1367                 $this->leading_context_lines = 2;
1368                 $this->trailing_context_lines = 2;
1369         }
1370
1371         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1372                 $l1 = wfMsg( 'lineno', $xbeg );
1373                 $l2 = wfMsg( 'lineno', $ybeg );
1374
1375                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1376                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1377                 return $r;
1378         }
1379
1380         function _start_block( $header ) {
1381                 global $wgOut;
1382                 echo $header;
1383         }
1384
1385         function _end_block() {
1386         }
1387
1388         function _lines( $lines, $prefix=' ', $color='white' ) {
1389         }
1390
1391         # HTML-escape parameter before calling this
1392         function addedLine( $line ) {
1393                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1394         }
1395
1396         # HTML-escape parameter before calling this
1397         function deletedLine( $line ) {
1398                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1399         }
1400
1401         # HTML-escape parameter before calling this
1402         function contextLine( $line ) {
1403                 return "<td> </td><td class='diff-context'>{$line}</td>";
1404         }
1405
1406         function emptyLine() {
1407                 return '<td colspan="2">&nbsp;</td>';
1408         }
1409
1410         function _added( $lines ) {
1411                 foreach ($lines as $line) {
1412                         echo '<tr>' . $this->emptyLine() .
1413                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1414                 }
1415         }
1416
1417         function _deleted($lines) {
1418                 foreach ($lines as $line) {
1419                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1420                           $this->emptyLine() . "</tr>\n";
1421                 }
1422         }
1423
1424         function _context( $lines ) {
1425                 foreach ($lines as $line) {
1426                         echo '<tr>' .
1427                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1428                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1429                 }
1430         }
1431
1432         function _changed( $orig, $closing ) {
1433                 $fname = 'TableDiffFormatter::_changed';
1434                 wfProfileIn( $fname );
1435
1436                 $diff = new WordLevelDiff( $orig, $closing );
1437                 $del = $diff->orig();
1438                 $add = $diff->closing();
1439
1440                 # Notice that WordLevelDiff returns HTML-escaped output.
1441                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1442
1443                 while ( $line = array_shift( $del ) ) {
1444                         $aline = array_shift( $add );
1445                         echo '<tr>' . $this->deletedLine( $line ) .
1446                                 $this->addedLine( $aline ) . "</tr>\n";
1447                 }
1448                 foreach ($add as $line) {       # If any leftovers
1449                         echo '<tr>' . $this->emptyLine() .
1450                                 $this->addedLine( $line ) . "</tr>\n";
1451                 }
1452                 wfProfileOut( $fname );
1453         }
1454 }
1455
1456 ?>