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