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