includes/DifferenceEngine.php

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