includes/DifferenceEngine.php

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