includes/diff/DifferenceEngine.php

   1 <?php
   2 /**
   3  * A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
   4  *
   5  * Copyright © 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
   6  * You may copy this code freely under the conditions of the GPL.
   7  *
   8  * @file
   9  * @ingroup DifferenceEngine
  10  * @defgroup DifferenceEngine DifferenceEngine
  11  */
  12
  13 /**
  14  * @todo document
  15  * @private
  16  * @ingroup DifferenceEngine
  17  */
  18 class _DiffOp {
  19         var $type;
  20         var $orig;
  21         var $closing;
  22
  23         function reverse() {
  24                 trigger_error('pure virtual', E_USER_ERROR);
  25         }
  26
  27         function norig() {
  28                 return $this->orig ? sizeof($this->orig) : 0;
  29         }
  30
  31         function nclosing() {
  32                 return $this->closing ? sizeof($this->closing) : 0;
  33         }
  34 }
  35
  36 /**
  37  * @todo document
  38  * @private
  39  * @ingroup DifferenceEngine
  40  */
  41 class _DiffOp_Copy extends _DiffOp {
  42         var $type = 'copy';
  43
  44         function _DiffOp_Copy ($orig, $closing = false) {
  45                 if (!is_array($closing))
  46                 $closing = $orig;
  47                 $this->orig = $orig;
  48                 $this->closing = $closing;
  49         }
  50
  51         function reverse() {
  52                 return new _DiffOp_Copy($this->closing, $this->orig);
  53         }
  54 }
  55
  56 /**
  57  * @todo document
  58  * @private
  59  * @ingroup DifferenceEngine
  60  */
  61 class _DiffOp_Delete extends _DiffOp {
  62         var $type = 'delete';
  63
  64         function _DiffOp_Delete ($lines) {
  65                 $this->orig = $lines;
  66                 $this->closing = false;
  67         }
  68
  69         function reverse() {
  70                 return new _DiffOp_Add($this->orig);
  71         }
  72 }
  73
  74 /**
  75  * @todo document
  76  * @private
  77  * @ingroup DifferenceEngine
  78  */
  79 class _DiffOp_Add extends _DiffOp {
  80         var $type = 'add';
  81
  82         function _DiffOp_Add ($lines) {
  83                 $this->closing = $lines;
  84                 $this->orig = false;
  85         }
  86
  87         function reverse() {
  88                 return new _DiffOp_Delete($this->closing);
  89         }
  90 }
  91
  92 /**
  93  * @todo document
  94  * @private
  95  * @ingroup DifferenceEngine
  96  */
  97 class _DiffOp_Change extends _DiffOp {
  98         var $type = 'change';
  99
 100         function _DiffOp_Change ($orig, $closing) {
 101                 $this->orig = $orig;
 102                 $this->closing = $closing;
 103         }
 104
 105         function reverse() {
 106                 return new _DiffOp_Change($this->closing, $this->orig);
 107         }
 108 }
 109
 110 /**
 111  * Class used internally by Diff to actually compute the diffs.
 112  *
 113  * The algorithm used here is mostly lifted from the perl module
 114  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 115  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 116  *
 117  * More ideas are taken from:
 118  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 119  *
 120  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 121  * diffutils-2.7, which can be found at:
 122  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 123  *
 124  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 125  * are my own.
 126  *
 127  * Line length limits for robustness added by Tim Starling, 2005-08-31
 128  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
 129  *
 130  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
 131  * @private
 132  * @ingroup DifferenceEngine
 133  */
 134 class _DiffEngine {
 135
 136         const MAX_XREF_LENGTH =  10000;
 137
 138         function diff ($from_lines, $to_lines){
 139                 wfProfileIn( __METHOD__ );
 140
 141                 // Diff and store locally
 142                 $this->diff_local($from_lines, $to_lines);
 143
 144                 // Merge edits when possible
 145                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 146                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 147
 148                 // Compute the edit operations.
 149                 $n_from = sizeof($from_lines);
 150                 $n_to = sizeof($to_lines);
 151
 152                 $edits = array();
 153                 $xi = $yi = 0;
 154                 while ($xi < $n_from || $yi < $n_to) {
 155                         assert($yi < $n_to || $this->xchanged[$xi]);
 156                         assert($xi < $n_from || $this->ychanged[$yi]);
 157
 158                         // Skip matching "snake".
 159                         $copy = array();
 160                         while ( $xi < $n_from && $yi < $n_to
 161                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 162                                 $copy[] = $from_lines[$xi++];
 163                                 ++$yi;
 164                         }
 165                         if ($copy)
 166                         $edits[] = new _DiffOp_Copy($copy);
 167
 168                         // Find deletes & adds.
 169                         $delete = array();
 170                         while ($xi < $n_from && $this->xchanged[$xi])
 171                         $delete[] = $from_lines[$xi++];
 172
 173                         $add = array();
 174                         while ($yi < $n_to && $this->ychanged[$yi])
 175                         $add[] = $to_lines[$yi++];
 176
 177                         if ($delete && $add)
 178                         $edits[] = new _DiffOp_Change($delete, $add);
 179                         elseif ($delete)
 180                         $edits[] = new _DiffOp_Delete($delete);
 181                         elseif ($add)
 182                         $edits[] = new _DiffOp_Add($add);
 183                 }
 184                 wfProfileOut( __METHOD__ );
 185                 return $edits;
 186         }
 187
 188         function diff_local ($from_lines, $to_lines) {
 189                 global $wgExternalDiffEngine;
 190                 wfProfileIn( __METHOD__);
 191
 192                 if($wgExternalDiffEngine == 'wikidiff3'){
 193                         // wikidiff3
 194                         $wikidiff3 = new WikiDiff3();
 195                         $wikidiff3->diff($from_lines, $to_lines);
 196                         $this->xchanged = $wikidiff3->removed;
 197                         $this->ychanged = $wikidiff3->added;
 198                         unset($wikidiff3);
 199                 }else{
 200                         // old diff
 201                         $n_from = sizeof($from_lines);
 202                         $n_to = sizeof($to_lines);
 203                         $this->xchanged = $this->ychanged = array();
 204                         $this->xv = $this->yv = array();
 205                         $this->xind = $this->yind = array();
 206                         unset($this->seq);
 207                         unset($this->in_seq);
 208                         unset($this->lcs);
 209
 210                         // Skip leading common lines.
 211                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 212                                 if ($from_lines[$skip] !== $to_lines[$skip])
 213                                 break;
 214                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 215                         }
 216                         // Skip trailing common lines.
 217                         $xi = $n_from; $yi = $n_to;
 218                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 219                                 if ($from_lines[$xi] !== $to_lines[$yi])
 220                                 break;
 221                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 222                         }
 223
 224                         // Ignore lines which do not exist in both files.
 225                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 226                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 227                         }
 228
 229                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 230                                 $line = $to_lines[$yi];
 231                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 232                                 continue;
 233                                 $yhash[$this->_line_hash($line)] = 1;
 234                                 $this->yv[] = $line;
 235                                 $this->yind[] = $yi;
 236                         }
 237                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 238                                 $line = $from_lines[$xi];
 239                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 240                                 continue;
 241                                 $this->xv[] = $line;
 242                                 $this->xind[] = $xi;
 243                         }
 244
 245                         // Find the LCS.
 246                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 247                 }
 248                 wfProfileOut( __METHOD__ );
 249         }
 250
 251         /**
 252          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 253          */
 254         function _line_hash( $line ) {
 255                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 256                         return md5( $line );
 257                 } else {
 258                         return $line;
 259                 }
 260         }
 261
 262         /* Divide the Largest Common Subsequence (LCS) of the sequences
 263          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 264          * sized segments.
 265          *
 266          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 267          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 268          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 269          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 270          * that (X0, Y0) == (XOFF, YOFF) and
 271          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 272          *
 273          * This function assumes that the first lines of the specified portions
 274          * of the two files do not match, and likewise that the last lines do not
 275          * match.  The caller must trim matching lines from the beginning and end
 276          * of the portions it is going to specify.
 277          */
 278         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 279                 $flip = false;
 280
 281                 if ($xlim - $xoff > $ylim - $yoff) {
 282                         // Things seems faster (I'm not sure I understand why)
 283                         // when the shortest sequence in X.
 284                         $flip = true;
 285                         list ($xoff, $xlim, $yoff, $ylim)
 286                         = array( $yoff, $ylim, $xoff, $xlim);
 287                 }
 288
 289                 if ($flip)
 290                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 291                 $ymatches[$this->xv[$i]][] = $i;
 292                 else
 293                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 294                 $ymatches[$this->yv[$i]][] = $i;
 295
 296                 $this->lcs = 0;
 297                 $this->seq[0]= $yoff - 1;
 298                 $this->in_seq = array();
 299                 $ymids[0] = array();
 300
 301                 $numer = $xlim - $xoff + $nchunks - 1;
 302                 $x = $xoff;
 303                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 304                         if ($chunk > 0)
 305                         for ($i = 0; $i <= $this->lcs; $i++)
 306                         $ymids[$i][$chunk-1] = $this->seq[$i];
 307
 308                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 309                         for ( ; $x < $x1; $x++) {
 310                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 311                                 if (empty($ymatches[$line]))
 312                                 continue;
 313                                 $matches = $ymatches[$line];
 314                                 reset($matches);
 315                                 while (list ($junk, $y) = each($matches))
 316                                 if (empty($this->in_seq[$y])) {
 317                                         $k = $this->_lcs_pos($y);
 318                                         assert($k > 0);
 319                                         $ymids[$k] = $ymids[$k-1];
 320                                         break;
 321                                 }
 322                                 while (list ( /* $junk */, $y) = each($matches)) {
 323                                         if ($y > $this->seq[$k-1]) {
 324                                                 assert($y < $this->seq[$k]);
 325                                                 // Optimization: this is a common case:
 326                                                 //      next match is just replacing previous match.
 327                                                 $this->in_seq[$this->seq[$k]] = false;
 328                                                 $this->seq[$k] = $y;
 329                                                 $this->in_seq[$y] = 1;
 330                                         } else if (empty($this->in_seq[$y])) {
 331                                                 $k = $this->_lcs_pos($y);
 332                                                 assert($k > 0);
 333                                                 $ymids[$k] = $ymids[$k-1];
 334                                         }
 335                                 }
 336                         }
 337                 }
 338
 339                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 340                 $ymid = $ymids[$this->lcs];
 341                 for ($n = 0; $n < $nchunks - 1; $n++) {
 342                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 343                         $y1 = $ymid[$n] + 1;
 344                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 345                 }
 346                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 347
 348                 return array($this->lcs, $seps);
 349         }
 350
 351         function _lcs_pos ($ypos) {
 352                 $end = $this->lcs;
 353                 if ($end == 0 || $ypos > $this->seq[$end]) {
 354                         $this->seq[++$this->lcs] = $ypos;
 355                         $this->in_seq[$ypos] = 1;
 356                         return $this->lcs;
 357                 }
 358
 359                 $beg = 1;
 360                 while ($beg < $end) {
 361                         $mid = (int)(($beg + $end) / 2);
 362                         if ( $ypos > $this->seq[$mid] )
 363                         $beg = $mid + 1;
 364                         else
 365                         $end = $mid;
 366                 }
 367
 368                 assert($ypos != $this->seq[$end]);
 369
 370                 $this->in_seq[$this->seq[$end]] = false;
 371                 $this->seq[$end] = $ypos;
 372                 $this->in_seq[$ypos] = 1;
 373                 return $end;
 374         }
 375
 376         /* Find LCS of two sequences.
 377          *
 378          * The results are recorded in the vectors $this->{x,y}changed[], by
 379          * storing a 1 in the element for each line that is an insertion
 380          * or deletion (ie. is not in the LCS).
 381          *
 382          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 383          *
 384          * Note that XLIM, YLIM are exclusive bounds.
 385          * All line numbers are origin-0 and discarded lines are not counted.
 386          */
 387         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 388                 // Slide down the bottom initial diagonal.
 389                 while ($xoff < $xlim && $yoff < $ylim
 390                 && $this->xv[$xoff] == $this->yv[$yoff]) {
 391                         ++$xoff;
 392                         ++$yoff;
 393                 }
 394
 395                 // Slide up the top initial diagonal.
 396                 while ($xlim > $xoff && $ylim > $yoff
 397                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 398                         --$xlim;
 399                         --$ylim;
 400                 }
 401
 402                 if ($xoff == $xlim || $yoff == $ylim)
 403                 $lcs = 0;
 404                 else {
 405                         // This is ad hoc but seems to work well.
 406                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 407                         //$nchunks = max(2,min(8,(int)$nchunks));
 408                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 409                         list ($lcs, $seps)
 410                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 411                 }
 412
 413                 if ($lcs == 0) {
 414                         // X and Y sequences have no common subsequence:
 415                         // mark all changed.
 416                         while ($yoff < $ylim)
 417                         $this->ychanged[$this->yind[$yoff++]] = 1;
 418                         while ($xoff < $xlim)
 419                         $this->xchanged[$this->xind[$xoff++]] = 1;
 420                 } else {
 421                         // Use the partitions to split this problem into subproblems.
 422                         reset($seps);
 423                         $pt1 = $seps[0];
 424                         while ($pt2 = next($seps)) {
 425                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 426                                 $pt1 = $pt2;
 427                         }
 428                 }
 429         }
 430
 431         /* Adjust inserts/deletes of identical lines to join changes
 432          * as much as possible.
 433          *
 434          * We do something when a run of changed lines include a
 435          * line at one end and has an excluded, identical line at the other.
 436          * We are free to choose which identical line is included.
 437          * `compareseq' usually chooses the one at the beginning,
 438          * but usually it is cleaner to consider the following identical line
 439          * to be the "change".
 440          *
 441          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 442          */
 443         function _shift_boundaries ($lines, &$changed, $other_changed) {
 444                 wfProfileIn( __METHOD__ );
 445                 $i = 0;
 446                 $j = 0;
 447
 448                 assert('sizeof($lines) == sizeof($changed)');
 449                 $len = sizeof($lines);
 450                 $other_len = sizeof($other_changed);
 451
 452                 while (1) {
 453                         /*
 454                          * Scan forwards to find beginning of another run of changes.
 455                          * Also keep track of the corresponding point in the other file.
 456                          *
 457                          * Throughout this code, $i and $j are adjusted together so that
 458                          * the first $i elements of $changed and the first $j elements
 459                          * of $other_changed both contain the same number of zeros
 460                          * (unchanged lines).
 461                          * Furthermore, $j is always kept so that $j == $other_len or
 462                          * $other_changed[$j] == false.
 463                          */
 464                         while ($j < $other_len && $other_changed[$j])
 465                         $j++;
 466
 467                         while ($i < $len && ! $changed[$i]) {
 468                                 assert('$j < $other_len && ! $other_changed[$j]');
 469                                 $i++; $j++;
 470                                 while ($j < $other_len && $other_changed[$j])
 471                                 $j++;
 472                         }
 473
 474                         if ($i == $len)
 475                         break;
 476
 477                         $start = $i;
 478
 479                         // Find the end of this run of changes.
 480                         while (++$i < $len && $changed[$i])
 481                         continue;
 482
 483                         do {
 484                                 /*
 485                                  * Record the length of this run of changes, so that
 486                                  * we can later determine whether the run has grown.
 487                                  */
 488                                 $runlength = $i - $start;
 489
 490                                 /*
 491                                  * Move the changed region back, so long as the
 492                                  * previous unchanged line matches the last changed one.
 493                                  * This merges with previous changed regions.
 494                                  */
 495                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 496                                         $changed[--$start] = 1;
 497                                         $changed[--$i] = false;
 498                                         while ($start > 0 && $changed[$start - 1])
 499                                         $start--;
 500                                         assert('$j > 0');
 501                                         while ($other_changed[--$j])
 502                                         continue;
 503                                         assert('$j >= 0 && !$other_changed[$j]');
 504                                 }
 505
 506                                 /*
 507                                  * Set CORRESPONDING to the end of the changed run, at the last
 508                                  * point where it corresponds to a changed run in the other file.
 509                                  * CORRESPONDING == LEN means no such point has been found.
 510                                  */
 511                                 $corresponding = $j < $other_len ? $i : $len;
 512
 513                                 /*
 514                                  * Move the changed region forward, so long as the
 515                                  * first changed line matches the following unchanged one.
 516                                  * This merges with following changed regions.
 517                                  * Do this second, so that if there are no merges,
 518                                  * the changed region is moved forward as far as possible.
 519                                  */
 520                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 521                                         $changed[$start++] = false;
 522                                         $changed[$i++] = 1;
 523                                         while ($i < $len && $changed[$i])
 524                                         $i++;
 525
 526                                         assert('$j < $other_len && ! $other_changed[$j]');
 527                                         $j++;
 528                                         if ($j < $other_len && $other_changed[$j]) {
 529                                                 $corresponding = $i;
 530                                                 while ($j < $other_len && $other_changed[$j])
 531                                                 $j++;
 532                                         }
 533                                 }
 534                         } while ($runlength != $i - $start);
 535
 536                         /*
 537                          * If possible, move the fully-merged run of changes
 538                          * back to a corresponding run in the other file.
 539                          */
 540                         while ($corresponding < $i) {
 541                                 $changed[--$start] = 1;
 542                                 $changed[--$i] = 0;
 543                                 assert('$j > 0');
 544                                 while ($other_changed[--$j])
 545                                 continue;
 546                                 assert('$j >= 0 && !$other_changed[$j]');
 547                         }
 548                 }
 549                 wfProfileOut( __METHOD__ );
 550         }
 551 }
 552
 553 /**
 554  * Class representing a 'diff' between two sequences of strings.
 555  * @todo document
 556  * @private
 557  * @ingroup DifferenceEngine
 558  */
 559 class Diff
 560 {
 561         var $edits;
 562
 563         /**
 564          * Constructor.
 565          * Computes diff between sequences of strings.
 566          *
 567          * @param $from_lines array An array of strings.
 568          *                (Typically these are lines from a file.)
 569          * @param $to_lines array An array of strings.
 570          */
 571         function Diff($from_lines, $to_lines) {
 572                 $eng = new _DiffEngine;
 573                 $this->edits = $eng->diff($from_lines, $to_lines);
 574                 //$this->_check($from_lines, $to_lines);
 575         }
 576
 577         /**
 578          * Compute reversed Diff.
 579          *
 580          * SYNOPSIS:
 581          *
 582          *      $diff = new Diff($lines1, $lines2);
 583          *      $rev = $diff->reverse();
 584          * @return object A Diff object representing the inverse of the
 585          *                                original diff.
 586          */
 587         function reverse () {
 588                 $rev = $this;
 589                 $rev->edits = array();
 590                 foreach ($this->edits as $edit) {
 591                         $rev->edits[] = $edit->reverse();
 592                 }
 593                 return $rev;
 594         }
 595
 596         /**
 597          * Check for empty diff.
 598          *
 599          * @return bool True iff two sequences were identical.
 600          */
 601         function isEmpty () {
 602                 foreach ($this->edits as $edit) {
 603                         if ($edit->type != 'copy')
 604                         return false;
 605                 }
 606                 return true;
 607         }
 608
 609         /**
 610          * Compute the length of the Longest Common Subsequence (LCS).
 611          *
 612          * This is mostly for diagnostic purposed.
 613          *
 614          * @return int The length of the LCS.
 615          */
 616         function lcs () {
 617                 $lcs = 0;
 618                 foreach ($this->edits as $edit) {
 619                         if ($edit->type == 'copy')
 620                         $lcs += sizeof($edit->orig);
 621                 }
 622                 return $lcs;
 623         }
 624
 625         /**
 626          * Get the original set of lines.
 627          *
 628          * This reconstructs the $from_lines parameter passed to the
 629          * constructor.
 630          *
 631          * @return array The original sequence of strings.
 632          */
 633         function orig() {
 634                 $lines = array();
 635
 636                 foreach ($this->edits as $edit) {
 637                         if ($edit->orig)
 638                         array_splice($lines, sizeof($lines), 0, $edit->orig);
 639                 }
 640                 return $lines;
 641         }
 642
 643         /**
 644          * Get the closing set of lines.
 645          *
 646          * This reconstructs the $to_lines parameter passed to the
 647          * constructor.
 648          *
 649          * @return array The sequence of strings.
 650          */
 651         function closing() {
 652                 $lines = array();
 653
 654                 foreach ($this->edits as $edit) {
 655                         if ($edit->closing)
 656                         array_splice($lines, sizeof($lines), 0, $edit->closing);
 657                 }
 658                 return $lines;
 659         }
 660
 661         /**
 662          * Check a Diff for validity.
 663          *
 664          * This is here only for debugging purposes.
 665          */
 666         function _check ($from_lines, $to_lines) {
 667                 wfProfileIn( __METHOD__ );
 668                 if (serialize($from_lines) != serialize($this->orig()))
 669                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 670                 if (serialize($to_lines) != serialize($this->closing()))
 671                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 672
 673                 $rev = $this->reverse();
 674                 if (serialize($to_lines) != serialize($rev->orig()))
 675                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
 676                 if (serialize($from_lines) != serialize($rev->closing()))
 677                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 678
 679
 680                 $prevtype = 'none';
 681                 foreach ($this->edits as $edit) {
 682                         if ( $prevtype == $edit->type )
 683                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 684                         $prevtype = $edit->type;
 685                 }
 686
 687                 $lcs = $this->lcs();
 688                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 689                 wfProfileOut( __METHOD__ );
 690         }
 691 }
 692
 693 /**
 694  * @todo document, bad name.
 695  * @private
 696  * @ingroup DifferenceEngine
 697  */
 698 class MappedDiff extends Diff
 699 {
 700         /**
 701          * Constructor.
 702          *
 703          * Computes diff between sequences of strings.
 704          *
 705          * This can be used to compute things like
 706          * case-insensitve diffs, or diffs which ignore
 707          * changes in white-space.
 708          *
 709          * @param $from_lines array An array of strings.
 710          *      (Typically these are lines from a file.)
 711          *
 712          * @param $to_lines array An array of strings.
 713          *
 714          * @param $mapped_from_lines array This array should
 715          *      have the same size number of elements as $from_lines.
 716          *      The elements in $mapped_from_lines and
 717          *      $mapped_to_lines are what is actually compared
 718          *      when computing the diff.
 719          *
 720          * @param $mapped_to_lines array This array should
 721          *      have the same number of elements as $to_lines.
 722          */
 723         function MappedDiff($from_lines, $to_lines,
 724         $mapped_from_lines, $mapped_to_lines) {
 725                 wfProfileIn( __METHOD__ );
 726
 727                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 728                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 729
 730                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 731
 732                 $xi = $yi = 0;
 733                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 734                         $orig = &$this->edits[$i]->orig;
 735                         if (is_array($orig)) {
 736                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 737                                 $xi += sizeof($orig);
 738                         }
 739
 740                         $closing = &$this->edits[$i]->closing;
 741                         if (is_array($closing)) {
 742                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 743                                 $yi += sizeof($closing);
 744                         }
 745                 }
 746                 wfProfileOut( __METHOD__ );
 747         }
 748 }
 749
 750 /**
 751  * A class to format Diffs
 752  *
 753  * This class formats the diff in classic diff format.
 754  * It is intended that this class be customized via inheritance,
 755  * to obtain fancier outputs.
 756  * @todo document
 757  * @private
 758  * @ingroup DifferenceEngine
 759  */
 760 class DiffFormatter {
 761         /**
 762          * Number of leading context "lines" to preserve.
 763          *
 764          * This should be left at zero for this class, but subclasses
 765          * may want to set this to other values.
 766          */
 767         var $leading_context_lines = 0;
 768
 769         /**
 770          * Number of trailing context "lines" to preserve.
 771          *
 772          * This should be left at zero for this class, but subclasses
 773          * may want to set this to other values.
 774          */
 775         var $trailing_context_lines = 0;
 776
 777         /**
 778          * Format a diff.
 779          *
 780          * @param $diff object A Diff object.
 781          * @return string The formatted output.
 782          */
 783         function format($diff) {
 784                 wfProfileIn( __METHOD__ );
 785
 786                 $xi = $yi = 1;
 787                 $block = false;
 788                 $context = array();
 789
 790                 $nlead = $this->leading_context_lines;
 791                 $ntrail = $this->trailing_context_lines;
 792
 793                 $this->_start_diff();
 794
 795                 foreach ($diff->edits as $edit) {
 796                         if ($edit->type == 'copy') {
 797                                 if (is_array($block)) {
 798                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 799                                                 $block[] = $edit;
 800                                         }
 801                                         else{
 802                                                 if ($ntrail) {
 803                                                         $context = array_slice($edit->orig, 0, $ntrail);
 804                                                         $block[] = new _DiffOp_Copy($context);
 805                                                 }
 806                                                 $this->_block($x0, $ntrail + $xi - $x0,
 807                                                 $y0, $ntrail + $yi - $y0,
 808                                                 $block);
 809                                                 $block = false;
 810                                         }
 811                                 }
 812                                 $context = $edit->orig;
 813                         }
 814                         else {
 815                                 if (! is_array($block)) {
 816                                         $context = array_slice($context, sizeof($context) - $nlead);
 817                                         $x0 = $xi - sizeof($context);
 818                                         $y0 = $yi - sizeof($context);
 819                                         $block = array();
 820                                         if ($context)
 821                                         $block[] = new _DiffOp_Copy($context);
 822                                 }
 823                                 $block[] = $edit;
 824                         }
 825
 826                         if ($edit->orig)
 827                         $xi += sizeof($edit->orig);
 828                         if ($edit->closing)
 829                         $yi += sizeof($edit->closing);
 830                 }
 831
 832                 if (is_array($block))
 833                 $this->_block($x0, $xi - $x0,
 834                 $y0, $yi - $y0,
 835                 $block);
 836
 837                 $end = $this->_end_diff();
 838                 wfProfileOut( __METHOD__ );
 839                 return $end;
 840         }
 841
 842         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 843                 wfProfileIn( __METHOD__ );
 844                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 845                 foreach ($edits as $edit) {
 846                         if ($edit->type == 'copy')
 847                         $this->_context($edit->orig);
 848                         elseif ($edit->type == 'add')
 849                         $this->_added($edit->closing);
 850                         elseif ($edit->type == 'delete')
 851                         $this->_deleted($edit->orig);
 852                         elseif ($edit->type == 'change')
 853                         $this->_changed($edit->orig, $edit->closing);
 854                         else
 855                         trigger_error('Unknown edit type', E_USER_ERROR);
 856                 }
 857                 $this->_end_block();
 858                 wfProfileOut( __METHOD__ );
 859         }
 860
 861         function _start_diff() {
 862                 ob_start();
 863         }
 864
 865         function _end_diff() {
 866                 $val = ob_get_contents();
 867                 ob_end_clean();
 868                 return $val;
 869         }
 870
 871         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 872                 if ($xlen > 1)
 873                 $xbeg .= "," . ($xbeg + $xlen - 1);
 874                 if ($ylen > 1)
 875                 $ybeg .= "," . ($ybeg + $ylen - 1);
 876
 877                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 878         }
 879
 880         function _start_block($header) {
 881                 echo $header . "\n";
 882         }
 883
 884         function _end_block() {
 885         }
 886
 887         function _lines($lines, $prefix = ' ') {
 888                 foreach ($lines as $line)
 889                 echo "$prefix $line\n";
 890         }
 891
 892         function _context($lines) {
 893                 $this->_lines($lines);
 894         }
 895
 896         function _added($lines) {
 897                 $this->_lines($lines, '>');
 898         }
 899         function _deleted($lines) {
 900                 $this->_lines($lines, '<');
 901         }
 902
 903         function _changed($orig, $closing) {
 904                 $this->_deleted($orig);
 905                 echo "---\n";
 906                 $this->_added($closing);
 907         }
 908 }
 909
 910 /**
 911  * A formatter that outputs unified diffs
 912  * @ingroup DifferenceEngine
 913  */
 914
 915 class UnifiedDiffFormatter extends DiffFormatter {
 916         var $leading_context_lines = 2;
 917         var $trailing_context_lines = 2;
 918
 919         function _added($lines) {
 920                 $this->_lines($lines, '+');
 921         }
 922         function _deleted($lines) {
 923                 $this->_lines($lines, '-');
 924         }
 925         function _changed($orig, $closing) {
 926                 $this->_deleted($orig);
 927                 $this->_added($closing);
 928         }
 929         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 930                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
 931         }
 932 }
 933
 934 /**
 935  * A pseudo-formatter that just passes along the Diff::$edits array
 936  * @ingroup DifferenceEngine
 937  */
 938 class ArrayDiffFormatter extends DiffFormatter {
 939         function format($diff) {
 940                 $oldline = 1;
 941                 $newline = 1;
 942                 $retval = array();
 943                 foreach($diff->edits as $edit)
 944                 switch($edit->type) {
 945                         case 'add':
 946                                 foreach($edit->closing as $l) {
 947                                         $retval[] = array(
 948                                                         'action' => 'add',
 949                                                         'new'=> $l,
 950                                                         'newline' => $newline++
 951                                         );
 952                                 }
 953                                 break;
 954                         case 'delete':
 955                                 foreach($edit->orig as $l) {
 956                                         $retval[] = array(
 957                                                         'action' => 'delete',
 958                                                         'old' => $l,
 959                                                         'oldline' => $oldline++,
 960                                         );
 961                                 }
 962                                 break;
 963                         case 'change':
 964                                 foreach($edit->orig as $i => $l) {
 965                                         $retval[] = array(
 966                                                         'action' => 'change',
 967                                                         'old' => $l,
 968                                                         'new' => @$edit->closing[$i],
 969                                                         'oldline' => $oldline++,
 970                                                         'newline' => $newline++,
 971                                         );
 972                                 }
 973                                 break;
 974                         case 'copy':
 975                                 $oldline += count($edit->orig);
 976                                 $newline += count($edit->orig);
 977                 }
 978                 return $retval;
 979         }
 980 }
 981
 982 /**
 983  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
 984  *
 985  */
 986
 987 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
 988
 989 /**
 990  * @todo document
 991  * @private
 992  * @ingroup DifferenceEngine
 993  */
 994 class _HWLDF_WordAccumulator {
 995         function _HWLDF_WordAccumulator () {
 996                 $this->_lines = array();
 997                 $this->_line = '';
 998                 $this->_group = '';
 999                 $this->_tag = '';
1000         }
1001
1002         function _flushGroup ($new_tag) {
1003                 if ($this->_group !== '') {
1004                         if ($this->_tag == 'ins')
1005                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1006                         htmlspecialchars ( $this->_group ) . '</ins>';
1007                         elseif ($this->_tag == 'del')
1008                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1009                         htmlspecialchars ( $this->_group ) . '</del>';
1010                         else
1011                         $this->_line .= htmlspecialchars ( $this->_group );
1012                 }
1013                 $this->_group = '';
1014                 $this->_tag = $new_tag;
1015         }
1016
1017         function _flushLine ($new_tag) {
1018                 $this->_flushGroup($new_tag);
1019                 if ($this->_line != '')
1020                 array_push ( $this->_lines, $this->_line );
1021                 else
1022                 # make empty lines visible by inserting an NBSP
1023                 array_push ( $this->_lines, NBSP );
1024                 $this->_line = '';
1025         }
1026
1027         function addWords ($words, $tag = '') {
1028                 if ($tag != $this->_tag)
1029                 $this->_flushGroup($tag);
1030
1031                 foreach ($words as $word) {
1032                         // new-line should only come as first char of word.
1033                         if ($word == '')
1034                         continue;
1035                         if ($word[0] == "\n") {
1036                                 $this->_flushLine($tag);
1037                                 $word = substr($word, 1);
1038                         }
1039                         assert(!strstr($word, "\n"));
1040                         $this->_group .= $word;
1041                 }
1042         }
1043
1044         function getLines() {
1045                 $this->_flushLine('~done');
1046                 return $this->_lines;
1047         }
1048 }
1049
1050 /**
1051  * @todo document
1052  * @private
1053  * @ingroup DifferenceEngine
1054  */
1055 class WordLevelDiff extends MappedDiff {
1056         const MAX_LINE_LENGTH = 10000;
1057
1058         function WordLevelDiff ($orig_lines, $closing_lines) {
1059                 wfProfileIn( __METHOD__ );
1060
1061                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1062                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1063
1064                 $this->MappedDiff($orig_words, $closing_words,
1065                 $orig_stripped, $closing_stripped);
1066                 wfProfileOut( __METHOD__ );
1067         }
1068
1069         function _split($lines) {
1070                 wfProfileIn( __METHOD__ );
1071
1072                 $words = array();
1073                 $stripped = array();
1074                 $first = true;
1075                 foreach ( $lines as $line ) {
1076                         # If the line is too long, just pretend the entire line is one big word
1077                         # This prevents resource exhaustion problems
1078                         if ( $first ) {
1079                                 $first = false;
1080                         } else {
1081                                 $words[] = "\n";
1082                                 $stripped[] = "\n";
1083                         }
1084                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1085                                 $words[] = $line;
1086                                 $stripped[] = $line;
1087                         } else {
1088                                 $m = array();
1089                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1090                                 $line, $m))
1091                                 {
1092                                         $words = array_merge( $words, $m[0] );
1093                                         $stripped = array_merge( $stripped, $m[1] );
1094                                 }
1095                         }
1096                 }
1097                 wfProfileOut( __METHOD__ );
1098                 return array($words, $stripped);
1099         }
1100
1101         function orig () {
1102                 wfProfileIn( __METHOD__ );
1103                 $orig = new _HWLDF_WordAccumulator;
1104
1105                 foreach ($this->edits as $edit) {
1106                         if ($edit->type == 'copy')
1107                         $orig->addWords($edit->orig);
1108                         elseif ($edit->orig)
1109                         $orig->addWords($edit->orig, 'del');
1110                 }
1111                 $lines = $orig->getLines();
1112                 wfProfileOut( __METHOD__ );
1113                 return $lines;
1114         }
1115
1116         function closing () {
1117                 wfProfileIn( __METHOD__ );
1118                 $closing = new _HWLDF_WordAccumulator;
1119
1120                 foreach ($this->edits as $edit) {
1121                         if ($edit->type == 'copy')
1122                         $closing->addWords($edit->closing);
1123                         elseif ($edit->closing)
1124                         $closing->addWords($edit->closing, 'ins');
1125                 }
1126                 $lines = $closing->getLines();
1127                 wfProfileOut( __METHOD__ );
1128                 return $lines;
1129         }
1130 }
1131
1132 /**
1133  * Wikipedia Table style diff formatter.
1134  * @todo document
1135  * @private
1136  * @ingroup DifferenceEngine
1137  */
1138 class TableDiffFormatter extends DiffFormatter {
1139         function TableDiffFormatter() {
1140                 $this->leading_context_lines = 2;
1141                 $this->trailing_context_lines = 2;
1142         }
1143
1144         public static function escapeWhiteSpace( $msg ) {
1145                 $msg = preg_replace( '/^ /m', '&#160; ', $msg );
1146                 $msg = preg_replace( '/ $/m', ' &#160;', $msg );
1147                 $msg = preg_replace( '/  /', '&#160; ', $msg );
1148                 return $msg;
1149         }
1150
1151         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1152                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1153                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1154                 return $r;
1155         }
1156
1157         function _start_block( $header ) {
1158                 echo $header;
1159         }
1160
1161         function _end_block() {
1162         }
1163
1164         function _lines( $lines, $prefix=' ', $color='white' ) {
1165         }
1166
1167         # HTML-escape parameter before calling this
1168         function addedLine( $line ) {
1169                 return $this->wrapLine( '+', 'diff-addedline', $line );
1170         }
1171
1172         # HTML-escape parameter before calling this
1173         function deletedLine( $line ) {
1174                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1175         }
1176
1177         # HTML-escape parameter before calling this
1178         function contextLine( $line ) {
1179                 return $this->wrapLine( ' ', 'diff-context', $line );
1180         }
1181
1182         private function wrapLine( $marker, $class, $line ) {
1183                 if( $line !== '' ) {
1184                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1185                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1186                 }
1187                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1188         }
1189
1190         function emptyLine() {
1191                 return '<td colspan="2">&#160;</td>';
1192         }
1193
1194         function _added( $lines ) {
1195                 foreach ($lines as $line) {
1196                         echo '<tr>' . $this->emptyLine() .
1197                         $this->addedLine( '<ins class="diffchange">' .
1198                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1199                 }
1200         }
1201
1202         function _deleted($lines) {
1203                 foreach ($lines as $line) {
1204                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1205                         htmlspecialchars ( $line ) . '</del>' ) .
1206                         $this->emptyLine() . "</tr>\n";
1207                 }
1208         }
1209
1210         function _context( $lines ) {
1211                 foreach ($lines as $line) {
1212                         echo '<tr>' .
1213                         $this->contextLine( htmlspecialchars ( $line ) ) .
1214                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1215                 }
1216         }
1217
1218         function _changed( $orig, $closing ) {
1219                 wfProfileIn( __METHOD__ );
1220
1221                 $diff = new WordLevelDiff( $orig, $closing );
1222                 $del = $diff->orig();
1223                 $add = $diff->closing();
1224
1225                 # Notice that WordLevelDiff returns HTML-escaped output.
1226                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1227
1228                 while ( $line = array_shift( $del ) ) {
1229                         $aline = array_shift( $add );
1230                         echo '<tr>' . $this->deletedLine( $line ) .
1231                         $this->addedLine( $aline ) . "</tr>\n";
1232                 }
1233                 foreach ($add as $line) {       # If any leftovers
1234                         echo '<tr>' . $this->emptyLine() .
1235                         $this->addedLine( $line ) . "</tr>\n";
1236                 }
1237                 wfProfileOut( __METHOD__ );
1238         }
1239 }