includes/diff/DifferenceEngine.php

   1 <?php
   2 /**
   3  * @defgroup DifferenceEngine DifferenceEngine
   4  */
   5
   6 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
   7 //
   8 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
   9 // You may copy this code freely under the conditions of the GPL.
  10 //
  11
  12 /**
  13  * @todo document
  14  * @private
  15  * @ingroup DifferenceEngine
  16  */
  17 class _DiffOp {
  18         var $type;
  19         var $orig;
  20         var $closing;
  21
  22         function reverse() {
  23                 trigger_error('pure virtual', E_USER_ERROR);
  24         }
  25
  26         function norig() {
  27                 return $this->orig ? sizeof($this->orig) : 0;
  28         }
  29
  30         function nclosing() {
  31                 return $this->closing ? sizeof($this->closing) : 0;
  32         }
  33 }
  34
  35 /**
  36  * @todo document
  37  * @private
  38  * @ingroup DifferenceEngine
  39  */
  40 class _DiffOp_Copy extends _DiffOp {
  41         var $type = 'copy';
  42
  43         function _DiffOp_Copy ($orig, $closing = false) {
  44                 if (!is_array($closing))
  45                 $closing = $orig;
  46                 $this->orig = $orig;
  47                 $this->closing = $closing;
  48         }
  49
  50         function reverse() {
  51                 return new _DiffOp_Copy($this->closing, $this->orig);
  52         }
  53 }
  54
  55 /**
  56  * @todo document
  57  * @private
  58  * @ingroup DifferenceEngine
  59  */
  60 class _DiffOp_Delete extends _DiffOp {
  61         var $type = 'delete';
  62
  63         function _DiffOp_Delete ($lines) {
  64                 $this->orig = $lines;
  65                 $this->closing = false;
  66         }
  67
  68         function reverse() {
  69                 return new _DiffOp_Add($this->orig);
  70         }
  71 }
  72
  73 /**
  74  * @todo document
  75  * @private
  76  * @ingroup DifferenceEngine
  77  */
  78 class _DiffOp_Add extends _DiffOp {
  79         var $type = 'add';
  80
  81         function _DiffOp_Add ($lines) {
  82                 $this->closing = $lines;
  83                 $this->orig = false;
  84         }
  85
  86         function reverse() {
  87                 return new _DiffOp_Delete($this->closing);
  88         }
  89 }
  90
  91 /**
  92  * @todo document
  93  * @private
  94  * @ingroup DifferenceEngine
  95  */
  96 class _DiffOp_Change extends _DiffOp {
  97         var $type = 'change';
  98
  99         function _DiffOp_Change ($orig, $closing) {
 100                 $this->orig = $orig;
 101                 $this->closing = $closing;
 102         }
 103
 104         function reverse() {
 105                 return new _DiffOp_Change($this->closing, $this->orig);
 106         }
 107 }
 108
 109 /**
 110  * Class used internally by Diff to actually compute the diffs.
 111  *
 112  * The algorithm used here is mostly lifted from the perl module
 113  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 114  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 115  *
 116  * More ideas are taken from:
 117  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 118  *
 119  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 120  * diffutils-2.7, which can be found at:
 121  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 122  *
 123  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 124  * are my own.
 125  *
 126  * Line length limits for robustness added by Tim Starling, 2005-08-31
 127  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
 128  *
 129  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
 130  * @private
 131  * @ingroup DifferenceEngine
 132  */
 133 class _DiffEngine {
 134
 135         const MAX_XREF_LENGTH =  10000;
 136
 137         function diff ($from_lines, $to_lines){
 138                 wfProfileIn( __METHOD__ );
 139
 140                 // Diff and store locally
 141                 $this->diff_local($from_lines, $to_lines);
 142
 143                 // Merge edits when possible
 144                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 145                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 146
 147                 // Compute the edit operations.
 148                 $n_from = sizeof($from_lines);
 149                 $n_to = sizeof($to_lines);
 150
 151                 $edits = array();
 152                 $xi = $yi = 0;
 153                 while ($xi < $n_from || $yi < $n_to) {
 154                         assert($yi < $n_to || $this->xchanged[$xi]);
 155                         assert($xi < $n_from || $this->ychanged[$yi]);
 156
 157                         // Skip matching "snake".
 158                         $copy = array();
 159                         while ( $xi < $n_from && $yi < $n_to
 160                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 161                                 $copy[] = $from_lines[$xi++];
 162                                 ++$yi;
 163                         }
 164                         if ($copy)
 165                         $edits[] = new _DiffOp_Copy($copy);
 166
 167                         // Find deletes & adds.
 168                         $delete = array();
 169                         while ($xi < $n_from && $this->xchanged[$xi])
 170                         $delete[] = $from_lines[$xi++];
 171
 172                         $add = array();
 173                         while ($yi < $n_to && $this->ychanged[$yi])
 174                         $add[] = $to_lines[$yi++];
 175
 176                         if ($delete && $add)
 177                         $edits[] = new _DiffOp_Change($delete, $add);
 178                         elseif ($delete)
 179                         $edits[] = new _DiffOp_Delete($delete);
 180                         elseif ($add)
 181                         $edits[] = new _DiffOp_Add($add);
 182                 }
 183                 wfProfileOut( __METHOD__ );
 184                 return $edits;
 185         }
 186
 187         function diff_local ($from_lines, $to_lines) {
 188                 global $wgExternalDiffEngine;
 189                 wfProfileIn( __METHOD__);
 190
 191                 if($wgExternalDiffEngine == 'wikidiff3'){
 192                         // wikidiff3
 193                         $wikidiff3 = new WikiDiff3();
 194                         $wikidiff3->diff($from_lines, $to_lines);
 195                         $this->xchanged = $wikidiff3->removed;
 196                         $this->ychanged = $wikidiff3->added;
 197                         unset($wikidiff3);
 198                 }else{
 199                         // old diff
 200                         $n_from = sizeof($from_lines);
 201                         $n_to = sizeof($to_lines);
 202                         $this->xchanged = $this->ychanged = array();
 203                         $this->xv = $this->yv = array();
 204                         $this->xind = $this->yind = array();
 205                         unset($this->seq);
 206                         unset($this->in_seq);
 207                         unset($this->lcs);
 208
 209                         // Skip leading common lines.
 210                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 211                                 if ($from_lines[$skip] !== $to_lines[$skip])
 212                                 break;
 213                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 214                         }
 215                         // Skip trailing common lines.
 216                         $xi = $n_from; $yi = $n_to;
 217                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 218                                 if ($from_lines[$xi] !== $to_lines[$yi])
 219                                 break;
 220                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 221                         }
 222
 223                         // Ignore lines which do not exist in both files.
 224                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 225                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 226                         }
 227
 228                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 229                                 $line = $to_lines[$yi];
 230                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 231                                 continue;
 232                                 $yhash[$this->_line_hash($line)] = 1;
 233                                 $this->yv[] = $line;
 234                                 $this->yind[] = $yi;
 235                         }
 236                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 237                                 $line = $from_lines[$xi];
 238                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 239                                 continue;
 240                                 $this->xv[] = $line;
 241                                 $this->xind[] = $xi;
 242                         }
 243
 244                         // Find the LCS.
 245                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 246                 }
 247                 wfProfileOut( __METHOD__ );
 248         }
 249
 250         /**
 251          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 252          */
 253         function _line_hash( $line ) {
 254                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 255                         return md5( $line );
 256                 } else {
 257                         return $line;
 258                 }
 259         }
 260
 261         /* Divide the Largest Common Subsequence (LCS) of the sequences
 262          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 263          * sized segments.
 264          *
 265          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 266          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 267          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 268          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 269          * that (X0, Y0) == (XOFF, YOFF) and
 270          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 271          *
 272          * This function assumes that the first lines of the specified portions
 273          * of the two files do not match, and likewise that the last lines do not
 274          * match.  The caller must trim matching lines from the beginning and end
 275          * of the portions it is going to specify.
 276          */
 277         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 278                 $flip = false;
 279
 280                 if ($xlim - $xoff > $ylim - $yoff) {
 281                         // Things seems faster (I'm not sure I understand why)
 282                         // when the shortest sequence in X.
 283                         $flip = true;
 284                         list ($xoff, $xlim, $yoff, $ylim)
 285                         = array( $yoff, $ylim, $xoff, $xlim);
 286                 }
 287
 288                 if ($flip)
 289                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 290                 $ymatches[$this->xv[$i]][] = $i;
 291                 else
 292                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 293                 $ymatches[$this->yv[$i]][] = $i;
 294
 295                 $this->lcs = 0;
 296                 $this->seq[0]= $yoff - 1;
 297                 $this->in_seq = array();
 298                 $ymids[0] = array();
 299
 300                 $numer = $xlim - $xoff + $nchunks - 1;
 301                 $x = $xoff;
 302                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 303                         if ($chunk > 0)
 304                         for ($i = 0; $i <= $this->lcs; $i++)
 305                         $ymids[$i][$chunk-1] = $this->seq[$i];
 306
 307                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 308                         for ( ; $x < $x1; $x++) {
 309                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 310                                 if (empty($ymatches[$line]))
 311                                 continue;
 312                                 $matches = $ymatches[$line];
 313                                 reset($matches);
 314                                 while (list ($junk, $y) = each($matches))
 315                                 if (empty($this->in_seq[$y])) {
 316                                         $k = $this->_lcs_pos($y);
 317                                         assert($k > 0);
 318                                         $ymids[$k] = $ymids[$k-1];
 319                                         break;
 320                                 }
 321                                 while (list ( /* $junk */, $y) = each($matches)) {
 322                                         if ($y > $this->seq[$k-1]) {
 323                                                 assert($y < $this->seq[$k]);
 324                                                 // Optimization: this is a common case:
 325                                                 //      next match is just replacing previous match.
 326                                                 $this->in_seq[$this->seq[$k]] = false;
 327                                                 $this->seq[$k] = $y;
 328                                                 $this->in_seq[$y] = 1;
 329                                         } else if (empty($this->in_seq[$y])) {
 330                                                 $k = $this->_lcs_pos($y);
 331                                                 assert($k > 0);
 332                                                 $ymids[$k] = $ymids[$k-1];
 333                                         }
 334                                 }
 335                         }
 336                 }
 337
 338                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 339                 $ymid = $ymids[$this->lcs];
 340                 for ($n = 0; $n < $nchunks - 1; $n++) {
 341                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 342                         $y1 = $ymid[$n] + 1;
 343                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 344                 }
 345                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 346
 347                 return array($this->lcs, $seps);
 348         }
 349
 350         function _lcs_pos ($ypos) {
 351                 $end = $this->lcs;
 352                 if ($end == 0 || $ypos > $this->seq[$end]) {
 353                         $this->seq[++$this->lcs] = $ypos;
 354                         $this->in_seq[$ypos] = 1;
 355                         return $this->lcs;
 356                 }
 357
 358                 $beg = 1;
 359                 while ($beg < $end) {
 360                         $mid = (int)(($beg + $end) / 2);
 361                         if ( $ypos > $this->seq[$mid] )
 362                         $beg = $mid + 1;
 363                         else
 364                         $end = $mid;
 365                 }
 366
 367                 assert($ypos != $this->seq[$end]);
 368
 369                 $this->in_seq[$this->seq[$end]] = false;
 370                 $this->seq[$end] = $ypos;
 371                 $this->in_seq[$ypos] = 1;
 372                 return $end;
 373         }
 374
 375         /* Find LCS of two sequences.
 376          *
 377          * The results are recorded in the vectors $this->{x,y}changed[], by
 378          * storing a 1 in the element for each line that is an insertion
 379          * or deletion (ie. is not in the LCS).
 380          *
 381          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 382          *
 383          * Note that XLIM, YLIM are exclusive bounds.
 384          * All line numbers are origin-0 and discarded lines are not counted.
 385          */
 386         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 387                 // Slide down the bottom initial diagonal.
 388                 while ($xoff < $xlim && $yoff < $ylim
 389                 && $this->xv[$xoff] == $this->yv[$yoff]) {
 390                         ++$xoff;
 391                         ++$yoff;
 392                 }
 393
 394                 // Slide up the top initial diagonal.
 395                 while ($xlim > $xoff && $ylim > $yoff
 396                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 397                         --$xlim;
 398                         --$ylim;
 399                 }
 400
 401                 if ($xoff == $xlim || $yoff == $ylim)
 402                 $lcs = 0;
 403                 else {
 404                         // This is ad hoc but seems to work well.
 405                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 406                         //$nchunks = max(2,min(8,(int)$nchunks));
 407                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 408                         list ($lcs, $seps)
 409                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 410                 }
 411
 412                 if ($lcs == 0) {
 413                         // X and Y sequences have no common subsequence:
 414                         // mark all changed.
 415                         while ($yoff < $ylim)
 416                         $this->ychanged[$this->yind[$yoff++]] = 1;
 417                         while ($xoff < $xlim)
 418                         $this->xchanged[$this->xind[$xoff++]] = 1;
 419                 } else {
 420                         // Use the partitions to split this problem into subproblems.
 421                         reset($seps);
 422                         $pt1 = $seps[0];
 423                         while ($pt2 = next($seps)) {
 424                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 425                                 $pt1 = $pt2;
 426                         }
 427                 }
 428         }
 429
 430         /* Adjust inserts/deletes of identical lines to join changes
 431          * as much as possible.
 432          *
 433          * We do something when a run of changed lines include a
 434          * line at one end and has an excluded, identical line at the other.
 435          * We are free to choose which identical line is included.
 436          * `compareseq' usually chooses the one at the beginning,
 437          * but usually it is cleaner to consider the following identical line
 438          * to be the "change".
 439          *
 440          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 441          */
 442         function _shift_boundaries ($lines, &$changed, $other_changed) {
 443                 wfProfileIn( __METHOD__ );
 444                 $i = 0;
 445                 $j = 0;
 446
 447                 assert('sizeof($lines) == sizeof($changed)');
 448                 $len = sizeof($lines);
 449                 $other_len = sizeof($other_changed);
 450
 451                 while (1) {
 452                         /*
 453                          * Scan forwards to find beginning of another run of changes.
 454                          * Also keep track of the corresponding point in the other file.
 455                          *
 456                          * Throughout this code, $i and $j are adjusted together so that
 457                          * the first $i elements of $changed and the first $j elements
 458                          * of $other_changed both contain the same number of zeros
 459                          * (unchanged lines).
 460                          * Furthermore, $j is always kept so that $j == $other_len or
 461                          * $other_changed[$j] == false.
 462                          */
 463                         while ($j < $other_len && $other_changed[$j])
 464                         $j++;
 465
 466                         while ($i < $len && ! $changed[$i]) {
 467                                 assert('$j < $other_len && ! $other_changed[$j]');
 468                                 $i++; $j++;
 469                                 while ($j < $other_len && $other_changed[$j])
 470                                 $j++;
 471                         }
 472
 473                         if ($i == $len)
 474                         break;
 475
 476                         $start = $i;
 477
 478                         // Find the end of this run of changes.
 479                         while (++$i < $len && $changed[$i])
 480                         continue;
 481
 482                         do {
 483                                 /*
 484                                  * Record the length of this run of changes, so that
 485                                  * we can later determine whether the run has grown.
 486                                  */
 487                                 $runlength = $i - $start;
 488
 489                                 /*
 490                                  * Move the changed region back, so long as the
 491                                  * previous unchanged line matches the last changed one.
 492                                  * This merges with previous changed regions.
 493                                  */
 494                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 495                                         $changed[--$start] = 1;
 496                                         $changed[--$i] = false;
 497                                         while ($start > 0 && $changed[$start - 1])
 498                                         $start--;
 499                                         assert('$j > 0');
 500                                         while ($other_changed[--$j])
 501                                         continue;
 502                                         assert('$j >= 0 && !$other_changed[$j]');
 503                                 }
 504
 505                                 /*
 506                                  * Set CORRESPONDING to the end of the changed run, at the last
 507                                  * point where it corresponds to a changed run in the other file.
 508                                  * CORRESPONDING == LEN means no such point has been found.
 509                                  */
 510                                 $corresponding = $j < $other_len ? $i : $len;
 511
 512                                 /*
 513                                  * Move the changed region forward, so long as the
 514                                  * first changed line matches the following unchanged one.
 515                                  * This merges with following changed regions.
 516                                  * Do this second, so that if there are no merges,
 517                                  * the changed region is moved forward as far as possible.
 518                                  */
 519                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 520                                         $changed[$start++] = false;
 521                                         $changed[$i++] = 1;
 522                                         while ($i < $len && $changed[$i])
 523                                         $i++;
 524
 525                                         assert('$j < $other_len && ! $other_changed[$j]');
 526                                         $j++;
 527                                         if ($j < $other_len && $other_changed[$j]) {
 528                                                 $corresponding = $i;
 529                                                 while ($j < $other_len && $other_changed[$j])
 530                                                 $j++;
 531                                         }
 532                                 }
 533                         } while ($runlength != $i - $start);
 534
 535                         /*
 536                          * If possible, move the fully-merged run of changes
 537                          * back to a corresponding run in the other file.
 538                          */
 539                         while ($corresponding < $i) {
 540                                 $changed[--$start] = 1;
 541                                 $changed[--$i] = 0;
 542                                 assert('$j > 0');
 543                                 while ($other_changed[--$j])
 544                                 continue;
 545                                 assert('$j >= 0 && !$other_changed[$j]');
 546                         }
 547                 }
 548                 wfProfileOut( __METHOD__ );
 549         }
 550 }
 551
 552 /**
 553  * Class representing a 'diff' between two sequences of strings.
 554  * @todo document
 555  * @private
 556  * @ingroup DifferenceEngine
 557  */
 558 class Diff
 559 {
 560         var $edits;
 561
 562         /**
 563          * Constructor.
 564          * Computes diff between sequences of strings.
 565          *
 566          * @param $from_lines array An array of strings.
 567          *                (Typically these are lines from a file.)
 568          * @param $to_lines array An array of strings.
 569          */
 570         function Diff($from_lines, $to_lines) {
 571                 $eng = new _DiffEngine;
 572                 $this->edits = $eng->diff($from_lines, $to_lines);
 573                 //$this->_check($from_lines, $to_lines);
 574         }
 575
 576         /**
 577          * Compute reversed Diff.
 578          *
 579          * SYNOPSIS:
 580          *
 581          *      $diff = new Diff($lines1, $lines2);
 582          *      $rev = $diff->reverse();
 583          * @return object A Diff object representing the inverse of the
 584          *                                original diff.
 585          */
 586         function reverse () {
 587                 $rev = $this;
 588                 $rev->edits = array();
 589                 foreach ($this->edits as $edit) {
 590                         $rev->edits[] = $edit->reverse();
 591                 }
 592                 return $rev;
 593         }
 594
 595         /**
 596          * Check for empty diff.
 597          *
 598          * @return bool True iff two sequences were identical.
 599          */
 600         function isEmpty () {
 601                 foreach ($this->edits as $edit) {
 602                         if ($edit->type != 'copy')
 603                         return false;
 604                 }
 605                 return true;
 606         }
 607
 608         /**
 609          * Compute the length of the Longest Common Subsequence (LCS).
 610          *
 611          * This is mostly for diagnostic purposed.
 612          *
 613          * @return int The length of the LCS.
 614          */
 615         function lcs () {
 616                 $lcs = 0;
 617                 foreach ($this->edits as $edit) {
 618                         if ($edit->type == 'copy')
 619                         $lcs += sizeof($edit->orig);
 620                 }
 621                 return $lcs;
 622         }
 623
 624         /**
 625          * Get the original set of lines.
 626          *
 627          * This reconstructs the $from_lines parameter passed to the
 628          * constructor.
 629          *
 630          * @return array The original sequence of strings.
 631          */
 632         function orig() {
 633                 $lines = array();
 634
 635                 foreach ($this->edits as $edit) {
 636                         if ($edit->orig)
 637                         array_splice($lines, sizeof($lines), 0, $edit->orig);
 638                 }
 639                 return $lines;
 640         }
 641
 642         /**
 643          * Get the closing set of lines.
 644          *
 645          * This reconstructs the $to_lines parameter passed to the
 646          * constructor.
 647          *
 648          * @return array The sequence of strings.
 649          */
 650         function closing() {
 651                 $lines = array();
 652
 653                 foreach ($this->edits as $edit) {
 654                         if ($edit->closing)
 655                         array_splice($lines, sizeof($lines), 0, $edit->closing);
 656                 }
 657                 return $lines;
 658         }
 659
 660         /**
 661          * Check a Diff for validity.
 662          *
 663          * This is here only for debugging purposes.
 664          */
 665         function _check ($from_lines, $to_lines) {
 666                 wfProfileIn( __METHOD__ );
 667                 if (serialize($from_lines) != serialize($this->orig()))
 668                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 669                 if (serialize($to_lines) != serialize($this->closing()))
 670                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 671
 672                 $rev = $this->reverse();
 673                 if (serialize($to_lines) != serialize($rev->orig()))
 674                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
 675                 if (serialize($from_lines) != serialize($rev->closing()))
 676                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 677
 678
 679                 $prevtype = 'none';
 680                 foreach ($this->edits as $edit) {
 681                         if ( $prevtype == $edit->type )
 682                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 683                         $prevtype = $edit->type;
 684                 }
 685
 686                 $lcs = $this->lcs();
 687                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 688                 wfProfileOut( __METHOD__ );
 689         }
 690 }
 691
 692 /**
 693  * @todo document, bad name.
 694  * @private
 695  * @ingroup DifferenceEngine
 696  */
 697 class MappedDiff extends Diff
 698 {
 699         /**
 700          * Constructor.
 701          *
 702          * Computes diff between sequences of strings.
 703          *
 704          * This can be used to compute things like
 705          * case-insensitve diffs, or diffs which ignore
 706          * changes in white-space.
 707          *
 708          * @param $from_lines array An array of strings.
 709          *      (Typically these are lines from a file.)
 710          *
 711          * @param $to_lines array An array of strings.
 712          *
 713          * @param $mapped_from_lines array This array should
 714          *      have the same size number of elements as $from_lines.
 715          *      The elements in $mapped_from_lines and
 716          *      $mapped_to_lines are what is actually compared
 717          *      when computing the diff.
 718          *
 719          * @param $mapped_to_lines array This array should
 720          *      have the same number of elements as $to_lines.
 721          */
 722         function MappedDiff($from_lines, $to_lines,
 723         $mapped_from_lines, $mapped_to_lines) {
 724                 wfProfileIn( __METHOD__ );
 725
 726                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 727                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 728
 729                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 730
 731                 $xi = $yi = 0;
 732                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 733                         $orig = &$this->edits[$i]->orig;
 734                         if (is_array($orig)) {
 735                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 736                                 $xi += sizeof($orig);
 737                         }
 738
 739                         $closing = &$this->edits[$i]->closing;
 740                         if (is_array($closing)) {
 741                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 742                                 $yi += sizeof($closing);
 743                         }
 744                 }
 745                 wfProfileOut( __METHOD__ );
 746         }
 747 }
 748
 749 /**
 750  * A class to format Diffs
 751  *
 752  * This class formats the diff in classic diff format.
 753  * It is intended that this class be customized via inheritance,
 754  * to obtain fancier outputs.
 755  * @todo document
 756  * @private
 757  * @ingroup DifferenceEngine
 758  */
 759 class DiffFormatter {
 760         /**
 761          * Number of leading context "lines" to preserve.
 762          *
 763          * This should be left at zero for this class, but subclasses
 764          * may want to set this to other values.
 765          */
 766         var $leading_context_lines = 0;
 767
 768         /**
 769          * Number of trailing context "lines" to preserve.
 770          *
 771          * This should be left at zero for this class, but subclasses
 772          * may want to set this to other values.
 773          */
 774         var $trailing_context_lines = 0;
 775
 776         /**
 777          * Format a diff.
 778          *
 779          * @param $diff object A Diff object.
 780          * @return string The formatted output.
 781          */
 782         function format($diff) {
 783                 wfProfileIn( __METHOD__ );
 784
 785                 $xi = $yi = 1;
 786                 $block = false;
 787                 $context = array();
 788
 789                 $nlead = $this->leading_context_lines;
 790                 $ntrail = $this->trailing_context_lines;
 791
 792                 $this->_start_diff();
 793
 794                 foreach ($diff->edits as $edit) {
 795                         if ($edit->type == 'copy') {
 796                                 if (is_array($block)) {
 797                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 798                                                 $block[] = $edit;
 799                                         }
 800                                         else{
 801                                                 if ($ntrail) {
 802                                                         $context = array_slice($edit->orig, 0, $ntrail);
 803                                                         $block[] = new _DiffOp_Copy($context);
 804                                                 }
 805                                                 $this->_block($x0, $ntrail + $xi - $x0,
 806                                                 $y0, $ntrail + $yi - $y0,
 807                                                 $block);
 808                                                 $block = false;
 809                                         }
 810                                 }
 811                                 $context = $edit->orig;
 812                         }
 813                         else {
 814                                 if (! is_array($block)) {
 815                                         $context = array_slice($context, sizeof($context) - $nlead);
 816                                         $x0 = $xi - sizeof($context);
 817                                         $y0 = $yi - sizeof($context);
 818                                         $block = array();
 819                                         if ($context)
 820                                         $block[] = new _DiffOp_Copy($context);
 821                                 }
 822                                 $block[] = $edit;
 823                         }
 824
 825                         if ($edit->orig)
 826                         $xi += sizeof($edit->orig);
 827                         if ($edit->closing)
 828                         $yi += sizeof($edit->closing);
 829                 }
 830
 831                 if (is_array($block))
 832                 $this->_block($x0, $xi - $x0,
 833                 $y0, $yi - $y0,
 834                 $block);
 835
 836                 $end = $this->_end_diff();
 837                 wfProfileOut( __METHOD__ );
 838                 return $end;
 839         }
 840
 841         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 842                 wfProfileIn( __METHOD__ );
 843                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 844                 foreach ($edits as $edit) {
 845                         if ($edit->type == 'copy')
 846                         $this->_context($edit->orig);
 847                         elseif ($edit->type == 'add')
 848                         $this->_added($edit->closing);
 849                         elseif ($edit->type == 'delete')
 850                         $this->_deleted($edit->orig);
 851                         elseif ($edit->type == 'change')
 852                         $this->_changed($edit->orig, $edit->closing);
 853                         else
 854                         trigger_error('Unknown edit type', E_USER_ERROR);
 855                 }
 856                 $this->_end_block();
 857                 wfProfileOut( __METHOD__ );
 858         }
 859
 860         function _start_diff() {
 861                 ob_start();
 862         }
 863
 864         function _end_diff() {
 865                 $val = ob_get_contents();
 866                 ob_end_clean();
 867                 return $val;
 868         }
 869
 870         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 871                 if ($xlen > 1)
 872                 $xbeg .= "," . ($xbeg + $xlen - 1);
 873                 if ($ylen > 1)
 874                 $ybeg .= "," . ($ybeg + $ylen - 1);
 875
 876                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 877         }
 878
 879         function _start_block($header) {
 880                 echo $header . "\n";
 881         }
 882
 883         function _end_block() {
 884         }
 885
 886         function _lines($lines, $prefix = ' ') {
 887                 foreach ($lines as $line)
 888                 echo "$prefix $line\n";
 889         }
 890
 891         function _context($lines) {
 892                 $this->_lines($lines);
 893         }
 894
 895         function _added($lines) {
 896                 $this->_lines($lines, '>');
 897         }
 898         function _deleted($lines) {
 899                 $this->_lines($lines, '<');
 900         }
 901
 902         function _changed($orig, $closing) {
 903                 $this->_deleted($orig);
 904                 echo "---\n";
 905                 $this->_added($closing);
 906         }
 907 }
 908
 909 /**
 910  * A formatter that outputs unified diffs
 911  * @ingroup DifferenceEngine
 912  */
 913
 914 class UnifiedDiffFormatter extends DiffFormatter {
 915         var $leading_context_lines = 2;
 916         var $trailing_context_lines = 2;
 917
 918         function _added($lines) {
 919                 $this->_lines($lines, '+');
 920         }
 921         function _deleted($lines) {
 922                 $this->_lines($lines, '-');
 923         }
 924         function _changed($orig, $closing) {
 925                 $this->_deleted($orig);
 926                 $this->_added($closing);
 927         }
 928         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 929                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
 930         }
 931 }
 932
 933 /**
 934  * A pseudo-formatter that just passes along the Diff::$edits array
 935  * @ingroup DifferenceEngine
 936  */
 937 class ArrayDiffFormatter extends DiffFormatter {
 938         function format($diff) {
 939                 $oldline = 1;
 940                 $newline = 1;
 941                 $retval = array();
 942                 foreach($diff->edits as $edit)
 943                 switch($edit->type) {
 944                         case 'add':
 945                                 foreach($edit->closing as $l) {
 946                                         $retval[] = array(
 947                                                         'action' => 'add',
 948                                                         'new'=> $l,
 949                                                         'newline' => $newline++
 950                                         );
 951                                 }
 952                                 break;
 953                         case 'delete':
 954                                 foreach($edit->orig as $l) {
 955                                         $retval[] = array(
 956                                                         'action' => 'delete',
 957                                                         'old' => $l,
 958                                                         'oldline' => $oldline++,
 959                                         );
 960                                 }
 961                                 break;
 962                         case 'change':
 963                                 foreach($edit->orig as $i => $l) {
 964                                         $retval[] = array(
 965                                                         'action' => 'change',
 966                                                         'old' => $l,
 967                                                         'new' => @$edit->closing[$i],
 968                                                         'oldline' => $oldline++,
 969                                                         'newline' => $newline++,
 970                                         );
 971                                 }
 972                                 break;
 973                         case 'copy':
 974                                 $oldline += count($edit->orig);
 975                                 $newline += count($edit->orig);
 976                 }
 977                 return $retval;
 978         }
 979 }
 980
 981 /**
 982  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
 983  *
 984  */
 985
 986 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
 987
 988 /**
 989  * @todo document
 990  * @private
 991  * @ingroup DifferenceEngine
 992  */
 993 class _HWLDF_WordAccumulator {
 994         function _HWLDF_WordAccumulator () {
 995                 $this->_lines = array();
 996                 $this->_line = '';
 997                 $this->_group = '';
 998                 $this->_tag = '';
 999         }
1000
1001         function _flushGroup ($new_tag) {
1002                 if ($this->_group !== '') {
1003                         if ($this->_tag == 'ins')
1004                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1005                         htmlspecialchars ( $this->_group ) . '</ins>';
1006                         elseif ($this->_tag == 'del')
1007                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1008                         htmlspecialchars ( $this->_group ) . '</del>';
1009                         else
1010                         $this->_line .= htmlspecialchars ( $this->_group );
1011                 }
1012                 $this->_group = '';
1013                 $this->_tag = $new_tag;
1014         }
1015
1016         function _flushLine ($new_tag) {
1017                 $this->_flushGroup($new_tag);
1018                 if ($this->_line != '')
1019                 array_push ( $this->_lines, $this->_line );
1020                 else
1021                 # make empty lines visible by inserting an NBSP
1022                 array_push ( $this->_lines, NBSP );
1023                 $this->_line = '';
1024         }
1025
1026         function addWords ($words, $tag = '') {
1027                 if ($tag != $this->_tag)
1028                 $this->_flushGroup($tag);
1029
1030                 foreach ($words as $word) {
1031                         // new-line should only come as first char of word.
1032                         if ($word == '')
1033                         continue;
1034                         if ($word[0] == "\n") {
1035                                 $this->_flushLine($tag);
1036                                 $word = substr($word, 1);
1037                         }
1038                         assert(!strstr($word, "\n"));
1039                         $this->_group .= $word;
1040                 }
1041         }
1042
1043         function getLines() {
1044                 $this->_flushLine('~done');
1045                 return $this->_lines;
1046         }
1047 }
1048
1049 /**
1050  * @todo document
1051  * @private
1052  * @ingroup DifferenceEngine
1053  */
1054 class WordLevelDiff extends MappedDiff {
1055         const MAX_LINE_LENGTH = 10000;
1056
1057         function WordLevelDiff ($orig_lines, $closing_lines) {
1058                 wfProfileIn( __METHOD__ );
1059
1060                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1061                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1062
1063                 $this->MappedDiff($orig_words, $closing_words,
1064                 $orig_stripped, $closing_stripped);
1065                 wfProfileOut( __METHOD__ );
1066         }
1067
1068         function _split($lines) {
1069                 wfProfileIn( __METHOD__ );
1070
1071                 $words = array();
1072                 $stripped = array();
1073                 $first = true;
1074                 foreach ( $lines as $line ) {
1075                         # If the line is too long, just pretend the entire line is one big word
1076                         # This prevents resource exhaustion problems
1077                         if ( $first ) {
1078                                 $first = false;
1079                         } else {
1080                                 $words[] = "\n";
1081                                 $stripped[] = "\n";
1082                         }
1083                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1084                                 $words[] = $line;
1085                                 $stripped[] = $line;
1086                         } else {
1087                                 $m = array();
1088                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1089                                 $line, $m))
1090                                 {
1091                                         $words = array_merge( $words, $m[0] );
1092                                         $stripped = array_merge( $stripped, $m[1] );
1093                                 }
1094                         }
1095                 }
1096                 wfProfileOut( __METHOD__ );
1097                 return array($words, $stripped);
1098         }
1099
1100         function orig () {
1101                 wfProfileIn( __METHOD__ );
1102                 $orig = new _HWLDF_WordAccumulator;
1103
1104                 foreach ($this->edits as $edit) {
1105                         if ($edit->type == 'copy')
1106                         $orig->addWords($edit->orig);
1107                         elseif ($edit->orig)
1108                         $orig->addWords($edit->orig, 'del');
1109                 }
1110                 $lines = $orig->getLines();
1111                 wfProfileOut( __METHOD__ );
1112                 return $lines;
1113         }
1114
1115         function closing () {
1116                 wfProfileIn( __METHOD__ );
1117                 $closing = new _HWLDF_WordAccumulator;
1118
1119                 foreach ($this->edits as $edit) {
1120                         if ($edit->type == 'copy')
1121                         $closing->addWords($edit->closing);
1122                         elseif ($edit->closing)
1123                         $closing->addWords($edit->closing, 'ins');
1124                 }
1125                 $lines = $closing->getLines();
1126                 wfProfileOut( __METHOD__ );
1127                 return $lines;
1128         }
1129 }
1130
1131 /**
1132  * Wikipedia Table style diff formatter.
1133  * @todo document
1134  * @private
1135  * @ingroup DifferenceEngine
1136  */
1137 class TableDiffFormatter extends DiffFormatter {
1138         function TableDiffFormatter() {
1139                 $this->leading_context_lines = 2;
1140                 $this->trailing_context_lines = 2;
1141         }
1142
1143         public static function escapeWhiteSpace( $msg ) {
1144                 $msg = preg_replace( '/^ /m', '&#160; ', $msg );
1145                 $msg = preg_replace( '/ $/m', ' &#160;', $msg );
1146                 $msg = preg_replace( '/  /', '&#160; ', $msg );
1147                 return $msg;
1148         }
1149
1150         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1151                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1152                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1153                 return $r;
1154         }
1155
1156         function _start_block( $header ) {
1157                 echo $header;
1158         }
1159
1160         function _end_block() {
1161         }
1162
1163         function _lines( $lines, $prefix=' ', $color='white' ) {
1164         }
1165
1166         # HTML-escape parameter before calling this
1167         function addedLine( $line ) {
1168                 return $this->wrapLine( '+', 'diff-addedline', $line );
1169         }
1170
1171         # HTML-escape parameter before calling this
1172         function deletedLine( $line ) {
1173                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1174         }
1175
1176         # HTML-escape parameter before calling this
1177         function contextLine( $line ) {
1178                 return $this->wrapLine( ' ', 'diff-context', $line );
1179         }
1180
1181         private function wrapLine( $marker, $class, $line ) {
1182                 if( $line !== '' ) {
1183                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1184                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1185                 }
1186                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1187         }
1188
1189         function emptyLine() {
1190                 return '<td colspan="2">&#160;</td>';
1191         }
1192
1193         function _added( $lines ) {
1194                 foreach ($lines as $line) {
1195                         echo '<tr>' . $this->emptyLine() .
1196                         $this->addedLine( '<ins class="diffchange">' .
1197                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1198                 }
1199         }
1200
1201         function _deleted($lines) {
1202                 foreach ($lines as $line) {
1203                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1204                         htmlspecialchars ( $line ) . '</del>' ) .
1205                         $this->emptyLine() . "</tr>\n";
1206                 }
1207         }
1208
1209         function _context( $lines ) {
1210                 foreach ($lines as $line) {
1211                         echo '<tr>' .
1212                         $this->contextLine( htmlspecialchars ( $line ) ) .
1213                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1214                 }
1215         }
1216
1217         function _changed( $orig, $closing ) {
1218                 wfProfileIn( __METHOD__ );
1219
1220                 $diff = new WordLevelDiff( $orig, $closing );
1221                 $del = $diff->orig();
1222                 $add = $diff->closing();
1223
1224                 # Notice that WordLevelDiff returns HTML-escaped output.
1225                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1226
1227                 while ( $line = array_shift( $del ) ) {
1228                         $aline = array_shift( $add );
1229                         echo '<tr>' . $this->deletedLine( $line ) .
1230                         $this->addedLine( $aline ) . "</tr>\n";
1231                 }
1232                 foreach ($add as $line) {       # If any leftovers
1233                         echo '<tr>' . $this->emptyLine() .
1234                         $this->addedLine( $line ) . "</tr>\n";
1235                 }
1236                 wfProfileOut( __METHOD__ );
1237         }
1238 }