includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @package MediaWiki
   5  * @subpackage DifferenceEngine
   6  */
   7
   8 /** */
   9 require_once( 'Revision.php' );
  10
  11 /**
  12  * @todo document
  13  * @access public
  14  * @package MediaWiki
  15  * @subpackage DifferenceEngine
  16  */
  17 class DifferenceEngine {
  18         /**#@+
  19          * @access private
  20          */
  21         var $mOldid, $mNewid;
  22         var $mOldtitle, $mNewtitle, $mPagetitle;
  23         var $mOldtext, $mNewtext;
  24         var $mOldUser, $mNewUser;
  25         var $mOldComment, $mNewComment;
  26         var $mOldPage, $mNewPage;
  27         var $mRcidMarkPatrolled;
  28         /**#@-*/
  29
  30         /**
  31          * Constructor
  32          * @param integer $old Old ID we want to show and diff with.
  33          * @param string $new Either 'prev' or 'next'.
  34          * @param integer $rcid ??? (default 0)
  35          */
  36         function DifferenceEngine( $old, $new, $rcid = 0 ) {
  37                 global $wgTitle;
  38
  39                 if ( 'prev' == $new ) {
  40                         # Show diff between revision $old and the previous one.
  41                         # Get previous one from DB.
  42                         #
  43                         $this->mNewid = intval($old);
  44
  45                         $this->mOldid = $wgTitle->getPreviousRevisionID( $this->mNewid );
  46
  47                 } elseif ( 'next' == $new ) {
  48                         # Show diff between revision $old and the previous one.
  49                         # Get previous one from DB.
  50                         #
  51                         $this->mOldid = intval($old);
  52                         $this->mNewid = $wgTitle->getNextRevisionID( $this->mOldid );
  53                         if ( false === $this->mNewid ) {
  54                                 # if no result, NewId points to the newest old revision. The only newer
  55                                 # revision is cur, which is "0".
  56                                 $this->mNewid = 0;
  57                         }
  58
  59                 } else {
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = intval($new);
  62                 }
  63                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  64         }
  65
  66         function showDiffPage() {
  67                 global $wgUser, $wgTitle, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol,
  68                        $wgUseExternalEditor, $wgUseRCPatrol;
  69                 $fname = 'DifferenceEngine::showDiffPage';
  70                 wfProfileIn( $fname );
  71
  72                 # If external diffs are enabled both globally and for the user,
  73                 # we'll use the application/x-external-editor interface to call
  74                 # an external diff tool like kompare, kdiff3, etc.
  75                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  76                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  77                         $wgOut->disable();
  78                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  79                         $url1=$wgTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  80                         $url2=$wgTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  81                         $special=$wgLang->getNsText(NS_SPECIAL);
  82                         $control=<<<CONTROL
  83 [Process]
  84 Type=Diff text
  85 Engine=MediaWiki
  86 Script={$wgServer}{$wgScript}
  87 Special namespace={$special}
  88
  89 [File]
  90 Extension=wiki
  91 URL=$url1
  92
  93 [File 2]
  94 Extension=wiki
  95 URL=$url2
  96 CONTROL;
  97                         echo($control);
  98                         return;
  99                 }
 100
 101                 # mOldid is false if the difference engine is called with a "vague" query for
 102                 # a diff between a version V and its previous version V' AND the version V
 103                 # is the first version of that article. In that case, V' does not exist.
 104                 if ( $this->mOldid === false ) {
 105                         $this->showFirstRevision();
 106                         wfProfileOut( $fname );
 107                         return;
 108                 }
 109
 110                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 111                   "{$this->mNewid})";
 112                 $mtext = wfMsg( 'missingarticle', $t );
 113
 114                 $wgOut->setArticleFlag( false );
 115                 if ( ! $this->loadText() ) {
 116                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 117                         $wgOut->addHTML( $mtext );
 118                         wfProfileOut( $fname );
 119                         return;
 120                 }
 121                 $wgOut->suppressQuickbar();
 122
 123                 $oldTitle = $this->mOldPage->getPrefixedText();
 124                 $newTitle = $this->mNewPage->getPrefixedText();
 125                 if( $oldTitle == $newTitle ) {
 126                         $wgOut->setPageTitle( $newTitle );
 127                 } else {
 128                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 129                 }
 130                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 131                 $wgOut->setRobotpolicy( 'noindex,follow' );
 132
 133                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 134                         $wgOut->loginToUse();
 135                         $wgOut->output();
 136                         wfProfileOut( $fname );
 137                         exit;
 138                 }
 139
 140                 $sk = $wgUser->getSkin();
 141                 $talk = $wgContLang->getNsText( NS_TALK );
 142                 $contribs = wfMsg( 'contribslink' );
 143
 144                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 145                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 146
 147                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 148                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 149                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 150                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 151                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 152                         'target=' . urlencode($this->mOldUser) );
 153                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 154                         'target=' . urlencode($this->mNewUser) );
 155                 if ( $this->newRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 156                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
 157                                 'action=rollback&from=' . urlencode($this->mNewUser) .
 158                                 '&token=' . urlencode( $wgUser->editToken( array( $wgTitle->getPrefixedText(), $this->mNewUser ) ) ) ) .
 159                                 ']</strong>';
 160                 } else {
 161                         $rollback = '';
 162                 }
 163                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isLoggedIn() &&
 164                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 165                 {
 166                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 167                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 168                 } else {
 169                         $patrol = '';
 170                 }
 171
 172                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'previousdiff' ), 'diff=prev&oldid='.$this->mOldid );
 173                 if ( $this->newRev->isCurrent() ) {
 174                         $nextlink = '';
 175                 } else {
 176                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 177                 }
 178
 179                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 180                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 181                         '<br />' . $prevlink;
 182                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 183                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 184                         '<br />' . $nextlink . $patrol;
 185
 186                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 187                   $oldHeader, $newHeader );
 188                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 189
 190                 if( !$this->newRev->isCurrent() ) {
 191                         $oldEditSectionSetting = $wgOut->mParserOptions->setEditSection( false );
 192                 }
 193                 $wgOut->addWikiText( $this->mNewtext );
 194                 if( !$this->newRev->isCurrent() ) {
 195                         $wgOut->mParserOptions->setEditSection( $oldEditSectionSetting );
 196                 }
 197
 198                 wfProfileOut( $fname );
 199         }
 200
 201         /**
 202          * Show the first revision of an article. Uses normal diff headers in
 203          * contrast to normal "old revision" display style.
 204          */
 205         function showFirstRevision() {
 206                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 207
 208                 $fname = 'DifferenceEngine::showFirstRevision';
 209                 wfProfileIn( $fname );
 210
 211
 212                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 213
 214                 # Get article text from the DB
 215                 #
 216                 if ( ! $this->loadText() ) {
 217                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 218                           "{$this->mNewid})";
 219                         $mtext = wfMsg( 'missingarticle', $t );
 220                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 221                         $wgOut->addHTML( $mtext );
 222                         wfProfileOut( $fname );
 223                         return;
 224                 }
 225
 226                 # Check if user is allowed to look at this page. If not, bail out.
 227                 #
 228                 if ( !( $this->mOldPage->userCanRead() ) ) {
 229                         $wgOut->loginToUse();
 230                         $wgOut->output();
 231                         wfProfileOut( $fname );
 232                         exit;
 233                 }
 234
 235                 # Prepare the header box
 236                 #
 237                 $sk = $wgUser->getSkin();
 238
 239                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 240                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 241                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 242                         'target=' . urlencode($this->mOldUser) );
 243                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 244                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 245                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 246                         '<br />' . $nextlink. "</div>\n";
 247
 248                 $wgOut->addHTML( $header );
 249
 250                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 251                 $wgOut->setRobotpolicy( 'noindex,follow' );
 252
 253
 254                 # Show current revision
 255                 #
 256                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 257                 $wgOut->addWikiText( $this->mNewtext );
 258
 259                 wfProfileOut( $fname );
 260         }
 261
 262         function showDiff( $otext, $ntext, $otitle, $ntitle ) {
 263                 global $wgOut;
 264                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 265         }
 266
 267         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 268                 global $wgUseExternalDiffEngine, $wgContLang;
 269                 $out = "
 270                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 271                         <tr>
 272                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 273                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 274                         </tr>
 275                 ";
 276                 $otext = $wgContLang->segmentForDiff($otext);
 277                 $ntext = $wgContLang->segmentForDiff($ntext);
 278                 $difftext='';
 279                 if ( $wgUseExternalDiffEngine ) {
 280                         # For historical reasons, external diff engine expects
 281                         # input text to be HTML-escaped already
 282                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 283                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 284                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 285                                 dl('php_wikidiff.so');
 286                         }
 287                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 288                 } else {
 289                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 290                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 291                         $diffs =& new Diff( $ota, $nta );
 292                         $formatter =& new TableDiffFormatter();
 293                         $difftext = $formatter->format( $diffs );
 294                 }
 295                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 296                 $out .= $difftext."</table>\n";
 297                 return $out;
 298         }
 299
 300         /**
 301          * Load the text of the articles to compare.  If newid is 0, then compare
 302          * the old article in oldid to the current article; if oldid is 0, then
 303          * compare the current article to the immediately previous one (ignoring the
 304          * value of newid).
 305          */
 306         function loadText() {
 307                 global $wgTitle, $wgOut, $wgLang;
 308                 $fname = 'DifferenceEngine::loadText';
 309
 310                 $dbr =& wfGetDB( DB_SLAVE );
 311                 if( $this->mNewid ) {
 312                         $this->newRev =& Revision::newFromId( $this->mNewid );
 313                 } else {
 314                         $this->newRev =& Revision::newFromTitle( $wgTitle );
 315                 }
 316                 if( is_null( $this->newRev ) ) {
 317                         return false;
 318                 }
 319
 320                 if( $this->newRev->isCurrent() ) {
 321                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 322                         $this->mNewPage = $wgTitle;
 323                         $newLink = $this->mNewPage->escapeLocalUrl();
 324                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 325                 } else {
 326                         $this->mNewPage = $this->newRev->getTitle();
 327                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 328                         $t = $wgLang->timeanddate( $this->newRev->getTimestamp(), true );
 329                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 330                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 331                 }
 332
 333                 if( $this->mOldid ) {
 334                         $this->oldRev =& Revision::newFromId( $this->mOldid );
 335                 } else {
 336                         $this->oldRev =& $this->newRev->getPrevious();
 337                         $this->mOldid = $this->oldRev->getId();
 338                 }
 339                 if( is_null( $this->oldRev ) ) {
 340                         return false;
 341                 }
 342
 343                 $this->mOldPage = $this->oldRev->getTitle();
 344
 345                 $t = $wgLang->timeanddate( $this->oldRev->getTimestamp(), true );
 346                 $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 347                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 348
 349                 $this->mNewUser = $this->newRev->getUserText();
 350                 $this->mNewComment = $this->newRev->getComment();
 351                 $this->mNewtext = $this->newRev->getText();
 352
 353                 $this->mOldUser = $this->oldRev->getUserText();
 354                 $this->mOldComment = $this->oldRev->getComment();
 355                 $this->mOldtext = $this->oldRev->getText();
 356
 357                 return true;
 358         }
 359 }
 360
 361 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 362 //
 363 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 364 // You may copy this code freely under the conditions of the GPL.
 365 //
 366
 367 define('USE_ASSERTS', function_exists('assert'));
 368
 369 /**
 370  * @todo document
 371  * @access private
 372  * @package MediaWiki
 373  * @subpackage DifferenceEngine
 374  */
 375 class _DiffOp {
 376         var $type;
 377         var $orig;
 378         var $closing;
 379
 380         function reverse() {
 381                 trigger_error('pure virtual', E_USER_ERROR);
 382         }
 383
 384         function norig() {
 385                 return $this->orig ? sizeof($this->orig) : 0;
 386         }
 387
 388         function nclosing() {
 389                 return $this->closing ? sizeof($this->closing) : 0;
 390         }
 391 }
 392
 393 /**
 394  * @todo document
 395  * @access private
 396  * @package MediaWiki
 397  * @subpackage DifferenceEngine
 398  */
 399 class _DiffOp_Copy extends _DiffOp {
 400         var $type = 'copy';
 401
 402         function _DiffOp_Copy ($orig, $closing = false) {
 403                 if (!is_array($closing))
 404                         $closing = $orig;
 405                 $this->orig = $orig;
 406                 $this->closing = $closing;
 407         }
 408
 409         function reverse() {
 410                 return new _DiffOp_Copy($this->closing, $this->orig);
 411         }
 412 }
 413
 414 /**
 415  * @todo document
 416  * @access private
 417  * @package MediaWiki
 418  * @subpackage DifferenceEngine
 419  */
 420 class _DiffOp_Delete extends _DiffOp {
 421         var $type = 'delete';
 422
 423         function _DiffOp_Delete ($lines) {
 424                 $this->orig = $lines;
 425                 $this->closing = false;
 426         }
 427
 428         function reverse() {
 429                 return new _DiffOp_Add($this->orig);
 430         }
 431 }
 432
 433 /**
 434  * @todo document
 435  * @access private
 436  * @package MediaWiki
 437  * @subpackage DifferenceEngine
 438  */
 439 class _DiffOp_Add extends _DiffOp {
 440         var $type = 'add';
 441
 442         function _DiffOp_Add ($lines) {
 443                 $this->closing = $lines;
 444                 $this->orig = false;
 445         }
 446
 447         function reverse() {
 448                 return new _DiffOp_Delete($this->closing);
 449         }
 450 }
 451
 452 /**
 453  * @todo document
 454  * @access private
 455  * @package MediaWiki
 456  * @subpackage DifferenceEngine
 457  */
 458 class _DiffOp_Change extends _DiffOp {
 459         var $type = 'change';
 460
 461         function _DiffOp_Change ($orig, $closing) {
 462                 $this->orig = $orig;
 463                 $this->closing = $closing;
 464         }
 465
 466         function reverse() {
 467                 return new _DiffOp_Change($this->closing, $this->orig);
 468         }
 469 }
 470
 471
 472 /**
 473  * Class used internally by Diff to actually compute the diffs.
 474  *
 475  * The algorithm used here is mostly lifted from the perl module
 476  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 477  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 478  *
 479  * More ideas are taken from:
 480  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 481  *
 482  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 483  * diffutils-2.7, which can be found at:
 484  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 485  *
 486  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 487  * are my own.
 488  *
 489  * @author Geoffrey T. Dairiki
 490  * @access private
 491  * @package MediaWiki
 492  * @subpackage DifferenceEngine
 493  */
 494 class _DiffEngine
 495 {
 496         function diff ($from_lines, $to_lines) {
 497                 $fname = '_DiffEngine::diff';
 498                 wfProfileIn( $fname );
 499
 500                 $n_from = sizeof($from_lines);
 501                 $n_to = sizeof($to_lines);
 502
 503                 $this->xchanged = $this->ychanged = array();
 504                 $this->xv = $this->yv = array();
 505                 $this->xind = $this->yind = array();
 506                 unset($this->seq);
 507                 unset($this->in_seq);
 508                 unset($this->lcs);
 509
 510                 // Skip leading common lines.
 511                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 512                         if ($from_lines[$skip] != $to_lines[$skip])
 513                                 break;
 514                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 515                 }
 516                 // Skip trailing common lines.
 517                 $xi = $n_from; $yi = $n_to;
 518                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 519                         if ($from_lines[$xi] != $to_lines[$yi])
 520                                 break;
 521                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 522                 }
 523
 524                 // Ignore lines which do not exist in both files.
 525                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 526                         $xhash[$from_lines[$xi]] = 1;
 527                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 528                         $line = $to_lines[$yi];
 529                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 530                                 continue;
 531                         $yhash[$line] = 1;
 532                         $this->yv[] = $line;
 533                         $this->yind[] = $yi;
 534                 }
 535                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 536                         $line = $from_lines[$xi];
 537                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 538                                 continue;
 539                         $this->xv[] = $line;
 540                         $this->xind[] = $xi;
 541                 }
 542
 543                 // Find the LCS.
 544                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 545
 546                 // Merge edits when possible
 547                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 548                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 549
 550                 // Compute the edit operations.
 551                 $edits = array();
 552                 $xi = $yi = 0;
 553                 while ($xi < $n_from || $yi < $n_to) {
 554                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 555                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 556
 557                         // Skip matching "snake".
 558                         $copy = array();
 559                         while ( $xi < $n_from && $yi < $n_to
 560                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 561                                 $copy[] = $from_lines[$xi++];
 562                                 ++$yi;
 563                         }
 564                         if ($copy)
 565                                 $edits[] = new _DiffOp_Copy($copy);
 566
 567                         // Find deletes & adds.
 568                         $delete = array();
 569                         while ($xi < $n_from && $this->xchanged[$xi])
 570                                 $delete[] = $from_lines[$xi++];
 571
 572                         $add = array();
 573                         while ($yi < $n_to && $this->ychanged[$yi])
 574                                 $add[] = $to_lines[$yi++];
 575
 576                         if ($delete && $add)
 577                                 $edits[] = new _DiffOp_Change($delete, $add);
 578                         elseif ($delete)
 579                                 $edits[] = new _DiffOp_Delete($delete);
 580                         elseif ($add)
 581                                 $edits[] = new _DiffOp_Add($add);
 582                 }
 583                 wfProfileOut( $fname );
 584                 return $edits;
 585         }
 586
 587
 588         /* Divide the Largest Common Subsequence (LCS) of the sequences
 589          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 590          * sized segments.
 591          *
 592          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 593          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 594          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 595          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 596          * that (X0, Y0) == (XOFF, YOFF) and
 597          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 598          *
 599          * This function assumes that the first lines of the specified portions
 600          * of the two files do not match, and likewise that the last lines do not
 601          * match.  The caller must trim matching lines from the beginning and end
 602          * of the portions it is going to specify.
 603          */
 604         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 605                 $fname = '_DiffEngine::_diag';
 606                 wfProfileIn( $fname );
 607                 $flip = false;
 608
 609                 if ($xlim - $xoff > $ylim - $yoff) {
 610                         // Things seems faster (I'm not sure I understand why)
 611                                 // when the shortest sequence in X.
 612                                 $flip = true;
 613                         list ($xoff, $xlim, $yoff, $ylim)
 614                         = array( $yoff, $ylim, $xoff, $xlim);
 615                 }
 616
 617                 if ($flip)
 618                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 619                                 $ymatches[$this->xv[$i]][] = $i;
 620                 else
 621                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 622                                 $ymatches[$this->yv[$i]][] = $i;
 623
 624                 $this->lcs = 0;
 625                 $this->seq[0]= $yoff - 1;
 626                 $this->in_seq = array();
 627                 $ymids[0] = array();
 628
 629                 $numer = $xlim - $xoff + $nchunks - 1;
 630                 $x = $xoff;
 631                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 632                         wfProfileIn( "$fname-chunk" );
 633                         if ($chunk > 0)
 634                                 for ($i = 0; $i <= $this->lcs; $i++)
 635                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 636
 637                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 638                         for ( ; $x < $x1; $x++) {
 639                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 640                                         if (empty($ymatches[$line]))
 641                                                 continue;
 642                                 $matches = $ymatches[$line];
 643                                 reset($matches);
 644                                 while (list ($junk, $y) = each($matches))
 645                                         if (empty($this->in_seq[$y])) {
 646                                                 $k = $this->_lcs_pos($y);
 647                                                 USE_ASSERTS && assert($k > 0);
 648                                                 $ymids[$k] = $ymids[$k-1];
 649                                                 break;
 650                                         }
 651                                 while (list ($junk, $y) = each($matches)) {
 652                                         if ($y > $this->seq[$k-1]) {
 653                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 654                                                 // Optimization: this is a common case:
 655                                                 //      next match is just replacing previous match.
 656                                                 $this->in_seq[$this->seq[$k]] = false;
 657                                                 $this->seq[$k] = $y;
 658                                                 $this->in_seq[$y] = 1;
 659                                         } else if (empty($this->in_seq[$y])) {
 660                                                 $k = $this->_lcs_pos($y);
 661                                                 USE_ASSERTS && assert($k > 0);
 662                                                 $ymids[$k] = $ymids[$k-1];
 663                                         }
 664                                 }
 665                         }
 666                         wfProfileOut( "$fname-chunk" );
 667                 }
 668
 669                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 670                 $ymid = $ymids[$this->lcs];
 671                 for ($n = 0; $n < $nchunks - 1; $n++) {
 672                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 673                         $y1 = $ymid[$n] + 1;
 674                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 675                 }
 676                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 677
 678                 wfProfileOut( $fname );
 679                 return array($this->lcs, $seps);
 680         }
 681
 682         function _lcs_pos ($ypos) {
 683                 $fname = '_DiffEngine::_lcs_pos';
 684                 wfProfileIn( $fname );
 685
 686                 $end = $this->lcs;
 687                 if ($end == 0 || $ypos > $this->seq[$end]) {
 688                         $this->seq[++$this->lcs] = $ypos;
 689                         $this->in_seq[$ypos] = 1;
 690                         wfProfileOut( $fname );
 691                         return $this->lcs;
 692                 }
 693
 694                 $beg = 1;
 695                 while ($beg < $end) {
 696                         $mid = (int)(($beg + $end) / 2);
 697                         if ( $ypos > $this->seq[$mid] )
 698                                 $beg = $mid + 1;
 699                         else
 700                                 $end = $mid;
 701                 }
 702
 703                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 704
 705                 $this->in_seq[$this->seq[$end]] = false;
 706                 $this->seq[$end] = $ypos;
 707                 $this->in_seq[$ypos] = 1;
 708                 wfProfileOut( $fname );
 709                 return $end;
 710         }
 711
 712         /* Find LCS of two sequences.
 713          *
 714          * The results are recorded in the vectors $this->{x,y}changed[], by
 715          * storing a 1 in the element for each line that is an insertion
 716          * or deletion (ie. is not in the LCS).
 717          *
 718          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 719          *
 720          * Note that XLIM, YLIM are exclusive bounds.
 721          * All line numbers are origin-0 and discarded lines are not counted.
 722          */
 723         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 724                 $fname = '_DiffEngine::_compareseq';
 725                 wfProfileIn( $fname );
 726
 727                 // Slide down the bottom initial diagonal.
 728                 while ($xoff < $xlim && $yoff < $ylim
 729                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 730                         ++$xoff;
 731                         ++$yoff;
 732                 }
 733
 734                 // Slide up the top initial diagonal.
 735                 while ($xlim > $xoff && $ylim > $yoff
 736                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 737                         --$xlim;
 738                         --$ylim;
 739                 }
 740
 741                 if ($xoff == $xlim || $yoff == $ylim)
 742                         $lcs = 0;
 743                 else {
 744                         // This is ad hoc but seems to work well.
 745                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 746                         //$nchunks = max(2,min(8,(int)$nchunks));
 747                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 748                         list ($lcs, $seps)
 749                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 750                 }
 751
 752                 if ($lcs == 0) {
 753                         // X and Y sequences have no common subsequence:
 754                         // mark all changed.
 755                         while ($yoff < $ylim)
 756                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 757                         while ($xoff < $xlim)
 758                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 759                 } else {
 760                         // Use the partitions to split this problem into subproblems.
 761                         reset($seps);
 762                         $pt1 = $seps[0];
 763                         while ($pt2 = next($seps)) {
 764                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 765                                 $pt1 = $pt2;
 766                         }
 767                 }
 768                 wfProfileOut( $fname );
 769         }
 770
 771         /* Adjust inserts/deletes of identical lines to join changes
 772          * as much as possible.
 773          *
 774          * We do something when a run of changed lines include a
 775          * line at one end and has an excluded, identical line at the other.
 776          * We are free to choose which identical line is included.
 777          * `compareseq' usually chooses the one at the beginning,
 778          * but usually it is cleaner to consider the following identical line
 779          * to be the "change".
 780          *
 781          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 782          */
 783         function _shift_boundaries ($lines, &$changed, $other_changed) {
 784                 $fname = '_DiffEngine::_shift_boundaries';
 785                 wfProfileIn( $fname );
 786                 $i = 0;
 787                 $j = 0;
 788
 789                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 790                 $len = sizeof($lines);
 791                 $other_len = sizeof($other_changed);
 792
 793                 while (1) {
 794                         /*
 795                          * Scan forwards to find beginning of another run of changes.
 796                          * Also keep track of the corresponding point in the other file.
 797                          *
 798                          * Throughout this code, $i and $j are adjusted together so that
 799                          * the first $i elements of $changed and the first $j elements
 800                          * of $other_changed both contain the same number of zeros
 801                          * (unchanged lines).
 802                          * Furthermore, $j is always kept so that $j == $other_len or
 803                          * $other_changed[$j] == false.
 804                          */
 805                         while ($j < $other_len && $other_changed[$j])
 806                                 $j++;
 807
 808                         while ($i < $len && ! $changed[$i]) {
 809                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 810                                 $i++; $j++;
 811                                 while ($j < $other_len && $other_changed[$j])
 812                                         $j++;
 813                         }
 814
 815                         if ($i == $len)
 816                                 break;
 817
 818                         $start = $i;
 819
 820                         // Find the end of this run of changes.
 821                         while (++$i < $len && $changed[$i])
 822                                 continue;
 823
 824                         do {
 825                                 /*
 826                                  * Record the length of this run of changes, so that
 827                                  * we can later determine whether the run has grown.
 828                                  */
 829                                 $runlength = $i - $start;
 830
 831                                 /*
 832                                  * Move the changed region back, so long as the
 833                                  * previous unchanged line matches the last changed one.
 834                                  * This merges with previous changed regions.
 835                                  */
 836                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 837                                         $changed[--$start] = 1;
 838                                         $changed[--$i] = false;
 839                                         while ($start > 0 && $changed[$start - 1])
 840                                                 $start--;
 841                                         USE_ASSERTS && assert('$j > 0');
 842                                         while ($other_changed[--$j])
 843                                                 continue;
 844                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 845                                 }
 846
 847                                 /*
 848                                  * Set CORRESPONDING to the end of the changed run, at the last
 849                                  * point where it corresponds to a changed run in the other file.
 850                                  * CORRESPONDING == LEN means no such point has been found.
 851                                  */
 852                                 $corresponding = $j < $other_len ? $i : $len;
 853
 854                                 /*
 855                                  * Move the changed region forward, so long as the
 856                                  * first changed line matches the following unchanged one.
 857                                  * This merges with following changed regions.
 858                                  * Do this second, so that if there are no merges,
 859                                  * the changed region is moved forward as far as possible.
 860                                  */
 861                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 862                                         $changed[$start++] = false;
 863                                         $changed[$i++] = 1;
 864                                         while ($i < $len && $changed[$i])
 865                                                 $i++;
 866
 867                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 868                                         $j++;
 869                                         if ($j < $other_len && $other_changed[$j]) {
 870                                                 $corresponding = $i;
 871                                                 while ($j < $other_len && $other_changed[$j])
 872                                                         $j++;
 873                                         }
 874                                 }
 875                         } while ($runlength != $i - $start);
 876
 877                         /*
 878                          * If possible, move the fully-merged run of changes
 879                          * back to a corresponding run in the other file.
 880                          */
 881                         while ($corresponding < $i) {
 882                                 $changed[--$start] = 1;
 883                                 $changed[--$i] = 0;
 884                                 USE_ASSERTS && assert('$j > 0');
 885                                 while ($other_changed[--$j])
 886                                         continue;
 887                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 888                         }
 889                 }
 890                 wfProfileOut( $fname );
 891         }
 892 }
 893
 894 /**
 895  * Class representing a 'diff' between two sequences of strings.
 896  * @todo document
 897  * @access private
 898  * @package MediaWiki
 899  * @subpackage DifferenceEngine
 900  */
 901 class Diff
 902 {
 903         var $edits;
 904
 905         /**
 906          * Constructor.
 907          * Computes diff between sequences of strings.
 908          *
 909          * @param $from_lines array An array of strings.
 910          *                (Typically these are lines from a file.)
 911          * @param $to_lines array An array of strings.
 912          */
 913         function Diff($from_lines, $to_lines) {
 914                 $eng = new _DiffEngine;
 915                 $this->edits = $eng->diff($from_lines, $to_lines);
 916                 //$this->_check($from_lines, $to_lines);
 917         }
 918
 919         /**
 920          * Compute reversed Diff.
 921          *
 922          * SYNOPSIS:
 923          *
 924          *      $diff = new Diff($lines1, $lines2);
 925          *      $rev = $diff->reverse();
 926          * @return object A Diff object representing the inverse of the
 927          *                                original diff.
 928          */
 929         function reverse () {
 930                 $rev = $this;
 931                 $rev->edits = array();
 932                 foreach ($this->edits as $edit) {
 933                         $rev->edits[] = $edit->reverse();
 934                 }
 935                 return $rev;
 936         }
 937
 938         /**
 939          * Check for empty diff.
 940          *
 941          * @return bool True iff two sequences were identical.
 942          */
 943         function isEmpty () {
 944                 foreach ($this->edits as $edit) {
 945                         if ($edit->type != 'copy')
 946                                 return false;
 947                 }
 948                 return true;
 949         }
 950
 951         /**
 952          * Compute the length of the Longest Common Subsequence (LCS).
 953          *
 954          * This is mostly for diagnostic purposed.
 955          *
 956          * @return int The length of the LCS.
 957          */
 958         function lcs () {
 959                 $lcs = 0;
 960                 foreach ($this->edits as $edit) {
 961                         if ($edit->type == 'copy')
 962                                 $lcs += sizeof($edit->orig);
 963                 }
 964                 return $lcs;
 965         }
 966
 967         /**
 968          * Get the original set of lines.
 969          *
 970          * This reconstructs the $from_lines parameter passed to the
 971          * constructor.
 972          *
 973          * @return array The original sequence of strings.
 974          */
 975         function orig() {
 976                 $lines = array();
 977
 978                 foreach ($this->edits as $edit) {
 979                         if ($edit->orig)
 980                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 981                 }
 982                 return $lines;
 983         }
 984
 985         /**
 986          * Get the closing set of lines.
 987          *
 988          * This reconstructs the $to_lines parameter passed to the
 989          * constructor.
 990          *
 991          * @return array The sequence of strings.
 992          */
 993         function closing() {
 994                 $lines = array();
 995
 996                 foreach ($this->edits as $edit) {
 997                         if ($edit->closing)
 998                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 999                 }
1000                 return $lines;
1001         }
1002
1003         /**
1004          * Check a Diff for validity.
1005          *
1006          * This is here only for debugging purposes.
1007          */
1008         function _check ($from_lines, $to_lines) {
1009                 $fname = 'Diff::_check';
1010                 wfProfileIn( $fname );
1011                 if (serialize($from_lines) != serialize($this->orig()))
1012                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1013                 if (serialize($to_lines) != serialize($this->closing()))
1014                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1015
1016                 $rev = $this->reverse();
1017                 if (serialize($to_lines) != serialize($rev->orig()))
1018                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1019                 if (serialize($from_lines) != serialize($rev->closing()))
1020                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1021
1022
1023                 $prevtype = 'none';
1024                 foreach ($this->edits as $edit) {
1025                         if ( $prevtype == $edit->type )
1026                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1027                         $prevtype = $edit->type;
1028                 }
1029
1030                 $lcs = $this->lcs();
1031                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1032                 wfProfileOut( $fname );
1033         }
1034 }
1035
1036 /**
1037  * FIXME: bad name.
1038  * @todo document
1039  * @access private
1040  * @package MediaWiki
1041  * @subpackage DifferenceEngine
1042  */
1043 class MappedDiff extends Diff
1044 {
1045         /**
1046          * Constructor.
1047          *
1048          * Computes diff between sequences of strings.
1049          *
1050          * This can be used to compute things like
1051          * case-insensitve diffs, or diffs which ignore
1052          * changes in white-space.
1053          *
1054          * @param $from_lines array An array of strings.
1055          *      (Typically these are lines from a file.)
1056          *
1057          * @param $to_lines array An array of strings.
1058          *
1059          * @param $mapped_from_lines array This array should
1060          *      have the same size number of elements as $from_lines.
1061          *      The elements in $mapped_from_lines and
1062          *      $mapped_to_lines are what is actually compared
1063          *      when computing the diff.
1064          *
1065          * @param $mapped_to_lines array This array should
1066          *      have the same number of elements as $to_lines.
1067          */
1068         function MappedDiff($from_lines, $to_lines,
1069                                                 $mapped_from_lines, $mapped_to_lines) {
1070                 $fname = 'MappedDiff::MappedDiff';
1071                 wfProfileIn( $fname );
1072
1073                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1074                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1075
1076                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1077
1078                 $xi = $yi = 0;
1079                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1080                         $orig = &$this->edits[$i]->orig;
1081                         if (is_array($orig)) {
1082                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1083                                 $xi += sizeof($orig);
1084                         }
1085
1086                         $closing = &$this->edits[$i]->closing;
1087                         if (is_array($closing)) {
1088                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1089                                 $yi += sizeof($closing);
1090                         }
1091                 }
1092                 wfProfileOut( $fname );
1093         }
1094 }
1095
1096 /**
1097  * A class to format Diffs
1098  *
1099  * This class formats the diff in classic diff format.
1100  * It is intended that this class be customized via inheritance,
1101  * to obtain fancier outputs.
1102  * @todo document
1103  * @access private
1104  * @package MediaWiki
1105  * @subpackage DifferenceEngine
1106  */
1107 class DiffFormatter
1108 {
1109         /**
1110          * Number of leading context "lines" to preserve.
1111          *
1112          * This should be left at zero for this class, but subclasses
1113          * may want to set this to other values.
1114          */
1115         var $leading_context_lines = 0;
1116
1117         /**
1118          * Number of trailing context "lines" to preserve.
1119          *
1120          * This should be left at zero for this class, but subclasses
1121          * may want to set this to other values.
1122          */
1123         var $trailing_context_lines = 0;
1124
1125         /**
1126          * Format a diff.
1127          *
1128          * @param $diff object A Diff object.
1129          * @return string The formatted output.
1130          */
1131         function format($diff) {
1132                 $fname = 'DiffFormatter::format';
1133                 wfProfileIn( $fname );
1134
1135                 $xi = $yi = 1;
1136                 $block = false;
1137                 $context = array();
1138
1139                 $nlead = $this->leading_context_lines;
1140                 $ntrail = $this->trailing_context_lines;
1141
1142                 $this->_start_diff();
1143
1144                 foreach ($diff->edits as $edit) {
1145                         if ($edit->type == 'copy') {
1146                                 if (is_array($block)) {
1147                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1148                                                 $block[] = $edit;
1149                                         }
1150                                         else{
1151                                                 if ($ntrail) {
1152                                                         $context = array_slice($edit->orig, 0, $ntrail);
1153                                                         $block[] = new _DiffOp_Copy($context);
1154                                                 }
1155                                                 $this->_block($x0, $ntrail + $xi - $x0,
1156                                                                           $y0, $ntrail + $yi - $y0,
1157                                                                           $block);
1158                                                 $block = false;
1159                                         }
1160                                 }
1161                                 $context = $edit->orig;
1162                         }
1163                         else {
1164                                 if (! is_array($block)) {
1165                                         $context = array_slice($context, sizeof($context) - $nlead);
1166                                         $x0 = $xi - sizeof($context);
1167                                         $y0 = $yi - sizeof($context);
1168                                         $block = array();
1169                                         if ($context)
1170                                                 $block[] = new _DiffOp_Copy($context);
1171                                 }
1172                                 $block[] = $edit;
1173                         }
1174
1175                         if ($edit->orig)
1176                                 $xi += sizeof($edit->orig);
1177                         if ($edit->closing)
1178                                 $yi += sizeof($edit->closing);
1179                 }
1180
1181                 if (is_array($block))
1182                         $this->_block($x0, $xi - $x0,
1183                                                   $y0, $yi - $y0,
1184                                                   $block);
1185
1186                 $end = $this->_end_diff();
1187                 wfProfileOut( $fname );
1188                 return $end;
1189         }
1190
1191         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1192                 $fname = 'DiffFormatter::_block';
1193                 wfProfileIn( $fname );
1194                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1195                 foreach ($edits as $edit) {
1196                         if ($edit->type == 'copy')
1197                                 $this->_context($edit->orig);
1198                         elseif ($edit->type == 'add')
1199                                 $this->_added($edit->closing);
1200                         elseif ($edit->type == 'delete')
1201                                 $this->_deleted($edit->orig);
1202                         elseif ($edit->type == 'change')
1203                                 $this->_changed($edit->orig, $edit->closing);
1204                         else
1205                                 trigger_error('Unknown edit type', E_USER_ERROR);
1206                 }
1207                 $this->_end_block();
1208                 wfProfileOut( $fname );
1209         }
1210
1211         function _start_diff() {
1212                 ob_start();
1213         }
1214
1215         function _end_diff() {
1216                 $val = ob_get_contents();
1217                 ob_end_clean();
1218                 return $val;
1219         }
1220
1221         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1222                 if ($xlen > 1)
1223                         $xbeg .= "," . ($xbeg + $xlen - 1);
1224                 if ($ylen > 1)
1225                         $ybeg .= "," . ($ybeg + $ylen - 1);
1226
1227                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1228         }
1229
1230         function _start_block($header) {
1231                 echo $header;
1232         }
1233
1234         function _end_block() {
1235         }
1236
1237         function _lines($lines, $prefix = ' ') {
1238                 foreach ($lines as $line)
1239                         echo "$prefix $line\n";
1240         }
1241
1242         function _context($lines) {
1243                 $this->_lines($lines);
1244         }
1245
1246         function _added($lines) {
1247                 $this->_lines($lines, '>');
1248         }
1249         function _deleted($lines) {
1250                 $this->_lines($lines, '<');
1251         }
1252
1253         function _changed($orig, $closing) {
1254                 $this->_deleted($orig);
1255                 echo "---\n";
1256                 $this->_added($closing);
1257         }
1258 }
1259
1260
1261 /**
1262  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1263  *
1264  */
1265
1266 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1267
1268 /**
1269  * @todo document
1270  * @access private
1271  * @package MediaWiki
1272  * @subpackage DifferenceEngine
1273  */
1274 class _HWLDF_WordAccumulator {
1275         function _HWLDF_WordAccumulator () {
1276                 $this->_lines = array();
1277                 $this->_line = '';
1278                 $this->_group = '';
1279                 $this->_tag = '';
1280         }
1281
1282         function _flushGroup ($new_tag) {
1283                 if ($this->_group !== '') {
1284                         if ($this->_tag == 'mark')
1285                                 $this->_line .= '<span class="diffchange">' .
1286                                         htmlspecialchars ( $this->_group ) . '</span>';
1287                         else
1288                                 $this->_line .= htmlspecialchars ( $this->_group );
1289                 }
1290                 $this->_group = '';
1291                 $this->_tag = $new_tag;
1292         }
1293
1294         function _flushLine ($new_tag) {
1295                 $this->_flushGroup($new_tag);
1296                 if ($this->_line != '')
1297                         array_push ( $this->_lines, $this->_line );
1298                 else
1299                         # make empty lines visible by inserting an NBSP
1300                         array_push ( $this->_lines, NBSP );
1301                 $this->_line = '';
1302         }
1303
1304         function addWords ($words, $tag = '') {
1305                 if ($tag != $this->_tag)
1306                         $this->_flushGroup($tag);
1307
1308                 foreach ($words as $word) {
1309                         // new-line should only come as first char of word.
1310                         if ($word == '')
1311                                 continue;
1312                         if ($word[0] == "\n") {
1313                                 $this->_flushLine($tag);
1314                                 $word = substr($word, 1);
1315                         }
1316                         assert(!strstr($word, "\n"));
1317                         $this->_group .= $word;
1318                 }
1319         }
1320
1321         function getLines() {
1322                 $this->_flushLine('~done');
1323                 return $this->_lines;
1324         }
1325 }
1326
1327 /**
1328  * @todo document
1329  * @access private
1330  * @package MediaWiki
1331  * @subpackage DifferenceEngine
1332  */
1333 class WordLevelDiff extends MappedDiff
1334 {
1335         function WordLevelDiff ($orig_lines, $closing_lines) {
1336                 $fname = 'WordLevelDiff::WordLevelDiff';
1337                 wfProfileIn( $fname );
1338
1339                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1340                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1341
1342
1343                 $this->MappedDiff($orig_words, $closing_words,
1344                                                   $orig_stripped, $closing_stripped);
1345                 wfProfileOut( $fname );
1346         }
1347
1348         function _split($lines) {
1349                 $fname = 'WordLevelDiff::_split';
1350                 wfProfileIn( $fname );
1351                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1352                                                         implode("\n", $lines),
1353                                                         $m)) {
1354                         wfProfileOut( $fname );
1355                         return array(array(''), array(''));
1356                 }
1357                 wfProfileOut( $fname );
1358                 return array($m[0], $m[1]);
1359         }
1360
1361         function orig () {
1362                 $fname = 'WordLevelDiff::orig';
1363                 wfProfileIn( $fname );
1364                 $orig = new _HWLDF_WordAccumulator;
1365
1366                 foreach ($this->edits as $edit) {
1367                         if ($edit->type == 'copy')
1368                                 $orig->addWords($edit->orig);
1369                         elseif ($edit->orig)
1370                                 $orig->addWords($edit->orig, 'mark');
1371                 }
1372                 $lines = $orig->getLines();
1373                 wfProfileOut( $fname );
1374                 return $lines;
1375         }
1376
1377         function closing () {
1378                 $fname = 'WordLevelDiff::closing';
1379                 wfProfileIn( $fname );
1380                 $closing = new _HWLDF_WordAccumulator;
1381
1382                 foreach ($this->edits as $edit) {
1383                         if ($edit->type == 'copy')
1384                                 $closing->addWords($edit->closing);
1385                         elseif ($edit->closing)
1386                                 $closing->addWords($edit->closing, 'mark');
1387                 }
1388                 $lines = $closing->getLines();
1389                 wfProfileOut( $fname );
1390                 return $lines;
1391         }
1392 }
1393
1394 /**
1395  *      Wikipedia Table style diff formatter.
1396  * @todo document
1397  * @access private
1398  * @package MediaWiki
1399  * @subpackage DifferenceEngine
1400  */
1401 class TableDiffFormatter extends DiffFormatter
1402 {
1403         function TableDiffFormatter() {
1404                 $this->leading_context_lines = 2;
1405                 $this->trailing_context_lines = 2;
1406         }
1407
1408         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1409                 $l1 = wfMsg( 'lineno', $xbeg );
1410                 $l2 = wfMsg( 'lineno', $ybeg );
1411
1412                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1413                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1414                 return $r;
1415         }
1416
1417         function _start_block( $header ) {
1418                 global $wgOut;
1419                 echo $header;
1420         }
1421
1422         function _end_block() {
1423         }
1424
1425         function _lines( $lines, $prefix=' ', $color='white' ) {
1426         }
1427
1428         # HTML-escape parameter before calling this
1429         function addedLine( $line ) {
1430                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1431         }
1432
1433         # HTML-escape parameter before calling this
1434         function deletedLine( $line ) {
1435                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1436         }
1437
1438         # HTML-escape parameter before calling this
1439         function contextLine( $line ) {
1440                 return "<td> </td><td class='diff-context'>{$line}</td>";
1441         }
1442
1443         function emptyLine() {
1444                 return '<td colspan="2">&nbsp;</td>';
1445         }
1446
1447         function _added( $lines ) {
1448                 foreach ($lines as $line) {
1449                         echo '<tr>' . $this->emptyLine() .
1450                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1451                 }
1452         }
1453
1454         function _deleted($lines) {
1455                 foreach ($lines as $line) {
1456                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1457                           $this->emptyLine() . "</tr>\n";
1458                 }
1459         }
1460
1461         function _context( $lines ) {
1462                 foreach ($lines as $line) {
1463                         echo '<tr>' .
1464                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1465                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1466                 }
1467         }
1468
1469         function _changed( $orig, $closing ) {
1470                 $fname = 'TableDiffFormatter::_changed';
1471                 wfProfileIn( $fname );
1472
1473                 $diff = new WordLevelDiff( $orig, $closing );
1474                 $del = $diff->orig();
1475                 $add = $diff->closing();
1476
1477                 # Notice that WordLevelDiff returns HTML-escaped output.
1478                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1479
1480                 while ( $line = array_shift( $del ) ) {
1481                         $aline = array_shift( $add );
1482                         echo '<tr>' . $this->deletedLine( $line ) .
1483                                 $this->addedLine( $aline ) . "</tr>\n";
1484                 }
1485                 foreach ($add as $line) {       # If any leftovers
1486                         echo '<tr>' . $this->emptyLine() .
1487                                 $this->addedLine( $line ) . "</tr>\n";
1488                 }
1489                 wfProfileOut( $fname );
1490         }
1491 }
1492
1493 ?>