includes/DifferenceEngine.php

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