includes/DifferenceEngine.php

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