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 $type = 'copy';
  43
  44         function __construct ($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 $type = 'delete';
  63
  64         function __construct ($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 $type = 'add';
  81
  82         function __construct ($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 $type = 'change';
  99
 100         function __construct ($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                                 }
 323                                 while (list ( /* $junk */, $y) = each($matches)) {
 324                                         if ($y > $this->seq[$k-1]) {
 325                                                 assert($y < $this->seq[$k]);
 326                                                 // Optimization: this is a common case:
 327                                                 //      next match is just replacing previous match.
 328                                                 $this->in_seq[$this->seq[$k]] = false;
 329                                                 $this->seq[$k] = $y;
 330                                                 $this->in_seq[$y] = 1;
 331                                         } else if (empty($this->in_seq[$y])) {
 332                                                 $k = $this->_lcs_pos($y);
 333                                                 assert($k > 0);
 334                                                 $ymids[$k] = $ymids[$k-1];
 335                                         }
 336                                 }
 337                         }
 338                 }
 339
 340                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 341                 $ymid = $ymids[$this->lcs];
 342                 for ($n = 0; $n < $nchunks - 1; $n++) {
 343                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 344                         $y1 = $ymid[$n] + 1;
 345                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 346                 }
 347                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 348
 349                 return array($this->lcs, $seps);
 350         }
 351
 352         function _lcs_pos ($ypos) {
 353                 $end = $this->lcs;
 354                 if ($end == 0 || $ypos > $this->seq[$end]) {
 355                         $this->seq[++$this->lcs] = $ypos;
 356                         $this->in_seq[$ypos] = 1;
 357                         return $this->lcs;
 358                 }
 359
 360                 $beg = 1;
 361                 while ($beg < $end) {
 362                         $mid = (int)(($beg + $end) / 2);
 363                         if ( $ypos > $this->seq[$mid] )
 364                         $beg = $mid + 1;
 365                         else
 366                         $end = $mid;
 367                 }
 368
 369                 assert($ypos != $this->seq[$end]);
 370
 371                 $this->in_seq[$this->seq[$end]] = false;
 372                 $this->seq[$end] = $ypos;
 373                 $this->in_seq[$ypos] = 1;
 374                 return $end;
 375         }
 376
 377         /* Find LCS of two sequences.
 378          *
 379          * The results are recorded in the vectors $this->{x,y}changed[], by
 380          * storing a 1 in the element for each line that is an insertion
 381          * or deletion (ie. is not in the LCS).
 382          *
 383          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 384          *
 385          * Note that XLIM, YLIM are exclusive bounds.
 386          * All line numbers are origin-0 and discarded lines are not counted.
 387          */
 388         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 389                 // Slide down the bottom initial diagonal.
 390                 while ($xoff < $xlim && $yoff < $ylim
 391                 && $this->xv[$xoff] == $this->yv[$yoff]) {
 392                         ++$xoff;
 393                         ++$yoff;
 394                 }
 395
 396                 // Slide up the top initial diagonal.
 397                 while ($xlim > $xoff && $ylim > $yoff
 398                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 399                         --$xlim;
 400                         --$ylim;
 401                 }
 402
 403                 if ($xoff == $xlim || $yoff == $ylim)
 404                 $lcs = 0;
 405                 else {
 406                         // This is ad hoc but seems to work well.
 407                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 408                         //$nchunks = max(2,min(8,(int)$nchunks));
 409                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 410                         list ($lcs, $seps)
 411                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 412                 }
 413
 414                 if ($lcs == 0) {
 415                         // X and Y sequences have no common subsequence:
 416                         // mark all changed.
 417                         while ($yoff < $ylim)
 418                         $this->ychanged[$this->yind[$yoff++]] = 1;
 419                         while ($xoff < $xlim)
 420                         $this->xchanged[$this->xind[$xoff++]] = 1;
 421                 } else {
 422                         // Use the partitions to split this problem into subproblems.
 423                         reset($seps);
 424                         $pt1 = $seps[0];
 425                         while ($pt2 = next($seps)) {
 426                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 427                                 $pt1 = $pt2;
 428                         }
 429                 }
 430         }
 431
 432         /* Adjust inserts/deletes of identical lines to join changes
 433          * as much as possible.
 434          *
 435          * We do something when a run of changed lines include a
 436          * line at one end and has an excluded, identical line at the other.
 437          * We are free to choose which identical line is included.
 438          * `compareseq' usually chooses the one at the beginning,
 439          * but usually it is cleaner to consider the following identical line
 440          * to be the "change".
 441          *
 442          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 443          */
 444         function _shift_boundaries ($lines, &$changed, $other_changed) {
 445                 wfProfileIn( __METHOD__ );
 446                 $i = 0;
 447                 $j = 0;
 448
 449                 assert('sizeof($lines) == sizeof($changed)');
 450                 $len = sizeof($lines);
 451                 $other_len = sizeof($other_changed);
 452
 453                 while (1) {
 454                         /*
 455                          * Scan forwards to find beginning of another run of changes.
 456                          * Also keep track of the corresponding point in the other file.
 457                          *
 458                          * Throughout this code, $i and $j are adjusted together so that
 459                          * the first $i elements of $changed and the first $j elements
 460                          * of $other_changed both contain the same number of zeros
 461                          * (unchanged lines).
 462                          * Furthermore, $j is always kept so that $j == $other_len or
 463                          * $other_changed[$j] == false.
 464                          */
 465                         while ($j < $other_len && $other_changed[$j])
 466                         $j++;
 467
 468                         while ($i < $len && ! $changed[$i]) {
 469                                 assert('$j < $other_len && ! $other_changed[$j]');
 470                                 $i++; $j++;
 471                                 while ($j < $other_len && $other_changed[$j])
 472                                 $j++;
 473                         }
 474
 475                         if ($i == $len)
 476                         break;
 477
 478                         $start = $i;
 479
 480                         // Find the end of this run of changes.
 481                         while (++$i < $len && $changed[$i])
 482                         continue;
 483
 484                         do {
 485                                 /*
 486                                  * Record the length of this run of changes, so that
 487                                  * we can later determine whether the run has grown.
 488                                  */
 489                                 $runlength = $i - $start;
 490
 491                                 /*
 492                                  * Move the changed region back, so long as the
 493                                  * previous unchanged line matches the last changed one.
 494                                  * This merges with previous changed regions.
 495                                  */
 496                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 497                                         $changed[--$start] = 1;
 498                                         $changed[--$i] = false;
 499                                         while ($start > 0 && $changed[$start - 1])
 500                                         $start--;
 501                                         assert('$j > 0');
 502                                         while ($other_changed[--$j])
 503                                         continue;
 504                                         assert('$j >= 0 && !$other_changed[$j]');
 505                                 }
 506
 507                                 /*
 508                                  * Set CORRESPONDING to the end of the changed run, at the last
 509                                  * point where it corresponds to a changed run in the other file.
 510                                  * CORRESPONDING == LEN means no such point has been found.
 511                                  */
 512                                 $corresponding = $j < $other_len ? $i : $len;
 513
 514                                 /*
 515                                  * Move the changed region forward, so long as the
 516                                  * first changed line matches the following unchanged one.
 517                                  * This merges with following changed regions.
 518                                  * Do this second, so that if there are no merges,
 519                                  * the changed region is moved forward as far as possible.
 520                                  */
 521                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 522                                         $changed[$start++] = false;
 523                                         $changed[$i++] = 1;
 524                                         while ($i < $len && $changed[$i])
 525                                         $i++;
 526
 527                                         assert('$j < $other_len && ! $other_changed[$j]');
 528                                         $j++;
 529                                         if ($j < $other_len && $other_changed[$j]) {
 530                                                 $corresponding = $i;
 531                                                 while ($j < $other_len && $other_changed[$j])
 532                                                 $j++;
 533                                         }
 534                                 }
 535                         } while ($runlength != $i - $start);
 536
 537                         /*
 538                          * If possible, move the fully-merged run of changes
 539                          * back to a corresponding run in the other file.
 540                          */
 541                         while ($corresponding < $i) {
 542                                 $changed[--$start] = 1;
 543                                 $changed[--$i] = 0;
 544                                 assert('$j > 0');
 545                                 while ($other_changed[--$j])
 546                                 continue;
 547                                 assert('$j >= 0 && !$other_changed[$j]');
 548                         }
 549                 }
 550                 wfProfileOut( __METHOD__ );
 551         }
 552 }
 553
 554 /**
 555  * Class representing a 'diff' between two sequences of strings.
 556  * @todo document
 557  * @private
 558  * @ingroup DifferenceEngine
 559  */
 560 class Diff
 561 {
 562         var $edits;
 563
 564         /**
 565          * Constructor.
 566          * Computes diff between sequences of strings.
 567          *
 568          * @param $from_lines array An array of strings.
 569          *                (Typically these are lines from a file.)
 570          * @param $to_lines array An array of strings.
 571          */
 572         function __construct($from_lines, $to_lines) {
 573                 $eng = new _DiffEngine;
 574                 $this->edits = $eng->diff($from_lines, $to_lines);
 575                 //$this->_check($from_lines, $to_lines);
 576         }
 577
 578         /**
 579          * Compute reversed Diff.
 580          *
 581          * SYNOPSIS:
 582          *
 583          *      $diff = new Diff($lines1, $lines2);
 584          *      $rev = $diff->reverse();
 585          * @return object A Diff object representing the inverse of the
 586          *                                original diff.
 587          */
 588         function reverse () {
 589                 $rev = $this;
 590                 $rev->edits = array();
 591                 foreach ($this->edits as $edit) {
 592                         $rev->edits[] = $edit->reverse();
 593                 }
 594                 return $rev;
 595         }
 596
 597         /**
 598          * Check for empty diff.
 599          *
 600          * @return bool True iff two sequences were identical.
 601          */
 602         function isEmpty () {
 603                 foreach ($this->edits as $edit) {
 604                         if ($edit->type != 'copy')
 605                         return false;
 606                 }
 607                 return true;
 608         }
 609
 610         /**
 611          * Compute the length of the Longest Common Subsequence (LCS).
 612          *
 613          * This is mostly for diagnostic purposed.
 614          *
 615          * @return int The length of the LCS.
 616          */
 617         function lcs () {
 618                 $lcs = 0;
 619                 foreach ($this->edits as $edit) {
 620                         if ($edit->type == 'copy')
 621                         $lcs += sizeof($edit->orig);
 622                 }
 623                 return $lcs;
 624         }
 625
 626         /**
 627          * Get the original set of lines.
 628          *
 629          * This reconstructs the $from_lines parameter passed to the
 630          * constructor.
 631          *
 632          * @return array The original sequence of strings.
 633          */
 634         function orig() {
 635                 $lines = array();
 636
 637                 foreach ($this->edits as $edit) {
 638                         if ($edit->orig)
 639                         array_splice($lines, sizeof($lines), 0, $edit->orig);
 640                 }
 641                 return $lines;
 642         }
 643
 644         /**
 645          * Get the closing set of lines.
 646          *
 647          * This reconstructs the $to_lines parameter passed to the
 648          * constructor.
 649          *
 650          * @return array The sequence of strings.
 651          */
 652         function closing() {
 653                 $lines = array();
 654
 655                 foreach ($this->edits as $edit) {
 656                         if ($edit->closing)
 657                         array_splice($lines, sizeof($lines), 0, $edit->closing);
 658                 }
 659                 return $lines;
 660         }
 661
 662         /**
 663          * Check a Diff for validity.
 664          *
 665          * This is here only for debugging purposes.
 666          */
 667         function _check ($from_lines, $to_lines) {
 668                 wfProfileIn( __METHOD__ );
 669                 if (serialize($from_lines) != serialize($this->orig()))
 670                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 671                 if (serialize($to_lines) != serialize($this->closing()))
 672                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 673
 674                 $rev = $this->reverse();
 675                 if (serialize($to_lines) != serialize($rev->orig()))
 676                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
 677                 if (serialize($from_lines) != serialize($rev->closing()))
 678                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 679
 680
 681                 $prevtype = 'none';
 682                 foreach ($this->edits as $edit) {
 683                         if ( $prevtype == $edit->type )
 684                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 685                         $prevtype = $edit->type;
 686                 }
 687
 688                 $lcs = $this->lcs();
 689                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 690                 wfProfileOut( __METHOD__ );
 691         }
 692 }
 693
 694 /**
 695  * @todo document, bad name.
 696  * @private
 697  * @ingroup DifferenceEngine
 698  */
 699 class MappedDiff extends Diff
 700 {
 701         /**
 702          * Constructor.
 703          *
 704          * Computes diff between sequences of strings.
 705          *
 706          * This can be used to compute things like
 707          * case-insensitve diffs, or diffs which ignore
 708          * changes in white-space.
 709          *
 710          * @param $from_lines array An array of strings.
 711          *      (Typically these are lines from a file.)
 712          *
 713          * @param $to_lines array An array of strings.
 714          *
 715          * @param $mapped_from_lines array This array should
 716          *      have the same size number of elements as $from_lines.
 717          *      The elements in $mapped_from_lines and
 718          *      $mapped_to_lines are what is actually compared
 719          *      when computing the diff.
 720          *
 721          * @param $mapped_to_lines array This array should
 722          *      have the same number of elements as $to_lines.
 723          */
 724         function __construct($from_lines, $to_lines,
 725         $mapped_from_lines, $mapped_to_lines) {
 726                 wfProfileIn( __METHOD__ );
 727
 728                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 729                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 730
 731                 parent::__construct($mapped_from_lines, $mapped_to_lines);
 732
 733                 $xi = $yi = 0;
 734                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 735                         $orig = &$this->edits[$i]->orig;
 736                         if (is_array($orig)) {
 737                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 738                                 $xi += sizeof($orig);
 739                         }
 740
 741                         $closing = &$this->edits[$i]->closing;
 742                         if (is_array($closing)) {
 743                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 744                                 $yi += sizeof($closing);
 745                         }
 746                 }
 747                 wfProfileOut( __METHOD__ );
 748         }
 749 }
 750
 751 /**
 752  * A class to format Diffs
 753  *
 754  * This class formats the diff in classic diff format.
 755  * It is intended that this class be customized via inheritance,
 756  * to obtain fancier outputs.
 757  * @todo document
 758  * @private
 759  * @ingroup DifferenceEngine
 760  */
 761 class DiffFormatter {
 762         /**
 763          * Number of leading context "lines" to preserve.
 764          *
 765          * This should be left at zero for this class, but subclasses
 766          * may want to set this to other values.
 767          */
 768         var $leading_context_lines = 0;
 769
 770         /**
 771          * Number of trailing context "lines" to preserve.
 772          *
 773          * This should be left at zero for this class, but subclasses
 774          * may want to set this to other values.
 775          */
 776         var $trailing_context_lines = 0;
 777
 778         /**
 779          * Format a diff.
 780          *
 781          * @param $diff object A Diff object.
 782          * @return string The formatted output.
 783          */
 784         function format($diff) {
 785                 wfProfileIn( __METHOD__ );
 786
 787                 $xi = $yi = 1;
 788                 $block = false;
 789                 $context = array();
 790
 791                 $nlead = $this->leading_context_lines;
 792                 $ntrail = $this->trailing_context_lines;
 793
 794                 $this->_start_diff();
 795
 796                 foreach ($diff->edits as $edit) {
 797                         if ($edit->type == 'copy') {
 798                                 if (is_array($block)) {
 799                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 800                                                 $block[] = $edit;
 801                                         }
 802                                         else{
 803                                                 if ($ntrail) {
 804                                                         $context = array_slice($edit->orig, 0, $ntrail);
 805                                                         $block[] = new _DiffOp_Copy($context);
 806                                                 }
 807                                                 $this->_block($x0, $ntrail + $xi - $x0,
 808                                                 $y0, $ntrail + $yi - $y0,
 809                                                 $block);
 810                                                 $block = false;
 811                                         }
 812                                 }
 813                                 $context = $edit->orig;
 814                         }
 815                         else {
 816                                 if (! is_array($block)) {
 817                                         $context = array_slice($context, sizeof($context) - $nlead);
 818                                         $x0 = $xi - sizeof($context);
 819                                         $y0 = $yi - sizeof($context);
 820                                         $block = array();
 821                                         if ($context)
 822                                         $block[] = new _DiffOp_Copy($context);
 823                                 }
 824                                 $block[] = $edit;
 825                         }
 826
 827                         if ($edit->orig)
 828                         $xi += sizeof($edit->orig);
 829                         if ($edit->closing)
 830                         $yi += sizeof($edit->closing);
 831                 }
 832
 833                 if (is_array($block))
 834                 $this->_block($x0, $xi - $x0,
 835                 $y0, $yi - $y0,
 836                 $block);
 837
 838                 $end = $this->_end_diff();
 839                 wfProfileOut( __METHOD__ );
 840                 return $end;
 841         }
 842
 843         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 844                 wfProfileIn( __METHOD__ );
 845                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 846                 foreach ($edits as $edit) {
 847                         if ($edit->type == 'copy')
 848                         $this->_context($edit->orig);
 849                         elseif ($edit->type == 'add')
 850                         $this->_added($edit->closing);
 851                         elseif ($edit->type == 'delete')
 852                         $this->_deleted($edit->orig);
 853                         elseif ($edit->type == 'change')
 854                         $this->_changed($edit->orig, $edit->closing);
 855                         else
 856                         trigger_error('Unknown edit type', E_USER_ERROR);
 857                 }
 858                 $this->_end_block();
 859                 wfProfileOut( __METHOD__ );
 860         }
 861
 862         function _start_diff() {
 863                 ob_start();
 864         }
 865
 866         function _end_diff() {
 867                 $val = ob_get_contents();
 868                 ob_end_clean();
 869                 return $val;
 870         }
 871
 872         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 873                 if ($xlen > 1)
 874                 $xbeg .= "," . ($xbeg + $xlen - 1);
 875                 if ($ylen > 1)
 876                 $ybeg .= "," . ($ybeg + $ylen - 1);
 877
 878                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 879         }
 880
 881         function _start_block($header) {
 882                 echo $header . "\n";
 883         }
 884
 885         function _end_block() {
 886         }
 887
 888         function _lines($lines, $prefix = ' ') {
 889                 foreach ($lines as $line)
 890                 echo "$prefix $line\n";
 891         }
 892
 893         function _context($lines) {
 894                 $this->_lines($lines);
 895         }
 896
 897         function _added($lines) {
 898                 $this->_lines($lines, '>');
 899         }
 900         function _deleted($lines) {
 901                 $this->_lines($lines, '<');
 902         }
 903
 904         function _changed($orig, $closing) {
 905                 $this->_deleted($orig);
 906                 echo "---\n";
 907                 $this->_added($closing);
 908         }
 909 }
 910
 911 /**
 912  * A formatter that outputs unified diffs
 913  * @ingroup DifferenceEngine
 914  */
 915
 916 class UnifiedDiffFormatter extends DiffFormatter {
 917         var $leading_context_lines = 2;
 918         var $trailing_context_lines = 2;
 919
 920         function _added($lines) {
 921                 $this->_lines($lines, '+');
 922         }
 923         function _deleted($lines) {
 924                 $this->_lines($lines, '-');
 925         }
 926         function _changed($orig, $closing) {
 927                 $this->_deleted($orig);
 928                 $this->_added($closing);
 929         }
 930         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 931                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
 932         }
 933 }
 934
 935 /**
 936  * A pseudo-formatter that just passes along the Diff::$edits array
 937  * @ingroup DifferenceEngine
 938  */
 939 class ArrayDiffFormatter extends DiffFormatter {
 940         function format($diff) {
 941                 $oldline = 1;
 942                 $newline = 1;
 943                 $retval = array();
 944                 foreach($diff->edits as $edit)
 945                 switch($edit->type) {
 946                         case 'add':
 947                                 foreach($edit->closing as $l) {
 948                                         $retval[] = array(
 949                                                         'action' => 'add',
 950                                                         'new'=> $l,
 951                                                         'newline' => $newline++
 952                                         );
 953                                 }
 954                                 break;
 955                         case 'delete':
 956                                 foreach($edit->orig as $l) {
 957                                         $retval[] = array(
 958                                                         'action' => 'delete',
 959                                                         'old' => $l,
 960                                                         'oldline' => $oldline++,
 961                                         );
 962                                 }
 963                                 break;
 964                         case 'change':
 965                                 foreach($edit->orig as $i => $l) {
 966                                         $retval[] = array(
 967                                                         'action' => 'change',
 968                                                         'old' => $l,
 969                                                         'new' => @$edit->closing[$i],
 970                                                         'oldline' => $oldline++,
 971                                                         'newline' => $newline++,
 972                                         );
 973                                 }
 974                                 break;
 975                         case 'copy':
 976                                 $oldline += count($edit->orig);
 977                                 $newline += count($edit->orig);
 978                 }
 979                 return $retval;
 980         }
 981 }
 982
 983 /**
 984  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
 985  *
 986  */
 987
 988 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
 989
 990 /**
 991  * @todo document
 992  * @private
 993  * @ingroup DifferenceEngine
 994  */
 995 class _HWLDF_WordAccumulator {
 996         function __construct () {
 997                 $this->_lines = array();
 998                 $this->_line = '';
 999                 $this->_group = '';
1000                 $this->_tag = '';
1001         }
1002
1003         function _flushGroup ($new_tag) {
1004                 if ($this->_group !== '') {
1005                         if ($this->_tag == 'ins')
1006                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1007                         htmlspecialchars ( $this->_group ) . '</ins>';
1008                         elseif ($this->_tag == 'del')
1009                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1010                         htmlspecialchars ( $this->_group ) . '</del>';
1011                         else
1012                         $this->_line .= htmlspecialchars ( $this->_group );
1013                 }
1014                 $this->_group = '';
1015                 $this->_tag = $new_tag;
1016         }
1017
1018         function _flushLine ($new_tag) {
1019                 $this->_flushGroup($new_tag);
1020                 if ($this->_line != '')
1021                 array_push ( $this->_lines, $this->_line );
1022                 else
1023                 # make empty lines visible by inserting an NBSP
1024                 array_push ( $this->_lines, NBSP );
1025                 $this->_line = '';
1026         }
1027
1028         function addWords ($words, $tag = '') {
1029                 if ($tag != $this->_tag)
1030                 $this->_flushGroup($tag);
1031
1032                 foreach ($words as $word) {
1033                         // new-line should only come as first char of word.
1034                         if ($word == '')
1035                         continue;
1036                         if ($word[0] == "\n") {
1037                                 $this->_flushLine($tag);
1038                                 $word = substr($word, 1);
1039                         }
1040                         assert(!strstr($word, "\n"));
1041                         $this->_group .= $word;
1042                 }
1043         }
1044
1045         function getLines() {
1046                 $this->_flushLine('~done');
1047                 return $this->_lines;
1048         }
1049 }
1050
1051 /**
1052  * @todo document
1053  * @private
1054  * @ingroup DifferenceEngine
1055  */
1056 class WordLevelDiff extends MappedDiff {
1057         const MAX_LINE_LENGTH = 10000;
1058
1059         function __construct ($orig_lines, $closing_lines) {
1060                 wfProfileIn( __METHOD__ );
1061
1062                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1063                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1064
1065                 parent::__construct($orig_words, $closing_words,
1066                 $orig_stripped, $closing_stripped);
1067                 wfProfileOut( __METHOD__ );
1068         }
1069
1070         function _split($lines) {
1071                 wfProfileIn( __METHOD__ );
1072
1073                 $words = array();
1074                 $stripped = array();
1075                 $first = true;
1076                 foreach ( $lines as $line ) {
1077                         # If the line is too long, just pretend the entire line is one big word
1078                         # This prevents resource exhaustion problems
1079                         if ( $first ) {
1080                                 $first = false;
1081                         } else {
1082                                 $words[] = "\n";
1083                                 $stripped[] = "\n";
1084                         }
1085                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1086                                 $words[] = $line;
1087                                 $stripped[] = $line;
1088                         } else {
1089                                 $m = array();
1090                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1091                                 $line, $m))
1092                                 {
1093                                         $words = array_merge( $words, $m[0] );
1094                                         $stripped = array_merge( $stripped, $m[1] );
1095                                 }
1096                         }
1097                 }
1098                 wfProfileOut( __METHOD__ );
1099                 return array($words, $stripped);
1100         }
1101
1102         function orig () {
1103                 wfProfileIn( __METHOD__ );
1104                 $orig = new _HWLDF_WordAccumulator;
1105
1106                 foreach ($this->edits as $edit) {
1107                         if ($edit->type == 'copy')
1108                         $orig->addWords($edit->orig);
1109                         elseif ($edit->orig)
1110                         $orig->addWords($edit->orig, 'del');
1111                 }
1112                 $lines = $orig->getLines();
1113                 wfProfileOut( __METHOD__ );
1114                 return $lines;
1115         }
1116
1117         function closing () {
1118                 wfProfileIn( __METHOD__ );
1119                 $closing = new _HWLDF_WordAccumulator;
1120
1121                 foreach ($this->edits as $edit) {
1122                         if ($edit->type == 'copy')
1123                         $closing->addWords($edit->closing);
1124                         elseif ($edit->closing)
1125                         $closing->addWords($edit->closing, 'ins');
1126                 }
1127                 $lines = $closing->getLines();
1128                 wfProfileOut( __METHOD__ );
1129                 return $lines;
1130         }
1131 }
1132
1133 /**
1134  * Wikipedia Table style diff formatter.
1135  * @todo document
1136  * @private
1137  * @ingroup DifferenceEngine
1138  */
1139 class TableDiffFormatter extends DiffFormatter {
1140         function __construct() {
1141                 $this->leading_context_lines = 2;
1142                 $this->trailing_context_lines = 2;
1143         }
1144
1145         public static function escapeWhiteSpace( $msg ) {
1146                 $msg = preg_replace( '/^ /m', '&#160; ', $msg );
1147                 $msg = preg_replace( '/ $/m', ' &#160;', $msg );
1148                 $msg = preg_replace( '/  /', '&#160; ', $msg );
1149                 return $msg;
1150         }
1151
1152         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1153                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1154                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1155                 return $r;
1156         }
1157
1158         function _start_block( $header ) {
1159                 echo $header;
1160         }
1161
1162         function _end_block() {
1163         }
1164
1165         function _lines( $lines, $prefix=' ', $color='white' ) {
1166         }
1167
1168         # HTML-escape parameter before calling this
1169         function addedLine( $line ) {
1170                 return $this->wrapLine( '+', 'diff-addedline', $line );
1171         }
1172
1173         # HTML-escape parameter before calling this
1174         function deletedLine( $line ) {
1175                 return $this->wrapLine( '&minus;', 'diff-deletedline', $line );
1176         }
1177
1178         # HTML-escape parameter before calling this
1179         function contextLine( $line ) {
1180                 return $this->wrapLine( '&#160;', 'diff-context', $line );
1181         }
1182
1183         private function wrapLine( $marker, $class, $line ) {
1184                 if( $line !== '' ) {
1185                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1186                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1187                 }
1188                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1189         }
1190
1191         function emptyLine() {
1192                 return '<td colspan="2">&#160;</td>';
1193         }
1194
1195         function _added( $lines ) {
1196                 foreach ($lines as $line) {
1197                         echo '<tr>' . $this->emptyLine() .
1198                         $this->addedLine( '<ins class="diffchange">' .
1199                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1200                 }
1201         }
1202
1203         function _deleted($lines) {
1204                 foreach ($lines as $line) {
1205                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1206                         htmlspecialchars ( $line ) . '</del>' ) .
1207                         $this->emptyLine() . "</tr>\n";
1208                 }
1209         }
1210
1211         function _context( $lines ) {
1212                 foreach ($lines as $line) {
1213                         echo '<tr>' .
1214                         $this->contextLine( htmlspecialchars ( $line ) ) .
1215                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1216                 }
1217         }
1218
1219         function _changed( $orig, $closing ) {
1220                 wfProfileIn( __METHOD__ );
1221
1222                 $diff = new WordLevelDiff( $orig, $closing );
1223                 $del = $diff->orig();
1224                 $add = $diff->closing();
1225
1226                 # Notice that WordLevelDiff returns HTML-escaped output.
1227                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1228
1229                 while ( $line = array_shift( $del ) ) {
1230                         $aline = array_shift( $add );
1231                         echo '<tr>' . $this->deletedLine( $line ) .
1232                         $this->addedLine( $aline ) . "</tr>\n";
1233                 }
1234                 foreach ($add as $line) {       # If any leftovers
1235                         echo '<tr>' . $this->emptyLine() .
1236                         $this->addedLine( $line ) . "</tr>\n";
1237                 }
1238                 wfProfileOut( __METHOD__ );
1239         }
1240 }