includes/DifferenceEngine.php

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