includes/DifferenceEngine.php

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