includes/DifferenceEngine.php

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