includes/diff/WikiDiff.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         protected $xchanged, $ychanged;
 139
 140         function diff ( $from_lines, $to_lines ) {
 141                 wfProfileIn( __METHOD__ );
 142
 143                 // Diff and store locally
 144                 $this->diff_local( $from_lines, $to_lines );
 145
 146                 // Merge edits when possible
 147                 $this->_shift_boundaries( $from_lines, $this->xchanged, $this->ychanged );
 148                 $this->_shift_boundaries( $to_lines, $this->ychanged, $this->xchanged );
 149
 150                 // Compute the edit operations.
 151                 $n_from = sizeof( $from_lines );
 152                 $n_to = sizeof( $to_lines );
 153
 154                 $edits = array();
 155                 $xi = $yi = 0;
 156                 while ( $xi < $n_from || $yi < $n_to ) {
 157                         assert( $yi < $n_to || $this->xchanged[$xi] );
 158                         assert( $xi < $n_from || $this->ychanged[$yi] );
 159
 160                         // Skip matching "snake".
 161                         $copy = array();
 162                         while ( $xi < $n_from && $yi < $n_to
 163                         && !$this->xchanged[$xi] && !$this->ychanged[$yi] ) {
 164                                 $copy[] = $from_lines[$xi++];
 165                                 ++$yi;
 166                         }
 167                         if ( $copy ) {
 168                                 $edits[] = new _DiffOp_Copy( $copy );
 169                         }
 170
 171                         // Find deletes & adds.
 172                         $delete = array();
 173                         while ( $xi < $n_from && $this->xchanged[$xi] ) {
 174                                 $delete[] = $from_lines[$xi++];
 175                         }
 176
 177                         $add = array();
 178                         while ( $yi < $n_to && $this->ychanged[$yi] )  {
 179                                 $add[] = $to_lines[$yi++];
 180                         }
 181
 182                         if ( $delete && $add ) {
 183                                 $edits[] = new _DiffOp_Change( $delete, $add );
 184                         } elseif ( $delete ) {
 185                                 $edits[] = new _DiffOp_Delete( $delete );
 186                         } elseif ( $add ) {
 187                                 $edits[] = new _DiffOp_Add( $add );
 188                         }
 189                 }
 190                 wfProfileOut( __METHOD__ );
 191                 return $edits;
 192         }
 193
 194         function diff_local ( $from_lines, $to_lines ) {
 195                 global $wgExternalDiffEngine;
 196                 wfProfileIn( __METHOD__ );
 197
 198                 if ( $wgExternalDiffEngine == 'wikidiff3' ) {
 199                         // wikidiff3
 200                         $wikidiff3 = new WikiDiff3();
 201                         $wikidiff3->diff( $from_lines, $to_lines );
 202                         $this->xchanged = $wikidiff3->removed;
 203                         $this->ychanged = $wikidiff3->added;
 204                         unset( $wikidiff3 );
 205                 } else {
 206                         // old diff
 207                         $n_from = sizeof( $from_lines );
 208                         $n_to = sizeof( $to_lines );
 209                         $this->xchanged = $this->ychanged = array();
 210                         $this->xv = $this->yv = array();
 211                         $this->xind = $this->yind = array();
 212                         unset( $this->seq );
 213                         unset( $this->in_seq );
 214                         unset( $this->lcs );
 215
 216                         // Skip leading common lines.
 217                         for ( $skip = 0; $skip < $n_from && $skip < $n_to; $skip++ ) {
 218                                 if ( $from_lines[$skip] !== $to_lines[$skip] )
 219                                 break;
 220                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 221                         }
 222                         // Skip trailing common lines.
 223                         $xi = $n_from; $yi = $n_to;
 224                         for ( $endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++ ) {
 225                                 if ( $from_lines[$xi] !== $to_lines[$yi] )
 226                                 break;
 227                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 228                         }
 229
 230                         // Ignore lines which do not exist in both files.
 231                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 232                                 $xhash[$this->_line_hash( $from_lines[$xi] )] = 1;
 233                         }
 234
 235                         for ( $yi = $skip; $yi < $n_to - $endskip; $yi++ ) {
 236                                 $line = $to_lines[$yi];
 237                                 if ( ( $this->ychanged[$yi] = empty( $xhash[$this->_line_hash( $line )] ) ) )
 238                                 continue;
 239                                 $yhash[$this->_line_hash( $line )] = 1;
 240                                 $this->yv[] = $line;
 241                                 $this->yind[] = $yi;
 242                         }
 243                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 244                                 $line = $from_lines[$xi];
 245                                 if ( ( $this->xchanged[$xi] = empty( $yhash[$this->_line_hash( $line )] ) ) )
 246                                 continue;
 247                                 $this->xv[] = $line;
 248                                 $this->xind[] = $xi;
 249                         }
 250
 251                         // Find the LCS.
 252                         $this->_compareseq( 0, sizeof( $this->xv ), 0, sizeof( $this->yv ) );
 253                 }
 254                 wfProfileOut( __METHOD__ );
 255         }
 256
 257         /**
 258          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 259          */
 260         function _line_hash( $line ) {
 261                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 262                         return md5( $line );
 263                 } else {
 264                         return $line;
 265                 }
 266         }
 267
 268         /* Divide the Largest Common Subsequence (LCS) of the sequences
 269          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 270          * sized segments.
 271          *
 272          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 273          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 274          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 275          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 276          * that (X0, Y0) == (XOFF, YOFF) and
 277          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 278          *
 279          * This function assumes that the first lines of the specified portions
 280          * of the two files do not match, and likewise that the last lines do not
 281          * match.  The caller must trim matching lines from the beginning and end
 282          * of the portions it is going to specify.
 283          */
 284         function _diag ( $xoff, $xlim, $yoff, $ylim, $nchunks ) {
 285                 $flip = false;
 286
 287                 if ( $xlim - $xoff > $ylim - $yoff ) {
 288                         // Things seems faster (I'm not sure I understand why)
 289                         // when the shortest sequence in X.
 290                         $flip = true;
 291                         list ( $xoff, $xlim, $yoff, $ylim ) = array( $yoff, $ylim, $xoff, $xlim );
 292                 }
 293
 294                 if ( $flip ) {
 295                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 296                                 $ymatches[$this->xv[$i]][] = $i;
 297                         }
 298                 } else {
 299                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 300                                 $ymatches[$this->yv[$i]][] = $i;
 301                         }
 302                 }
 303
 304                 $this->lcs = 0;
 305                 $this->seq[0] = $yoff - 1;
 306                 $this->in_seq = array();
 307                 $ymids[0] = array();
 308
 309                 $numer = $xlim - $xoff + $nchunks - 1;
 310                 $x = $xoff;
 311                 for ( $chunk = 0; $chunk < $nchunks; $chunk++ ) {
 312                         if ( $chunk > 0 ) {
 313                                 for ( $i = 0; $i <= $this->lcs; $i++ ) {
 314                                         $ymids[$i][$chunk -1] = $this->seq[$i];
 315                                 }
 316                         }
 317
 318                         $x1 = $xoff + (int)( ( $numer + ( $xlim -$xoff ) * $chunk ) / $nchunks );
 319                         for ( ; $x < $x1; $x++ ) {
 320                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 321                                 if ( empty( $ymatches[$line] ) ) {
 322                                         continue;
 323                                 }
 324                                 $matches = $ymatches[$line];
 325                                 reset( $matches );
 326                                 while ( list( , $y ) = each( $matches ) ) {
 327                                         if ( empty( $this->in_seq[$y] ) ) {
 328                                                 $k = $this->_lcs_pos( $y );
 329                                                 assert( $k > 0 );
 330                                                 $ymids[$k] = $ymids[$k -1];
 331                                                 break;
 332                                         }
 333                                 }
 334                                 while ( list ( , $y ) = each( $matches ) ) {
 335                                         if ( $y > $this->seq[$k -1] ) {
 336                                                 assert( $y < $this->seq[$k] );
 337                                                 // Optimization: this is a common case:
 338                                                 //      next match is just replacing previous match.
 339                                                 $this->in_seq[$this->seq[$k]] = false;
 340                                                 $this->seq[$k] = $y;
 341                                                 $this->in_seq[$y] = 1;
 342                                         } else if ( empty( $this->in_seq[$y] ) ) {
 343                                                 $k = $this->_lcs_pos( $y );
 344                                                 assert( $k > 0 );
 345                                                 $ymids[$k] = $ymids[$k -1];
 346                                         }
 347                                 }
 348                         }
 349                 }
 350
 351                 $seps[] = $flip ? array( $yoff, $xoff ) : array( $xoff, $yoff );
 352                 $ymid = $ymids[$this->lcs];
 353                 for ( $n = 0; $n < $nchunks - 1; $n++ ) {
 354                         $x1 = $xoff + (int)( ( $numer + ( $xlim - $xoff ) * $n ) / $nchunks );
 355                         $y1 = $ymid[$n] + 1;
 356                         $seps[] = $flip ? array( $y1, $x1 ) : array( $x1, $y1 );
 357                 }
 358                 $seps[] = $flip ? array( $ylim, $xlim ) : array( $xlim, $ylim );
 359
 360                 return array( $this->lcs, $seps );
 361         }
 362
 363         function _lcs_pos ( $ypos ) {
 364                 $end = $this->lcs;
 365                 if ( $end == 0 || $ypos > $this->seq[$end] ) {
 366                         $this->seq[++$this->lcs] = $ypos;
 367                         $this->in_seq[$ypos] = 1;
 368                         return $this->lcs;
 369                 }
 370
 371                 $beg = 1;
 372                 while ( $beg < $end ) {
 373                         $mid = (int)( ( $beg + $end ) / 2 );
 374                         if ( $ypos > $this->seq[$mid] )
 375                         $beg = $mid + 1;
 376                         else
 377                         $end = $mid;
 378                 }
 379
 380                 assert( $ypos != $this->seq[$end] );
 381
 382                 $this->in_seq[$this->seq[$end]] = false;
 383                 $this->seq[$end] = $ypos;
 384                 $this->in_seq[$ypos] = 1;
 385                 return $end;
 386         }
 387
 388         /* Find LCS of two sequences.
 389          *
 390          * The results are recorded in the vectors $this->{x,y}changed[], by
 391          * storing a 1 in the element for each line that is an insertion
 392          * or deletion (ie. is not in the LCS).
 393          *
 394          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 395          *
 396          * Note that XLIM, YLIM are exclusive bounds.
 397          * All line numbers are origin-0 and discarded lines are not counted.
 398          */
 399         function _compareseq ( $xoff, $xlim, $yoff, $ylim ) {
 400                 // Slide down the bottom initial diagonal.
 401                 while ( $xoff < $xlim && $yoff < $ylim
 402                 && $this->xv[$xoff] == $this->yv[$yoff] ) {
 403                         ++$xoff;
 404                         ++$yoff;
 405                 }
 406
 407                 // Slide up the top initial diagonal.
 408                 while ( $xlim > $xoff && $ylim > $yoff
 409                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1] ) {
 410                         --$xlim;
 411                         --$ylim;
 412                 }
 413
 414                 if ( $xoff == $xlim || $yoff == $ylim ) {
 415                         $lcs = 0;
 416                 } else {
 417                         // This is ad hoc but seems to work well.
 418                         // $nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 419                         // $nchunks = max(2,min(8,(int)$nchunks));
 420                         $nchunks = min( 7, $xlim - $xoff, $ylim - $yoff ) + 1;
 421                         list ( $lcs, $seps )
 422                         = $this->_diag( $xoff, $xlim, $yoff, $ylim, $nchunks );
 423                 }
 424
 425                 if ( $lcs == 0 ) {
 426                         // X and Y sequences have no common subsequence:
 427                         // mark all changed.
 428                         while ( $yoff < $ylim ) {
 429                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 430                         }
 431                         while ( $xoff < $xlim ) {
 432                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 433                         }
 434                 } else {
 435                         // Use the partitions to split this problem into subproblems.
 436                         reset( $seps );
 437                         $pt1 = $seps[0];
 438                         while ( $pt2 = next( $seps ) ) {
 439                                 $this->_compareseq ( $pt1[0], $pt2[0], $pt1[1], $pt2[1] );
 440                                 $pt1 = $pt2;
 441                         }
 442                 }
 443         }
 444
 445         /* Adjust inserts/deletes of identical lines to join changes
 446          * as much as possible.
 447          *
 448          * We do something when a run of changed lines include a
 449          * line at one end and has an excluded, identical line at the other.
 450          * We are free to choose which identical line is included.
 451          * `compareseq' usually chooses the one at the beginning,
 452          * but usually it is cleaner to consider the following identical line
 453          * to be the "change".
 454          *
 455          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 456          */
 457         function _shift_boundaries ( $lines, &$changed, $other_changed ) {
 458                 wfProfileIn( __METHOD__ );
 459                 $i = 0;
 460                 $j = 0;
 461
 462                 assert( 'sizeof($lines) == sizeof($changed)' );
 463                 $len = sizeof( $lines );
 464                 $other_len = sizeof( $other_changed );
 465
 466                 while ( 1 ) {
 467                         /*
 468                          * Scan forwards to find beginning of another run of changes.
 469                          * Also keep track of the corresponding point in the other file.
 470                          *
 471                          * Throughout this code, $i and $j are adjusted together so that
 472                          * the first $i elements of $changed and the first $j elements
 473                          * of $other_changed both contain the same number of zeros
 474                          * (unchanged lines).
 475                          * Furthermore, $j is always kept so that $j == $other_len or
 476                          * $other_changed[$j] == false.
 477                          */
 478                         while ( $j < $other_len && $other_changed[$j] ) {
 479                                 $j++;
 480                         }
 481
 482                         while ( $i < $len && ! $changed[$i] ) {
 483                                 assert( '$j < $other_len && ! $other_changed[$j]' );
 484                                 $i++; $j++;
 485                                 while ( $j < $other_len && $other_changed[$j] )
 486                                 $j++;
 487                         }
 488
 489                         if ( $i == $len ) {
 490                                 break;
 491                         }
 492
 493                         $start = $i;
 494
 495                         // Find the end of this run of changes.
 496                         while ( ++$i < $len && $changed[$i] ) {
 497                                 continue;
 498                         }
 499
 500                         do {
 501                                 /*
 502                                  * Record the length of this run of changes, so that
 503                                  * we can later determine whether the run has grown.
 504                                  */
 505                                 $runlength = $i - $start;
 506
 507                                 /*
 508                                  * Move the changed region back, so long as the
 509                                  * previous unchanged line matches the last changed one.
 510                                  * This merges with previous changed regions.
 511                                  */
 512                                 while ( $start > 0 && $lines[$start - 1] == $lines[$i - 1] ) {
 513                                         $changed[--$start] = 1;
 514                                         $changed[--$i] = false;
 515                                         while ( $start > 0 && $changed[$start - 1] ) {
 516                                                 $start--;
 517                                         }
 518                                         assert( '$j > 0' );
 519                                         while ( $other_changed[--$j] ) {
 520                                                 continue;
 521                                         }
 522                                         assert( '$j >= 0 && !$other_changed[$j]' );
 523                                 }
 524
 525                                 /*
 526                                  * Set CORRESPONDING to the end of the changed run, at the last
 527                                  * point where it corresponds to a changed run in the other file.
 528                                  * CORRESPONDING == LEN means no such point has been found.
 529                                  */
 530                                 $corresponding = $j < $other_len ? $i : $len;
 531
 532                                 /*
 533                                  * Move the changed region forward, so long as the
 534                                  * first changed line matches the following unchanged one.
 535                                  * This merges with following changed regions.
 536                                  * Do this second, so that if there are no merges,
 537                                  * the changed region is moved forward as far as possible.
 538                                  */
 539                                 while ( $i < $len && $lines[$start] == $lines[$i] ) {
 540                                         $changed[$start++] = false;
 541                                         $changed[$i++] = 1;
 542                                         while ( $i < $len && $changed[$i] ) {
 543                                                 $i++;
 544                                         }
 545
 546                                         assert( '$j < $other_len && ! $other_changed[$j]' );
 547                                         $j++;
 548                                         if ( $j < $other_len && $other_changed[$j] ) {
 549                                                 $corresponding = $i;
 550                                                 while ( $j < $other_len && $other_changed[$j] ) {
 551                                                         $j++;
 552                                                 }
 553                                         }
 554                                 }
 555                         } while ( $runlength != $i - $start );
 556
 557                         /*
 558                          * If possible, move the fully-merged run of changes
 559                          * back to a corresponding run in the other file.
 560                          */
 561                         while ( $corresponding < $i ) {
 562                                 $changed[--$start] = 1;
 563                                 $changed[--$i] = 0;
 564                                 assert( '$j > 0' );
 565                                 while ( $other_changed[--$j] ) {
 566                                         continue;
 567                                 }
 568                                 assert( '$j >= 0 && !$other_changed[$j]' );
 569                         }
 570                 }
 571                 wfProfileOut( __METHOD__ );
 572         }
 573 }
 574
 575 /**
 576  * Class representing a 'diff' between two sequences of strings.
 577  * @todo document
 578  * @private
 579  * @ingroup DifferenceEngine
 580  */
 581 class Diff
 582 {
 583         var $edits;
 584
 585         /**
 586          * Constructor.
 587          * Computes diff between sequences of strings.
 588          *
 589          * @param $from_lines array An array of strings.
 590          *                (Typically these are lines from a file.)
 591          * @param $to_lines array An array of strings.
 592          */
 593         function __construct( $from_lines, $to_lines ) {
 594                 $eng = new _DiffEngine;
 595                 $this->edits = $eng->diff( $from_lines, $to_lines );
 596                 // $this->_check($from_lines, $to_lines);
 597         }
 598
 599         /**
 600          * Compute reversed Diff.
 601          *
 602          * SYNOPSIS:
 603          *
 604          *      $diff = new Diff($lines1, $lines2);
 605          *      $rev = $diff->reverse();
 606          * @return object A Diff object representing the inverse of the
 607          *                                original diff.
 608          */
 609         function reverse () {
 610                 $rev = $this;
 611                 $rev->edits = array();
 612                 foreach ( $this->edits as $edit ) {
 613                         $rev->edits[] = $edit->reverse();
 614                 }
 615                 return $rev;
 616         }
 617
 618         /**
 619          * Check for empty diff.
 620          *
 621          * @return bool True iff two sequences were identical.
 622          */
 623         function isEmpty () {
 624                 foreach ( $this->edits as $edit ) {
 625                         if ( $edit->type != 'copy' ) {
 626                                 return false;
 627                         }
 628                 }
 629                 return true;
 630         }
 631
 632         /**
 633          * Compute the length of the Longest Common Subsequence (LCS).
 634          *
 635          * This is mostly for diagnostic purposed.
 636          *
 637          * @return int The length of the LCS.
 638          */
 639         function lcs () {
 640                 $lcs = 0;
 641                 foreach ( $this->edits as $edit ) {
 642                         if ( $edit->type == 'copy' ) {
 643                                 $lcs += sizeof( $edit->orig );
 644                         }
 645                 }
 646                 return $lcs;
 647         }
 648
 649         /**
 650          * Get the original set of lines.
 651          *
 652          * This reconstructs the $from_lines parameter passed to the
 653          * constructor.
 654          *
 655          * @return array The original sequence of strings.
 656          */
 657         function orig() {
 658                 $lines = array();
 659
 660                 foreach ( $this->edits as $edit ) {
 661                         if ( $edit->orig ) {
 662                                 array_splice( $lines, sizeof( $lines ), 0, $edit->orig );
 663                         }
 664                 }
 665                 return $lines;
 666         }
 667
 668         /**
 669          * Get the closing set of lines.
 670          *
 671          * This reconstructs the $to_lines parameter passed to the
 672          * constructor.
 673          *
 674          * @return array The sequence of strings.
 675          */
 676         function closing() {
 677                 $lines = array();
 678
 679                 foreach ( $this->edits as $edit ) {
 680                         if ( $edit->closing ) {
 681                                 array_splice( $lines, sizeof( $lines ), 0, $edit->closing );
 682                         }
 683                 }
 684                 return $lines;
 685         }
 686
 687         /**
 688          * Check a Diff for validity.
 689          *
 690          * This is here only for debugging purposes.
 691          */
 692         function _check ( $from_lines, $to_lines ) {
 693                 wfProfileIn( __METHOD__ );
 694                 if ( serialize( $from_lines ) != serialize( $this->orig() ) ) {
 695                         trigger_error( "Reconstructed original doesn't match", E_USER_ERROR );
 696                 }
 697                 if ( serialize( $to_lines ) != serialize( $this->closing() ) ) {
 698                         trigger_error( "Reconstructed closing doesn't match", E_USER_ERROR );
 699                 }
 700
 701                 $rev = $this->reverse();
 702                 if ( serialize( $to_lines ) != serialize( $rev->orig() ) ) {
 703                         trigger_error( "Reversed original doesn't match", E_USER_ERROR );
 704                 }
 705                 if ( serialize( $from_lines ) != serialize( $rev->closing() ) ) {
 706                         trigger_error( "Reversed closing doesn't match", E_USER_ERROR );
 707                 }
 708
 709
 710                 $prevtype = 'none';
 711                 foreach ( $this->edits as $edit ) {
 712                         if ( $prevtype == $edit->type ) {
 713                                 trigger_error( "Edit sequence is non-optimal", E_USER_ERROR );
 714                         }
 715                         $prevtype = $edit->type;
 716                 }
 717
 718                 $lcs = $this->lcs();
 719                 trigger_error( 'Diff okay: LCS = ' . $lcs, E_USER_NOTICE );
 720                 wfProfileOut( __METHOD__ );
 721         }
 722 }
 723
 724 /**
 725  * @todo document, bad name.
 726  * @private
 727  * @ingroup DifferenceEngine
 728  */
 729 class MappedDiff extends Diff
 730 {
 731         /**
 732          * Constructor.
 733          *
 734          * Computes diff between sequences of strings.
 735          *
 736          * This can be used to compute things like
 737          * case-insensitve diffs, or diffs which ignore
 738          * changes in white-space.
 739          *
 740          * @param $from_lines array An array of strings.
 741          *      (Typically these are lines from a file.)
 742          *
 743          * @param $to_lines array An array of strings.
 744          *
 745          * @param $mapped_from_lines array This array should
 746          *      have the same size number of elements as $from_lines.
 747          *      The elements in $mapped_from_lines and
 748          *      $mapped_to_lines are what is actually compared
 749          *      when computing the diff.
 750          *
 751          * @param $mapped_to_lines array This array should
 752          *      have the same number of elements as $to_lines.
 753          */
 754         function __construct( $from_lines, $to_lines,
 755                 $mapped_from_lines, $mapped_to_lines ) {
 756                 wfProfileIn( __METHOD__ );
 757
 758                 assert( sizeof( $from_lines ) == sizeof( $mapped_from_lines ) );
 759                 assert( sizeof( $to_lines ) == sizeof( $mapped_to_lines ) );
 760
 761                 parent::__construct( $mapped_from_lines, $mapped_to_lines );
 762
 763                 $xi = $yi = 0;
 764                 for ( $i = 0; $i < sizeof( $this->edits ); $i++ ) {
 765                         $orig = &$this->edits[$i]->orig;
 766                         if ( is_array( $orig ) ) {
 767                                 $orig = array_slice( $from_lines, $xi, sizeof( $orig ) );
 768                                 $xi += sizeof( $orig );
 769                         }
 770
 771                         $closing = &$this->edits[$i]->closing;
 772                         if ( is_array( $closing ) ) {
 773                                 $closing = array_slice( $to_lines, $yi, sizeof( $closing ) );
 774                                 $yi += sizeof( $closing );
 775                         }
 776                 }
 777                 wfProfileOut( __METHOD__ );
 778         }
 779 }
 780
 781 /**
 782  * A class to format Diffs
 783  *
 784  * This class formats the diff in classic diff format.
 785  * It is intended that this class be customized via inheritance,
 786  * to obtain fancier outputs.
 787  * @todo document
 788  * @private
 789  * @ingroup DifferenceEngine
 790  */
 791 class DiffFormatter {
 792         /**
 793          * Number of leading context "lines" to preserve.
 794          *
 795          * This should be left at zero for this class, but subclasses
 796          * may want to set this to other values.
 797          */
 798         var $leading_context_lines = 0;
 799
 800         /**
 801          * Number of trailing context "lines" to preserve.
 802          *
 803          * This should be left at zero for this class, but subclasses
 804          * may want to set this to other values.
 805          */
 806         var $trailing_context_lines = 0;
 807
 808         /**
 809          * Format a diff.
 810          *
 811          * @param $diff Diff A Diff object.
 812          * @return string The formatted output.
 813          */
 814         function format( $diff ) {
 815                 wfProfileIn( __METHOD__ );
 816
 817                 $xi = $yi = 1;
 818                 $block = false;
 819                 $context = array();
 820
 821                 $nlead = $this->leading_context_lines;
 822                 $ntrail = $this->trailing_context_lines;
 823
 824                 $this->_start_diff();
 825
 826                 foreach ( $diff->edits as $edit ) {
 827                         if ( $edit->type == 'copy' ) {
 828                                 if ( is_array( $block ) ) {
 829                                         if ( sizeof( $edit->orig ) <= $nlead + $ntrail ) {
 830                                                 $block[] = $edit;
 831                                         }
 832                                         else {
 833                                                 if ( $ntrail ) {
 834                                                         $context = array_slice( $edit->orig, 0, $ntrail );
 835                                                         $block[] = new _DiffOp_Copy( $context );
 836                                                 }
 837                                                 $this->_block( $x0, $ntrail + $xi - $x0,
 838                                                         $y0, $ntrail + $yi - $y0,
 839                                                         $block );
 840                                                 $block = false;
 841                                         }
 842                                 }
 843                                 $context = $edit->orig;
 844                         }
 845                         else {
 846                                 if ( ! is_array( $block ) ) {
 847                                         $context = array_slice( $context, sizeof( $context ) - $nlead );
 848                                         $x0 = $xi - sizeof( $context );
 849                                         $y0 = $yi - sizeof( $context );
 850                                         $block = array();
 851                                         if ( $context ) {
 852                                                 $block[] = new _DiffOp_Copy( $context );
 853                                         }
 854                                 }
 855                                 $block[] = $edit;
 856                         }
 857
 858                         if ( $edit->orig ) {
 859                                 $xi += sizeof( $edit->orig );
 860                         }
 861                         if ( $edit->closing ) {
 862                                 $yi += sizeof( $edit->closing );
 863                         }
 864                 }
 865
 866                 if ( is_array( $block ) ) {
 867                         $this->_block( $x0, $xi - $x0,
 868                                 $y0, $yi - $y0,
 869                                 $block );
 870                 }
 871
 872                 $end = $this->_end_diff();
 873                 wfProfileOut( __METHOD__ );
 874                 return $end;
 875         }
 876
 877         function _block( $xbeg, $xlen, $ybeg, $ylen, &$edits ) {
 878                 wfProfileIn( __METHOD__ );
 879                 $this->_start_block( $this->_block_header( $xbeg, $xlen, $ybeg, $ylen ) );
 880                 foreach ( $edits as $edit ) {
 881                         if ( $edit->type == 'copy' ) {
 882                                 $this->_context( $edit->orig );
 883                         } elseif ( $edit->type == 'add' ) {
 884                                 $this->_added( $edit->closing );
 885                         } elseif ( $edit->type == 'delete' ) {
 886                                 $this->_deleted( $edit->orig );
 887                         } elseif ( $edit->type == 'change' ) {
 888                                 $this->_changed( $edit->orig, $edit->closing );
 889                         } else {
 890                                 trigger_error( 'Unknown edit type', E_USER_ERROR );
 891                         }
 892                 }
 893                 $this->_end_block();
 894                 wfProfileOut( __METHOD__ );
 895         }
 896
 897         function _start_diff() {
 898                 ob_start();
 899         }
 900
 901         function _end_diff() {
 902                 $val = ob_get_contents();
 903                 ob_end_clean();
 904                 return $val;
 905         }
 906
 907         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
 908                 if ( $xlen > 1 ) {
 909                         $xbeg .= "," . ( $xbeg + $xlen - 1 );
 910                 }
 911                 if ( $ylen > 1 ) {
 912                         $ybeg .= "," . ( $ybeg + $ylen - 1 );
 913                 }
 914
 915                 return $xbeg . ( $xlen ? ( $ylen ? 'c' : 'd' ) : 'a' ) . $ybeg;
 916         }
 917
 918         function _start_block( $header ) {
 919                 echo $header . "\n";
 920         }
 921
 922         function _end_block() {
 923         }
 924
 925         function _lines( $lines, $prefix = ' ' ) {
 926                 foreach ( $lines as $line ) {
 927                         echo "$prefix $line\n";
 928                 }
 929         }
 930
 931         function _context( $lines ) {
 932                 $this->_lines( $lines );
 933         }
 934
 935         function _added( $lines ) {
 936                 $this->_lines( $lines, '>' );
 937         }
 938         function _deleted( $lines ) {
 939                 $this->_lines( $lines, '<' );
 940         }
 941
 942         function _changed( $orig, $closing ) {
 943                 $this->_deleted( $orig );
 944                 echo "---\n";
 945                 $this->_added( $closing );
 946         }
 947 }
 948
 949 /**
 950  * A formatter that outputs unified diffs
 951  * @ingroup DifferenceEngine
 952  */
 953
 954 class UnifiedDiffFormatter extends DiffFormatter {
 955         var $leading_context_lines = 2;
 956         var $trailing_context_lines = 2;
 957
 958         function _added( $lines ) {
 959                 $this->_lines( $lines, '+' );
 960         }
 961         function _deleted( $lines ) {
 962                 $this->_lines( $lines, '-' );
 963         }
 964         function _changed( $orig, $closing ) {
 965                 $this->_deleted( $orig );
 966                 $this->_added( $closing );
 967         }
 968         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
 969                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
 970         }
 971 }
 972
 973 /**
 974  * A pseudo-formatter that just passes along the Diff::$edits array
 975  * @ingroup DifferenceEngine
 976  */
 977 class ArrayDiffFormatter extends DiffFormatter {
 978         function format( $diff ) {
 979                 $oldline = 1;
 980                 $newline = 1;
 981                 $retval = array();
 982                 foreach ( $diff->edits as $edit ) {
 983                         switch( $edit->type ) {
 984                                 case 'add':
 985                                         foreach ( $edit->closing as $l ) {
 986                                                 $retval[] = array(
 987                                                                 'action' => 'add',
 988                                                                 'new' => $l,
 989                                                                 'newline' => $newline++
 990                                                 );
 991                                         }
 992                                         break;
 993                                 case 'delete':
 994                                         foreach ( $edit->orig as $l ) {
 995                                                 $retval[] = array(
 996                                                                 'action' => 'delete',
 997                                                                 'old' => $l,
 998                                                                 'oldline' => $oldline++,
 999                                                 );
1000                                         }
1001                                         break;
1002                                 case 'change':
1003                                         foreach ( $edit->orig as $i => $l ) {
1004                                                 $retval[] = array(
1005                                                                 'action' => 'change',
1006                                                                 'old' => $l,
1007                                                                 'new' => @$edit->closing[$i],
1008                                                                 'oldline' => $oldline++,
1009                                                                 'newline' => $newline++,
1010                                                 );
1011                                         }
1012                                         break;
1013                                 case 'copy':
1014                                         $oldline += count( $edit->orig );
1015                                         $newline += count( $edit->orig );
1016                         }
1017                 }
1018                 return $retval;
1019         }
1020 }
1021
1022 /**
1023  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1024  *
1025  */
1026
1027 define( 'NBSP', '&#160;' ); // iso-8859-x non-breaking space.
1028
1029 /**
1030  * @todo document
1031  * @private
1032  * @ingroup DifferenceEngine
1033  */
1034 class _HWLDF_WordAccumulator {
1035         function __construct () {
1036                 $this->_lines = array();
1037                 $this->_line = '';
1038                 $this->_group = '';
1039                 $this->_tag = '';
1040         }
1041
1042         function _flushGroup ( $new_tag ) {
1043                 if ( $this->_group !== '' ) {
1044                         if ( $this->_tag == 'ins' ) {
1045                                 $this->_line .= '<ins class="diffchange diffchange-inline">' .
1046                                                 htmlspecialchars ( $this->_group ) . '</ins>';
1047                         } elseif ( $this->_tag == 'del' ) {
1048                                 $this->_line .= '<del class="diffchange diffchange-inline">' .
1049                                                 htmlspecialchars ( $this->_group ) . '</del>';
1050                         } else {
1051                                 $this->_line .= htmlspecialchars ( $this->_group );
1052                         }
1053                 }
1054                 $this->_group = '';
1055                 $this->_tag = $new_tag;
1056         }
1057
1058         function _flushLine ( $new_tag ) {
1059                 $this->_flushGroup( $new_tag );
1060                 if ( $this->_line != '' ) {
1061                         array_push ( $this->_lines, $this->_line );
1062                 } else {
1063                         # make empty lines visible by inserting an NBSP
1064                         array_push ( $this->_lines, NBSP );
1065                 }
1066                 $this->_line = '';
1067         }
1068
1069         function addWords ( $words, $tag = '' ) {
1070                 if ( $tag != $this->_tag ) {
1071                         $this->_flushGroup( $tag );
1072                 }
1073
1074                 foreach ( $words as $word ) {
1075                         // new-line should only come as first char of word.
1076                         if ( $word == '' ) {
1077                                 continue;
1078                         }
1079                         if ( $word[0] == "\n" ) {
1080                                 $this->_flushLine( $tag );
1081                                 $word = substr( $word, 1 );
1082                         }
1083                         assert( !strstr( $word, "\n" ) );
1084                         $this->_group .= $word;
1085                 }
1086         }
1087
1088         function getLines() {
1089                 $this->_flushLine( '~done' );
1090                 return $this->_lines;
1091         }
1092 }
1093
1094 /**
1095  * @todo document
1096  * @private
1097  * @ingroup DifferenceEngine
1098  */
1099 class WordLevelDiff extends MappedDiff {
1100         const MAX_LINE_LENGTH = 10000;
1101
1102         function __construct ( $orig_lines, $closing_lines ) {
1103                 wfProfileIn( __METHOD__ );
1104
1105                 list ( $orig_words, $orig_stripped ) = $this->_split( $orig_lines );
1106                 list ( $closing_words, $closing_stripped ) = $this->_split( $closing_lines );
1107
1108                 parent::__construct( $orig_words, $closing_words,
1109                 $orig_stripped, $closing_stripped );
1110                 wfProfileOut( __METHOD__ );
1111         }
1112
1113         function _split( $lines ) {
1114                 wfProfileIn( __METHOD__ );
1115
1116                 $words = array();
1117                 $stripped = array();
1118                 $first = true;
1119                 foreach ( $lines as $line ) {
1120                         # If the line is too long, just pretend the entire line is one big word
1121                         # This prevents resource exhaustion problems
1122                         if ( $first ) {
1123                                 $first = false;
1124                         } else {
1125                                 $words[] = "\n";
1126                                 $stripped[] = "\n";
1127                         }
1128                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1129                                 $words[] = $line;
1130                                 $stripped[] = $line;
1131                         } else {
1132                                 $m = array();
1133                                 if ( preg_match_all( '/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1134                                         $line, $m ) )
1135                                 {
1136                                         $words = array_merge( $words, $m[0] );
1137                                         $stripped = array_merge( $stripped, $m[1] );
1138                                 }
1139                         }
1140                 }
1141                 wfProfileOut( __METHOD__ );
1142                 return array( $words, $stripped );
1143         }
1144
1145         function orig () {
1146                 wfProfileIn( __METHOD__ );
1147                 $orig = new _HWLDF_WordAccumulator;
1148
1149                 foreach ( $this->edits as $edit ) {
1150                         if ( $edit->type == 'copy' ) {
1151                                 $orig->addWords( $edit->orig );
1152                         } elseif ( $edit->orig ) {
1153                                 $orig->addWords( $edit->orig, 'del' );
1154                         }
1155                 }
1156                 $lines = $orig->getLines();
1157                 wfProfileOut( __METHOD__ );
1158                 return $lines;
1159         }
1160
1161         function closing () {
1162                 wfProfileIn( __METHOD__ );
1163                 $closing = new _HWLDF_WordAccumulator;
1164
1165                 foreach ( $this->edits as $edit ) {
1166                         if ( $edit->type == 'copy' ) {
1167                                 $closing->addWords( $edit->closing );
1168                         } elseif ( $edit->closing ) {
1169                                 $closing->addWords( $edit->closing, 'ins' );
1170                         }
1171                 }
1172                 $lines = $closing->getLines();
1173                 wfProfileOut( __METHOD__ );
1174                 return $lines;
1175         }
1176 }
1177
1178 /**
1179  * Wikipedia Table style diff formatter.
1180  * @todo document
1181  * @private
1182  * @ingroup DifferenceEngine
1183  */
1184 class TableDiffFormatter extends DiffFormatter {
1185         function __construct() {
1186                 $this->leading_context_lines = 2;
1187                 $this->trailing_context_lines = 2;
1188         }
1189
1190         public static function escapeWhiteSpace( $msg ) {
1191                 $msg = preg_replace( '/^ /m', '&#160; ', $msg );
1192                 $msg = preg_replace( '/ $/m', ' &#160;', $msg );
1193                 $msg = preg_replace( '/  /', '&#160; ', $msg );
1194                 return $msg;
1195         }
1196
1197         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1198                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE ' . $xbeg . "--></td>\n" .
1199                   '<td colspan="2" class="diff-lineno"><!--LINE ' . $ybeg . "--></td></tr>\n";
1200                 return $r;
1201         }
1202
1203         function _start_block( $header ) {
1204                 echo $header;
1205         }
1206
1207         function _end_block() {
1208         }
1209
1210         function _lines( $lines, $prefix = ' ', $color = 'white' ) {
1211         }
1212
1213         # HTML-escape parameter before calling this
1214         function addedLine( $line ) {
1215                 return $this->wrapLine( '+', 'diff-addedline', $line );
1216         }
1217
1218         # HTML-escape parameter before calling this
1219         function deletedLine( $line ) {
1220                 return $this->wrapLine( '−', 'diff-deletedline', $line );
1221         }
1222
1223         # HTML-escape parameter before calling this
1224         function contextLine( $line ) {
1225                 return $this->wrapLine( '&#160;', 'diff-context', $line );
1226         }
1227
1228         private function wrapLine( $marker, $class, $line ) {
1229                 if ( $line !== '' ) {
1230                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1231                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1232                 }
1233                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1234         }
1235
1236         function emptyLine() {
1237                 return '<td colspan="2">&#160;</td>';
1238         }
1239
1240         function _added( $lines ) {
1241                 foreach ( $lines as $line ) {
1242                         echo '<tr>' . $this->emptyLine() .
1243                         $this->addedLine( '<ins class="diffchange">' .
1244                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1245                 }
1246         }
1247
1248         function _deleted( $lines ) {
1249                 foreach ( $lines as $line ) {
1250                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1251                         htmlspecialchars ( $line ) . '</del>' ) .
1252                         $this->emptyLine() . "</tr>\n";
1253                 }
1254         }
1255
1256         function _context( $lines ) {
1257                 foreach ( $lines as $line ) {
1258                         echo '<tr>' .
1259                         $this->contextLine( htmlspecialchars ( $line ) ) .
1260                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1261                 }
1262         }
1263
1264         function _changed( $orig, $closing ) {
1265                 wfProfileIn( __METHOD__ );
1266
1267                 $diff = new WordLevelDiff( $orig, $closing );
1268                 $del = $diff->orig();
1269                 $add = $diff->closing();
1270
1271                 # Notice that WordLevelDiff returns HTML-escaped output.
1272                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1273
1274                 while ( $line = array_shift( $del ) ) {
1275                         $aline = array_shift( $add );
1276                         echo '<tr>' . $this->deletedLine( $line ) .
1277                         $this->addedLine( $aline ) . "</tr>\n";
1278                 }
1279                 foreach ( $add as $line ) {     # If any leftovers
1280                         echo '<tr>' . $this->emptyLine() .
1281                         $this->addedLine( $line ) . "</tr>\n";
1282                 }
1283                 wfProfileOut( __METHOD__ );
1284         }
1285 }