includes/DifferenceEngine.php

   1 <?php
   2 # See diff.doc
   3
   4 class DifferenceEngine {
   5         /* private */ var $mOldid, $mNewid;
   6         /* private */ var $mOldtitle, $mNewtitle;
   7         /* private */ var $mOldtext, $mNewtext;
   8         /* private */ var $mOldUser, $mNewUser;
   9         /* private */ var $mOldComment, $mNewComment;
  10         /* private */ var $mOldPage, $mNewPage;
  11
  12         function DifferenceEngine( $old, $new )
  13         {
  14                 $this->mOldid = $old;
  15                 $this->mNewid = $new;
  16         }
  17
  18         function showDiffPage()
  19         {
  20                 global $wgUser, $wgTitle, $wgOut, $wgLang;
  21                 $fname = "DifferenceEngine::showDiffPage";
  22                 wfProfileIn( $fname );
  23
  24                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  25                   "{$this->mNewid})";
  26                 $mtext = wfMsg( "missingarticle", $t );
  27
  28                 $wgOut->setArticleFlag( false );
  29                 if ( ! $this->loadText() ) {
  30                         $wgOut->setPagetitle( wfMsg( "errorpagetitle" ) );
  31                         $wgOut->addHTML( $mtext );
  32                         wfProfileOut( $fname );
  33                         return;
  34                 }
  35                 $wgOut->suppressQuickbar();
  36
  37                 $oldTitle = $this->mOldPage->getPrefixedText();
  38                 $newTitle = $this->mNewPage->getPrefixedText();
  39                 if( $oldTitle == $newTitle ) {
  40                         $wgOut->setPageTitle( $newTitle );
  41                 } else {
  42                         $wgOut->setPageTitle( $oldTitle . ", " . $newTitle );
  43                 }
  44                 $wgOut->setSubtitle( wfMsg( "difference" ) );
  45                 $wgOut->setRobotpolicy( "noindex,follow" );
  46
  47                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  48                         $wgOut->loginToUse();
  49                         $wgOut->output();
  50                         wfProfileOut( $fname );
  51                         exit;
  52                 }
  53
  54                 $sk = $wgUser->getSkin();
  55                 $talk = $wgLang->getNsText( NS_TALK );
  56                 $contribs = wfMsg( "contribslink" );
  57
  58
  59                 $this->mOldComment = $sk->formatComment($this->mOldComment);
  60                 $this->mNewComment = $sk->formatComment($this->mNewComment);
  61
  62
  63                 $oldUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mOldUser ), $this->mOldUser );
  64                 $newUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mNewUser ), $this->mNewUser );
  65                 $oldUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mOldUser ), $talk );
  66                 $newUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mNewUser ), $talk );
  67                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  68                         "target=" . urlencode($this->mOldUser) );
  69                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  70                         "target=" . urlencode($this->mNewUser) );
  71                 if ( !$this->mNewid && $wgUser->isSysop() ) {
  72                         $rollback = "&nbsp;&nbsp;&nbsp;<strong>[" . $sk->makeKnownLinkObj( $wgTitle, wfMsg( "rollbacklink" ),
  73                                 "action=rollback&from=" . urlencode($this->mNewUser) ) . "]</strong>";
  74                 } else {
  75                         $rollback = "";
  76                 }
  77
  78                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink ($oldUTLink | $oldContribs)<br />".
  79                   $this->mOldComment;
  80                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink ($newUTLink | $newContribs) $rollback<br />".
  81                   $this->mNewComment;
  82
  83                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
  84                   $oldHeader, $newHeader );
  85                 $wgOut->addHTML( "<hr /><h2>{$this->mNewtitle}</h2>\n" );
  86                 $wgOut->addWikiText( $this->mNewtext );
  87
  88                 wfProfileOut( $fname );
  89         }
  90
  91         function showDiff( $otext, $ntext, $otitle, $ntitle )
  92         {
  93                 global $wgOut;
  94
  95                 $ota = explode( "\n", str_replace( "\r\n", "\n",
  96                   htmlspecialchars( $otext ) ) );
  97                 $nta = explode( "\n", str_replace( "\r\n", "\n",
  98                   htmlspecialchars( $ntext ) ) );
  99
 100                 $wgOut->addHTML( "<table border='0' width='98%'
 101 cellpadding='0' cellspacing='4px' class='diff'><tr>
 102 <td colspan='2' width='50%' align='center' class='diff-otitle'>
 103 {$otitle}</td>
 104 <td colspan='2' width='50%' align='center' class='diff-ntitle'>
 105 {$ntitle}</td>
 106 </tr>\n" );
 107
 108                 $diffs = new Diff( $ota, $nta );
 109                 $formatter = new TableDiffFormatter();
 110                 $formatter->format( $diffs );
 111                 $wgOut->addHTML( "</table>\n" );
 112         }
 113
 114         # Load the text of the articles to compare.  If newid is 0, then compare
 115         # the old article in oldid to the current article; if oldid is 0, then
 116         # compare the current article to the immediately previous one (ignoring
 117         # the value of newid).
 118         #
 119         function loadText()
 120         {
 121                 global $wgTitle, $wgOut, $wgLang, $wgIsMySQL;
 122                 $fname = "DifferenceEngine::loadText";
 123
 124                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 125                         $wgOut->setArticleFlag( true );
 126                         $this->mNewtitle = wfMsg( "currentrev" );
 127                         $id = $wgTitle->getArticleID();
 128
 129                         $sql = "SELECT cur_text, cur_user_text, cur_comment FROM cur WHERE cur_id={$id}";
 130                         $res = wfQuery( $sql, DB_READ, $fname );
 131                         if ( 0 == wfNumRows( $res ) ) { return false; }
 132
 133                         $s = wfFetchObject( $res );
 134                         $this->mNewPage = &$wgTitle;
 135                         $this->mNewtext = $s->cur_text;
 136                         $this->mNewUser = $s->cur_user_text;
 137                         $this->mNewComment = $s->cur_comment;
 138                 } else {
 139                         $sql = "SELECT old_namespace,old_title,old_timestamp,old_text,old_flags,old_user_text,old_comment FROM old WHERE " .
 140                           "old_id={$this->mNewid}";
 141
 142                         $res = wfQuery( $sql, DB_READ, $fname );
 143                         if ( 0 == wfNumRows( $res ) ) { return false; }
 144
 145                         $s = wfFetchObject( $res );
 146                         $this->mNewtext = Article::getRevisionText( $s );
 147
 148                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 149                         $this->mNewPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 150                         $this->mNewtitle = wfMsg( "revisionasof", $t );
 151                         $this->mNewUser = $s->old_user_text;
 152                         $this->mNewComment = $s->old_comment;
 153                 }
 154                 if ( 0 == $this->mOldid ) {
 155                         $use_index=$wgIsMySQL?"USE INDEX (name_title_timestamp)":"";
 156                         $sql = "SELECT old_namespace,old_title,old_timestamp,old_text,old_flags,old_user_text,old_comment " .
 157                           "FROM old $use_index WHERE " .
 158                           "old_namespace=" . $this->mNewPage->getNamespace() . " AND " .
 159                           "old_title='" . wfStrencode( $this->mNewPage->getDBkey() ) .
 160                           "' ORDER BY inverse_timestamp LIMIT 1";
 161                         $res = wfQuery( $sql, DB_READ, $fname );
 162                 } else {
 163                         $sql = "SELECT old_namespace,old_title,old_timestamp,old_text,old_flags,old_user_text,old_comment FROM old WHERE " .
 164                           "old_id={$this->mOldid}";
 165                         $res = wfQuery( $sql, DB_READ, $fname );
 166                 }
 167                 if ( 0 == wfNumRows( $res ) ) { return false; }
 168
 169                 $s = wfFetchObject( $res );
 170                 $this->mOldPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 171                 $this->mOldtext = Article::getRevisionText( $s );
 172
 173                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 174                 $this->mOldtitle = wfMsg( "revisionasof", $t );
 175                 $this->mOldUser = $s->old_user_text;
 176                 $this->mOldComment = $s->old_comment;
 177
 178                 return true;
 179         }
 180 }
 181
 182 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 183 //
 184 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 185 // You may copy this code freely under the conditions of the GPL.
 186 //
 187
 188 define('USE_ASSERTS', function_exists('assert'));
 189
 190 class _DiffOp {
 191         var $type;
 192         var $orig;
 193         var $closing;
 194
 195         function reverse() {
 196                 trigger_error("pure virtual", E_USER_ERROR);
 197         }
 198
 199         function norig() {
 200                 return $this->orig ? sizeof($this->orig) : 0;
 201         }
 202
 203         function nclosing() {
 204                 return $this->closing ? sizeof($this->closing) : 0;
 205         }
 206 }
 207
 208 class _DiffOp_Copy extends _DiffOp {
 209         var $type = 'copy';
 210
 211         function _DiffOp_Copy ($orig, $closing = false) {
 212                 if (!is_array($closing))
 213                         $closing = $orig;
 214                 $this->orig = $orig;
 215                 $this->closing = $closing;
 216         }
 217
 218         function reverse() {
 219                 return new _DiffOp_Copy($this->closing, $this->orig);
 220         }
 221 }
 222
 223 class _DiffOp_Delete extends _DiffOp {
 224         var $type = 'delete';
 225
 226         function _DiffOp_Delete ($lines) {
 227                 $this->orig = $lines;
 228                 $this->closing = false;
 229         }
 230
 231         function reverse() {
 232                 return new _DiffOp_Add($this->orig);
 233         }
 234 }
 235
 236 class _DiffOp_Add extends _DiffOp {
 237         var $type = 'add';
 238
 239         function _DiffOp_Add ($lines) {
 240                 $this->closing = $lines;
 241                 $this->orig = false;
 242         }
 243
 244         function reverse() {
 245                 return new _DiffOp_Delete($this->closing);
 246         }
 247 }
 248
 249 class _DiffOp_Change extends _DiffOp {
 250         var $type = 'change';
 251
 252         function _DiffOp_Change ($orig, $closing) {
 253                 $this->orig = $orig;
 254                 $this->closing = $closing;
 255         }
 256
 257         function reverse() {
 258                 return new _DiffOp_Change($this->closing, $this->orig);
 259         }
 260 }
 261
 262
 263 /**
 264  * Class used internally by Diff to actually compute the diffs.
 265  *
 266  * The algorithm used here is mostly lifted from the perl module
 267  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 268  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 269  *
 270  * More ideas are taken from:
 271  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 272  *
 273  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 274  * diffutils-2.7, which can be found at:
 275  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 276  *
 277  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 278  * are my own.
 279  *
 280  * @author Geoffrey T. Dairiki
 281  * @access private
 282  */
 283 class _DiffEngine
 284 {
 285         function diff ($from_lines, $to_lines) {
 286                 $n_from = sizeof($from_lines);
 287                 $n_to = sizeof($to_lines);
 288
 289                 $this->xchanged = $this->ychanged = array();
 290                 $this->xv = $this->yv = array();
 291                 $this->xind = $this->yind = array();
 292                 unset($this->seq);
 293                 unset($this->in_seq);
 294                 unset($this->lcs);
 295
 296                 // Skip leading common lines.
 297                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 298                         if ($from_lines[$skip] != $to_lines[$skip])
 299                                 break;
 300                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 301                 }
 302                 // Skip trailing common lines.
 303                 $xi = $n_from; $yi = $n_to;
 304                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 305                         if ($from_lines[$xi] != $to_lines[$yi])
 306                                 break;
 307                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 308                 }
 309
 310                 // Ignore lines which do not exist in both files.
 311                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 312                         $xhash[$from_lines[$xi]] = 1;
 313                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 314                         $line = $to_lines[$yi];
 315                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 316                                 continue;
 317                         $yhash[$line] = 1;
 318                         $this->yv[] = $line;
 319                         $this->yind[] = $yi;
 320                 }
 321                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 322                         $line = $from_lines[$xi];
 323                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 324                                 continue;
 325                         $this->xv[] = $line;
 326                         $this->xind[] = $xi;
 327                 }
 328
 329                 // Find the LCS.
 330                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 331
 332                 // Merge edits when possible
 333                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 334                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 335
 336                 // Compute the edit operations.
 337                 $edits = array();
 338                 $xi = $yi = 0;
 339                 while ($xi < $n_from || $yi < $n_to) {
 340                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 341                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 342
 343                         // Skip matching "snake".
 344                         $copy = array();
 345                         while ( $xi < $n_from && $yi < $n_to
 346                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 347                                 $copy[] = $from_lines[$xi++];
 348                                 ++$yi;
 349                         }
 350                         if ($copy)
 351                                 $edits[] = new _DiffOp_Copy($copy);
 352
 353                         // Find deletes & adds.
 354                         $delete = array();
 355                         while ($xi < $n_from && $this->xchanged[$xi])
 356                                 $delete[] = $from_lines[$xi++];
 357
 358                         $add = array();
 359                         while ($yi < $n_to && $this->ychanged[$yi])
 360                                 $add[] = $to_lines[$yi++];
 361
 362                         if ($delete && $add)
 363                                 $edits[] = new _DiffOp_Change($delete, $add);
 364                         elseif ($delete)
 365                                 $edits[] = new _DiffOp_Delete($delete);
 366                         elseif ($add)
 367                                 $edits[] = new _DiffOp_Add($add);
 368                 }
 369                 return $edits;
 370         }
 371
 372
 373         /* Divide the Largest Common Subsequence (LCS) of the sequences
 374          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 375          * sized segments.
 376          *
 377          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 378          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 379          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 380          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 381          * that (X0, Y0) == (XOFF, YOFF) and
 382          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 383          *
 384          * This function assumes that the first lines of the specified portions
 385          * of the two files do not match, and likewise that the last lines do not
 386          * match.  The caller must trim matching lines from the beginning and end
 387          * of the portions it is going to specify.
 388          */
 389         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 390         $flip = false;
 391
 392         if ($xlim - $xoff > $ylim - $yoff) {
 393                 // Things seems faster (I'm not sure I understand why)
 394                         // when the shortest sequence in X.
 395                         $flip = true;
 396                 list ($xoff, $xlim, $yoff, $ylim)
 397                 = array( $yoff, $ylim, $xoff, $xlim);
 398                 }
 399
 400         if ($flip)
 401                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 402                 $ymatches[$this->xv[$i]][] = $i;
 403         else
 404                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 405                 $ymatches[$this->yv[$i]][] = $i;
 406
 407         $this->lcs = 0;
 408         $this->seq[0]= $yoff - 1;
 409         $this->in_seq = array();
 410         $ymids[0] = array();
 411
 412         $numer = $xlim - $xoff + $nchunks - 1;
 413         $x = $xoff;
 414         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 415                 if ($chunk > 0)
 416                 for ($i = 0; $i <= $this->lcs; $i++)
 417                         $ymids[$i][$chunk-1] = $this->seq[$i];
 418
 419                 $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 420                 for ( ; $x < $x1; $x++) {
 421                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 422                                 if (empty($ymatches[$line]))
 423                         continue;
 424                 $matches = $ymatches[$line];
 425                                 reset($matches);
 426                 while (list ($junk, $y) = each($matches))
 427                         if (empty($this->in_seq[$y])) {
 428                         $k = $this->_lcs_pos($y);
 429                         USE_ASSERTS && assert($k > 0);
 430                         $ymids[$k] = $ymids[$k-1];
 431                         break;
 432                                         }
 433                 while (list ($junk, $y) = each($matches)) {
 434                         if ($y > $this->seq[$k-1]) {
 435                         USE_ASSERTS && assert($y < $this->seq[$k]);
 436                         // Optimization: this is a common case:
 437                         //      next match is just replacing previous match.
 438                         $this->in_seq[$this->seq[$k]] = false;
 439                         $this->seq[$k] = $y;
 440                         $this->in_seq[$y] = 1;
 441                                         }
 442                         else if (empty($this->in_seq[$y])) {
 443                         $k = $this->_lcs_pos($y);
 444                         USE_ASSERTS && assert($k > 0);
 445                         $ymids[$k] = $ymids[$k-1];
 446                                         }
 447                                 }
 448                         }
 449                 }
 450
 451         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 452         $ymid = $ymids[$this->lcs];
 453         for ($n = 0; $n < $nchunks - 1; $n++) {
 454                 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 455                 $y1 = $ymid[$n] + 1;
 456                 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 457                 }
 458         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 459
 460         return array($this->lcs, $seps);
 461         }
 462
 463         function _lcs_pos ($ypos) {
 464         $end = $this->lcs;
 465         if ($end == 0 || $ypos > $this->seq[$end]) {
 466                 $this->seq[++$this->lcs] = $ypos;
 467                 $this->in_seq[$ypos] = 1;
 468                 return $this->lcs;
 469                 }
 470
 471         $beg = 1;
 472         while ($beg < $end) {
 473                 $mid = (int)(($beg + $end) / 2);
 474                 if ( $ypos > $this->seq[$mid] )
 475                 $beg = $mid + 1;
 476                 else
 477                 $end = $mid;
 478                 }
 479
 480         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 481
 482         $this->in_seq[$this->seq[$end]] = false;
 483         $this->seq[$end] = $ypos;
 484         $this->in_seq[$ypos] = 1;
 485         return $end;
 486         }
 487
 488         /* Find LCS of two sequences.
 489          *
 490          * The results are recorded in the vectors $this->{x,y}changed[], by
 491          * storing a 1 in the element for each line that is an insertion
 492          * or deletion (ie. is not in the LCS).
 493          *
 494          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 495          *
 496          * Note that XLIM, YLIM are exclusive bounds.
 497          * All line numbers are origin-0 and discarded lines are not counted.
 498          */
 499         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 500         // Slide down the bottom initial diagonal.
 501         while ($xoff < $xlim && $yoff < $ylim
 502                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 503                 ++$xoff;
 504                 ++$yoff;
 505                 }
 506
 507         // Slide up the top initial diagonal.
 508         while ($xlim > $xoff && $ylim > $yoff
 509                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 510                 --$xlim;
 511                 --$ylim;
 512                 }
 513
 514         if ($xoff == $xlim || $yoff == $ylim)
 515                 $lcs = 0;
 516         else {
 517                 // This is ad hoc but seems to work well.
 518                 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 519                 //$nchunks = max(2,min(8,(int)$nchunks));
 520                 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 521                 list ($lcs, $seps)
 522                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 523                 }
 524
 525         if ($lcs == 0) {
 526                 // X and Y sequences have no common subsequence:
 527                 // mark all changed.
 528                 while ($yoff < $ylim)
 529                 $this->ychanged[$this->yind[$yoff++]] = 1;
 530                 while ($xoff < $xlim)
 531                 $this->xchanged[$this->xind[$xoff++]] = 1;
 532                 }
 533         else {
 534                 // Use the partitions to split this problem into subproblems.
 535                 reset($seps);
 536                 $pt1 = $seps[0];
 537                 while ($pt2 = next($seps)) {
 538                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 539                 $pt1 = $pt2;
 540                         }
 541                 }
 542         }
 543
 544         /* Adjust inserts/deletes of identical lines to join changes
 545          * as much as possible.
 546          *
 547          * We do something when a run of changed lines include a
 548          * line at one end and has an excluded, identical line at the other.
 549          * We are free to choose which identical line is included.
 550          * `compareseq' usually chooses the one at the beginning,
 551          * but usually it is cleaner to consider the following identical line
 552          * to be the "change".
 553          *
 554          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 555          */
 556         function _shift_boundaries ($lines, &$changed, $other_changed) {
 557         $i = 0;
 558         $j = 0;
 559
 560         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 561         $len = sizeof($lines);
 562         $other_len = sizeof($other_changed);
 563
 564         while (1) {
 565                 /*
 566                  * Scan forwards to find beginning of another run of changes.
 567                  * Also keep track of the corresponding point in the other file.
 568                  *
 569                  * Throughout this code, $i and $j are adjusted together so that
 570                  * the first $i elements of $changed and the first $j elements
 571                  * of $other_changed both contain the same number of zeros
 572                  * (unchanged lines).
 573                  * Furthermore, $j is always kept so that $j == $other_len or
 574                  * $other_changed[$j] == false.
 575                  */
 576                 while ($j < $other_len && $other_changed[$j])
 577                 $j++;
 578
 579                 while ($i < $len && ! $changed[$i]) {
 580                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 581                 $i++; $j++;
 582                 while ($j < $other_len && $other_changed[$j])
 583                         $j++;
 584                         }
 585
 586                 if ($i == $len)
 587                 break;
 588
 589                 $start = $i;
 590
 591                 // Find the end of this run of changes.
 592                 while (++$i < $len && $changed[$i])
 593                 continue;
 594
 595                 do {
 596                 /*
 597                  * Record the length of this run of changes, so that
 598                  * we can later determine whether the run has grown.
 599                  */
 600                 $runlength = $i - $start;
 601
 602                 /*
 603                  * Move the changed region back, so long as the
 604                  * previous unchanged line matches the last changed one.
 605                  * This merges with previous changed regions.
 606                  */
 607                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 608                         $changed[--$start] = 1;
 609                         $changed[--$i] = false;
 610                         while ($start > 0 && $changed[$start - 1])
 611                         $start--;
 612                         USE_ASSERTS && assert('$j > 0');
 613                         while ($other_changed[--$j])
 614                         continue;
 615                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 616                                 }
 617
 618                 /*
 619                  * Set CORRESPONDING to the end of the changed run, at the last
 620                  * point where it corresponds to a changed run in the other file.
 621                  * CORRESPONDING == LEN means no such point has been found.
 622                  */
 623                 $corresponding = $j < $other_len ? $i : $len;
 624
 625                 /*
 626                  * Move the changed region forward, so long as the
 627                  * first changed line matches the following unchanged one.
 628                  * This merges with following changed regions.
 629                  * Do this second, so that if there are no merges,
 630                  * the changed region is moved forward as far as possible.
 631                  */
 632                 while ($i < $len && $lines[$start] == $lines[$i]) {
 633                         $changed[$start++] = false;
 634                         $changed[$i++] = 1;
 635                         while ($i < $len && $changed[$i])
 636                         $i++;
 637
 638                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 639                         $j++;
 640                         if ($j < $other_len && $other_changed[$j]) {
 641                         $corresponding = $i;
 642                         while ($j < $other_len && $other_changed[$j])
 643                                 $j++;
 644                                         }
 645                                 }
 646                         } while ($runlength != $i - $start);
 647
 648                 /*
 649                  * If possible, move the fully-merged run of changes
 650                  * back to a corresponding run in the other file.
 651                  */
 652                 while ($corresponding < $i) {
 653                 $changed[--$start] = 1;
 654                 $changed[--$i] = 0;
 655                 USE_ASSERTS && assert('$j > 0');
 656                 while ($other_changed[--$j])
 657                         continue;
 658                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 659                         }
 660                 }
 661         }
 662 }
 663
 664 /**
 665  * Class representing a 'diff' between two sequences of strings.
 666  */
 667 class Diff
 668 {
 669         var $edits;
 670
 671         /**
 672          * Constructor.
 673          * Computes diff between sequences of strings.
 674          *
 675          * @param $from_lines array An array of strings.
 676          *                (Typically these are lines from a file.)
 677          * @param $to_lines array An array of strings.
 678          */
 679         function Diff($from_lines, $to_lines) {
 680                 $eng = new _DiffEngine;
 681                 $this->edits = $eng->diff($from_lines, $to_lines);
 682                 //$this->_check($from_lines, $to_lines);
 683         }
 684
 685         /**
 686          * Compute reversed Diff.
 687          *
 688          * SYNOPSIS:
 689          *
 690          *      $diff = new Diff($lines1, $lines2);
 691          *      $rev = $diff->reverse();
 692          * @return object A Diff object representing the inverse of the
 693          *                                original diff.
 694          */
 695         function reverse () {
 696         $rev = $this;
 697                 $rev->edits = array();
 698                 foreach ($this->edits as $edit) {
 699                         $rev->edits[] = $edit->reverse();
 700                 }
 701         return $rev;
 702         }
 703
 704         /**
 705          * Check for empty diff.
 706          *
 707          * @return bool True iff two sequences were identical.
 708          */
 709         function isEmpty () {
 710                 foreach ($this->edits as $edit) {
 711                         if ($edit->type != 'copy')
 712                                 return false;
 713                 }
 714                 return true;
 715         }
 716
 717         /**
 718          * Compute the length of the Longest Common Subsequence (LCS).
 719          *
 720          * This is mostly for diagnostic purposed.
 721          *
 722          * @return int The length of the LCS.
 723          */
 724         function lcs () {
 725         $lcs = 0;
 726                 foreach ($this->edits as $edit) {
 727                         if ($edit->type == 'copy')
 728                                 $lcs += sizeof($edit->orig);
 729                 }
 730         return $lcs;
 731         }
 732
 733         /**
 734          * Get the original set of lines.
 735          *
 736          * This reconstructs the $from_lines parameter passed to the
 737          * constructor.
 738          *
 739          * @return array The original sequence of strings.
 740          */
 741         function orig() {
 742                 $lines = array();
 743
 744                 foreach ($this->edits as $edit) {
 745                         if ($edit->orig)
 746                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 747                 }
 748                 return $lines;
 749         }
 750
 751         /**
 752          * Get the closing set of lines.
 753          *
 754          * This reconstructs the $to_lines parameter passed to the
 755          * constructor.
 756          *
 757          * @return array The sequence of strings.
 758          */
 759         function closing() {
 760                 $lines = array();
 761
 762                 foreach ($this->edits as $edit) {
 763                         if ($edit->closing)
 764                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 765                 }
 766                 return $lines;
 767         }
 768
 769         /**
 770          * Check a Diff for validity.
 771          *
 772          * This is here only for debugging purposes.
 773          */
 774         function _check ($from_lines, $to_lines) {
 775                 if (serialize($from_lines) != serialize($this->orig()))
 776                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 777                 if (serialize($to_lines) != serialize($this->closing()))
 778                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 779
 780                 $rev = $this->reverse();
 781                 if (serialize($to_lines) != serialize($rev->orig()))
 782                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 783                 if (serialize($from_lines) != serialize($rev->closing()))
 784                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 785
 786
 787                 $prevtype = 'none';
 788                 foreach ($this->edits as $edit) {
 789                         if ( $prevtype == $edit->type )
 790                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 791                         $prevtype = $edit->type;
 792                 }
 793
 794                 $lcs = $this->lcs();
 795                 trigger_error("Diff okay: LCS = $lcs", E_USER_NOTICE);
 796         }
 797 }
 798
 799 /**
 800  * FIXME: bad name.
 801  */
 802 class MappedDiff
 803 extends Diff
 804 {
 805         /**
 806          * Constructor.
 807          *
 808          * Computes diff between sequences of strings.
 809          *
 810          * This can be used to compute things like
 811          * case-insensitve diffs, or diffs which ignore
 812          * changes in white-space.
 813          *
 814          * @param $from_lines array An array of strings.
 815          *      (Typically these are lines from a file.)
 816          *
 817          * @param $to_lines array An array of strings.
 818          *
 819          * @param $mapped_from_lines array This array should
 820          *      have the same size number of elements as $from_lines.
 821          *      The elements in $mapped_from_lines and
 822          *      $mapped_to_lines are what is actually compared
 823          *      when computing the diff.
 824          *
 825          * @param $mapped_to_lines array This array should
 826          *      have the same number of elements as $to_lines.
 827          */
 828         function MappedDiff($from_lines, $to_lines,
 829                                                 $mapped_from_lines, $mapped_to_lines) {
 830
 831                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 832                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 833
 834                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 835
 836                 $xi = $yi = 0;
 837                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 838                         $orig = &$this->edits[$i]->orig;
 839                         if (is_array($orig)) {
 840                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 841                                 $xi += sizeof($orig);
 842                         }
 843
 844                         $closing = &$this->edits[$i]->closing;
 845                         if (is_array($closing)) {
 846                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 847                                 $yi += sizeof($closing);
 848                         }
 849                 }
 850         }
 851 }
 852
 853 /**
 854  * A class to format Diffs
 855  *
 856  * This class formats the diff in classic diff format.
 857  * It is intended that this class be customized via inheritance,
 858  * to obtain fancier outputs.
 859  */
 860 class DiffFormatter
 861 {
 862         /**
 863          * Number of leading context "lines" to preserve.
 864          *
 865          * This should be left at zero for this class, but subclasses
 866          * may want to set this to other values.
 867          */
 868         var $leading_context_lines = 0;
 869
 870         /**
 871          * Number of trailing context "lines" to preserve.
 872          *
 873          * This should be left at zero for this class, but subclasses
 874          * may want to set this to other values.
 875          */
 876         var $trailing_context_lines = 0;
 877
 878         /**
 879          * Format a diff.
 880          *
 881          * @param $diff object A Diff object.
 882          * @return string The formatted output.
 883          */
 884         function format($diff) {
 885
 886                 $xi = $yi = 1;
 887                 $block = false;
 888                 $context = array();
 889
 890                 $nlead = $this->leading_context_lines;
 891                 $ntrail = $this->trailing_context_lines;
 892
 893                 $this->_start_diff();
 894
 895                 foreach ($diff->edits as $edit) {
 896                         if ($edit->type == 'copy') {
 897                                 if (is_array($block)) {
 898                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 899                                                 $block[] = $edit;
 900                                         }
 901                                         else{
 902                                                 if ($ntrail) {
 903                                                         $context = array_slice($edit->orig, 0, $ntrail);
 904                                                         $block[] = new _DiffOp_Copy($context);
 905                                                 }
 906                                                 $this->_block($x0, $ntrail + $xi - $x0,
 907                                                                           $y0, $ntrail + $yi - $y0,
 908                                                                           $block);
 909                                                 $block = false;
 910                                         }
 911                                 }
 912                                 $context = $edit->orig;
 913                         }
 914                         else {
 915                                 if (! is_array($block)) {
 916                                         $context = array_slice($context, sizeof($context) - $nlead);
 917                                         $x0 = $xi - sizeof($context);
 918                                         $y0 = $yi - sizeof($context);
 919                                         $block = array();
 920                                         if ($context)
 921                                                 $block[] = new _DiffOp_Copy($context);
 922                                 }
 923                                 $block[] = $edit;
 924                         }
 925
 926                         if ($edit->orig)
 927                                 $xi += sizeof($edit->orig);
 928                         if ($edit->closing)
 929                                 $yi += sizeof($edit->closing);
 930                 }
 931
 932                 if (is_array($block))
 933                         $this->_block($x0, $xi - $x0,
 934                                                   $y0, $yi - $y0,
 935                                                   $block);
 936
 937                 return $this->_end_diff();
 938         }
 939
 940         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 941                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 942                 foreach ($edits as $edit) {
 943                         if ($edit->type == 'copy')
 944                                 $this->_context($edit->orig);
 945                         elseif ($edit->type == 'add')
 946                                 $this->_added($edit->closing);
 947                         elseif ($edit->type == 'delete')
 948                                 $this->_deleted($edit->orig);
 949                         elseif ($edit->type == 'change')
 950                                 $this->_changed($edit->orig, $edit->closing);
 951                         else
 952                                 trigger_error("Unknown edit type", E_USER_ERROR);
 953                 }
 954                 $this->_end_block();
 955         }
 956
 957         function _start_diff() {
 958                 ob_start();
 959         }
 960
 961         function _end_diff() {
 962                 $val = ob_get_contents();
 963                 ob_end_clean();
 964                 return $val;
 965         }
 966
 967         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 968                 if ($xlen > 1)
 969                         $xbeg .= "," . ($xbeg + $xlen - 1);
 970                 if ($ylen > 1)
 971                         $ybeg .= "," . ($ybeg + $ylen - 1);
 972
 973                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 974         }
 975
 976         function _start_block($header) {
 977                 echo $header;
 978         }
 979
 980         function _end_block() {
 981         }
 982
 983         function _lines($lines, $prefix = ' ') {
 984                 foreach ($lines as $line)
 985                         echo "$prefix $line\n";
 986         }
 987
 988         function _context($lines) {
 989                 $this->_lines($lines);
 990         }
 991
 992         function _added($lines) {
 993                 $this->_lines($lines, ">");
 994         }
 995         function _deleted($lines) {
 996                 $this->_lines($lines, "<");
 997         }
 998
 999         function _changed($orig, $closing) {
1000                 $this->_deleted($orig);
1001                 echo "---\n";
1002                 $this->_added($closing);
1003         }
1004 }
1005
1006
1007 /**
1008  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1009  *
1010  */
1011
1012 define('NBSP', "\xA0");                 // iso-8859-x non-breaking space.
1013
1014 class _HWLDF_WordAccumulator {
1015         function _HWLDF_WordAccumulator () {
1016                 $this->_lines = array();
1017                 $this->_line = '';
1018                 $this->_group = '';
1019                 $this->_tag = '';
1020         }
1021
1022         function _flushGroup ($new_tag) {
1023                 if ($this->_group !== '') {
1024           if ($this->_tag == 'mark')
1025                         $this->_line .= "<font color=\"red\">$this->_group</font>";
1026           else
1027                 $this->_line .= $this->_group;
1028         }
1029                 $this->_group = '';
1030                 $this->_tag = $new_tag;
1031         }
1032
1033         function _flushLine ($new_tag) {
1034                 $this->_flushGroup($new_tag);
1035                 if ($this->_line != '')
1036                         $this->_lines[] = $this->_line;
1037                 $this->_line = '';
1038         }
1039
1040         function addWords ($words, $tag = '') {
1041                 if ($tag != $this->_tag)
1042                         $this->_flushGroup($tag);
1043
1044                 foreach ($words as $word) {
1045                         // new-line should only come as first char of word.
1046                         if ($word == '')
1047                                 continue;
1048                         if ($word[0] == "\n") {
1049                                 $this->_group .= NBSP;
1050                                 $this->_flushLine($tag);
1051                                 $word = substr($word, 1);
1052                         }
1053                         assert(!strstr($word, "\n"));
1054                         $this->_group .= $word;
1055                 }
1056         }
1057
1058         function getLines() {
1059                 $this->_flushLine('~done');
1060                 return $this->_lines;
1061         }
1062 }
1063
1064 class WordLevelDiff extends MappedDiff
1065 {
1066         function WordLevelDiff ($orig_lines, $closing_lines) {
1067                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1068                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1069
1070
1071                 $this->MappedDiff($orig_words, $closing_words,
1072                                                   $orig_stripped, $closing_stripped);
1073         }
1074
1075         function _split($lines) {
1076                 // FIXME: fix POSIX char class.
1077 #                if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1078                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1079                                                         implode("\n", $lines),
1080                                                         $m)) {
1081                         return array(array(''), array(''));
1082                 }
1083                 return array($m[0], $m[1]);
1084         }
1085
1086         function orig () {
1087                 $orig = new _HWLDF_WordAccumulator;
1088
1089                 foreach ($this->edits as $edit) {
1090                         if ($edit->type == 'copy')
1091                                 $orig->addWords($edit->orig);
1092                         elseif ($edit->orig)
1093                                 $orig->addWords($edit->orig, 'mark');
1094                 }
1095                 return $orig->getLines();
1096         }
1097
1098         function closing () {
1099                 $closing = new _HWLDF_WordAccumulator;
1100
1101                 foreach ($this->edits as $edit) {
1102                         if ($edit->type == 'copy')
1103                                 $closing->addWords($edit->closing);
1104                         elseif ($edit->closing)
1105                                 $closing->addWords($edit->closing, 'mark');
1106                 }
1107                 return $closing->getLines();
1108         }
1109 }
1110
1111 /**
1112  *      Wikipedia Table style diff formatter.
1113  *
1114  */
1115 class TableDiffFormatter extends DiffFormatter
1116 {
1117         function TableDiffFormatter() {
1118                 $this->leading_context_lines = 2;
1119                 $this->trailing_context_lines = 2;
1120         }
1121
1122         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1123                 $l1 = wfMsg( "lineno", $xbeg );
1124                 $l2 = wfMsg( "lineno", $ybeg );
1125
1126                 $r = "<tr><td colspan='2' align='left'><strong>{$l1}</strong></td>\n" .
1127                   "<td colspan='2' align='left'><strong>{$l2}</strong></td></tr>\n";
1128                 return $r;
1129         }
1130
1131         function _start_block( $header ) {
1132                 global $wgOut;
1133                 $wgOut->addHTML( $header );
1134         }
1135
1136         function _end_block() {
1137         }
1138
1139         function _lines( $lines, $prefix=' ', $color="white" ) {
1140         }
1141
1142         function addedLine( $line ) {
1143                 return "<td>+</td><td class='diff-addedline'>" .
1144                   "<small>{$line}</small></td>";
1145         }
1146
1147         function deletedLine( $line ) {
1148                 return "<td>-</td><td class='diff-deletedline'>" .
1149                   "<small>{$line}</small></td>";
1150         }
1151
1152         function emptyLine() {
1153                 return "<td colspan='2'>&nbsp;</td>";
1154         }
1155
1156         function contextLine( $line ) {
1157                 return "<td> </td><td class='diff-context'><small>{$line}</small></td>";
1158         }
1159
1160         function _added($lines) {
1161                 global $wgOut;
1162                 foreach ($lines as $line) {
1163                         $wgOut->addHTML( "<tr>" . $this->emptyLine() .
1164                           $this->addedLine( $line ) . "</tr>\n" );
1165                 }
1166         }
1167
1168         function _deleted($lines) {
1169                 global $wgOut;
1170                 foreach ($lines as $line) {
1171                         $wgOut->addHTML( "<tr>" . $this->deletedLine( $line ) .
1172                           $this->emptyLine() . "</tr>\n" );
1173                 }
1174         }
1175
1176         function _context( $lines ) {
1177                 global $wgOut;
1178                 foreach ($lines as $line) {
1179                         $wgOut->addHTML( "<tr>" . $this->contextLine( $line ) .
1180                           $this->contextLine( $line ) . "</tr>\n" );
1181                 }
1182         }
1183
1184         function _changed( $orig, $closing ) {
1185                 global $wgOut;
1186                 $diff = new WordLevelDiff( $orig, $closing );
1187                 $del = $diff->orig();
1188                 $add = $diff->closing();
1189
1190                 while ( $line = array_shift( $del ) ) {
1191                         $aline = array_shift( $add );
1192                         $wgOut->addHTML( "<tr>" . $this->deletedLine( $line ) .
1193                           $this->addedLine( $aline ) . "</tr>\n" );
1194                 }
1195                 $this->_added( $add ); # If any leftovers
1196         }
1197 }
1198
1199 ?>