includes/diff/DairikiDiff.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                 }
  48                 $this->orig = $orig;
  49                 $this->closing = $closing;
  50         }
  51
  52         function reverse() {
  53                 return new _DiffOp_Copy( $this->closing, $this->orig );
  54         }
  55 }
  56
  57 /**
  58  * @todo document
  59  * @private
  60  * @ingroup DifferenceEngine
  61  */
  62 class _DiffOp_Delete extends _DiffOp {
  63         var $type = 'delete';
  64
  65         function __construct( $lines ) {
  66                 $this->orig = $lines;
  67                 $this->closing = false;
  68         }
  69
  70         function reverse() {
  71                 return new _DiffOp_Add( $this->orig );
  72         }
  73 }
  74
  75 /**
  76  * @todo document
  77  * @private
  78  * @ingroup DifferenceEngine
  79  */
  80 class _DiffOp_Add extends _DiffOp {
  81         var $type = 'add';
  82
  83         function __construct( $lines ) {
  84                 $this->closing = $lines;
  85                 $this->orig = false;
  86         }
  87
  88         function reverse() {
  89                 return new _DiffOp_Delete( $this->closing );
  90         }
  91 }
  92
  93 /**
  94  * @todo document
  95  * @private
  96  * @ingroup DifferenceEngine
  97  */
  98 class _DiffOp_Change extends _DiffOp {
  99         var $type = 'change';
 100
 101         function __construct( $orig, $closing ) {
 102                 $this->orig = $orig;
 103                 $this->closing = $closing;
 104         }
 105
 106         function reverse() {
 107                 return new _DiffOp_Change( $this->closing, $this->orig );
 108         }
 109 }
 110
 111 /**
 112  * Class used internally by Diff to actually compute the diffs.
 113  *
 114  * The algorithm used here is mostly lifted from the perl module
 115  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 116  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 117  *
 118  * More ideas are taken from:
 119  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 120  *
 121  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 122  * diffutils-2.7, which can be found at:
 123  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 124  *
 125  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 126  * are my own.
 127  *
 128  * Line length limits for robustness added by Tim Starling, 2005-08-31
 129  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
 130  *
 131  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
 132  * @private
 133  * @ingroup DifferenceEngine
 134  */
 135 class _DiffEngine {
 136
 137         const MAX_XREF_LENGTH =  10000;
 138
 139         protected $xchanged, $ychanged;
 140
 141         protected $xv = array(), $yv = array();
 142         protected $xind = array(), $yind = array();
 143
 144         protected $seq = array(), $in_seq = array();
 145
 146         protected $lcs = 0;
 147
 148         function diff ( $from_lines, $to_lines ) {
 149                 wfProfileIn( __METHOD__ );
 150
 151                 // Diff and store locally
 152                 $this->diff_local( $from_lines, $to_lines );
 153
 154                 // Merge edits when possible
 155                 $this->_shift_boundaries( $from_lines, $this->xchanged, $this->ychanged );
 156                 $this->_shift_boundaries( $to_lines, $this->ychanged, $this->xchanged );
 157
 158                 // Compute the edit operations.
 159                 $n_from = sizeof( $from_lines );
 160                 $n_to = sizeof( $to_lines );
 161
 162                 $edits = array();
 163                 $xi = $yi = 0;
 164                 while ( $xi < $n_from || $yi < $n_to ) {
 165                         assert( $yi < $n_to || $this->xchanged[$xi] );
 166                         assert( $xi < $n_from || $this->ychanged[$yi] );
 167
 168                         // Skip matching "snake".
 169                         $copy = array();
 170                         while ( $xi < $n_from && $yi < $n_to
 171                         && !$this->xchanged[$xi] && !$this->ychanged[$yi] ) {
 172                                 $copy[] = $from_lines[$xi++];
 173                                 ++$yi;
 174                         }
 175                         if ( $copy ) {
 176                                 $edits[] = new _DiffOp_Copy( $copy );
 177                         }
 178
 179                         // Find deletes & adds.
 180                         $delete = array();
 181                         while ( $xi < $n_from && $this->xchanged[$xi] ) {
 182                                 $delete[] = $from_lines[$xi++];
 183                         }
 184
 185                         $add = array();
 186                         while ( $yi < $n_to && $this->ychanged[$yi] )  {
 187                                 $add[] = $to_lines[$yi++];
 188                         }
 189
 190                         if ( $delete && $add ) {
 191                                 $edits[] = new _DiffOp_Change( $delete, $add );
 192                         } elseif ( $delete ) {
 193                                 $edits[] = new _DiffOp_Delete( $delete );
 194                         } elseif ( $add ) {
 195                                 $edits[] = new _DiffOp_Add( $add );
 196                         }
 197                 }
 198                 wfProfileOut( __METHOD__ );
 199                 return $edits;
 200         }
 201
 202         function diff_local ( $from_lines, $to_lines ) {
 203                 global $wgExternalDiffEngine;
 204                 wfProfileIn( __METHOD__ );
 205
 206                 if ( $wgExternalDiffEngine == 'wikidiff3' ) {
 207                         // wikidiff3
 208                         $wikidiff3 = new WikiDiff3();
 209                         $wikidiff3->diff( $from_lines, $to_lines );
 210                         $this->xchanged = $wikidiff3->removed;
 211                         $this->ychanged = $wikidiff3->added;
 212                         unset( $wikidiff3 );
 213                 } else {
 214                         // old diff
 215                         $n_from = sizeof( $from_lines );
 216                         $n_to = sizeof( $to_lines );
 217                         $this->xchanged = $this->ychanged = array();
 218                         $this->xv = $this->yv = array();
 219                         $this->xind = $this->yind = array();
 220                         unset( $this->seq );
 221                         unset( $this->in_seq );
 222                         unset( $this->lcs );
 223
 224                         // Skip leading common lines.
 225                         for ( $skip = 0; $skip < $n_from && $skip < $n_to; $skip++ ) {
 226                                 if ( $from_lines[$skip] !== $to_lines[$skip] ) {
 227                                         break;
 228                                 }
 229                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 230                         }
 231                         // Skip trailing common lines.
 232                         $xi = $n_from; $yi = $n_to;
 233                         for ( $endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++ ) {
 234                                 if ( $from_lines[$xi] !== $to_lines[$yi] ) {
 235                                         break;
 236                                 }
 237                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 238                         }
 239
 240                         // Ignore lines which do not exist in both files.
 241                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 242                                 $xhash[$this->_line_hash( $from_lines[$xi] )] = 1;
 243                         }
 244
 245                         for ( $yi = $skip; $yi < $n_to - $endskip; $yi++ ) {
 246                                 $line = $to_lines[$yi];
 247                                 if ( ( $this->ychanged[$yi] = empty( $xhash[$this->_line_hash( $line )] ) ) ) {
 248                                         continue;
 249                                 }
 250                                 $yhash[$this->_line_hash( $line )] = 1;
 251                                 $this->yv[] = $line;
 252                                 $this->yind[] = $yi;
 253                         }
 254                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 255                                 $line = $from_lines[$xi];
 256                                 if ( ( $this->xchanged[$xi] = empty( $yhash[$this->_line_hash( $line )] ) ) ) {
 257                                         continue;
 258                                 }
 259                                 $this->xv[] = $line;
 260                                 $this->xind[] = $xi;
 261                         }
 262
 263                         // Find the LCS.
 264                         $this->_compareseq( 0, sizeof( $this->xv ), 0, sizeof( $this->yv ) );
 265                 }
 266                 wfProfileOut( __METHOD__ );
 267         }
 268
 269         /**
 270          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 271          */
 272         function _line_hash( $line ) {
 273                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 274                         return md5( $line );
 275                 } else {
 276                         return $line;
 277                 }
 278         }
 279
 280         /**
 281          * Divide the Largest Common Subsequence (LCS) of the sequences
 282          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 283          * sized segments.
 284          *
 285          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 286          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 287          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 288          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 289          * that (X0, Y0) == (XOFF, YOFF) and
 290          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 291          *
 292          * This function assumes that the first lines of the specified portions
 293          * of the two files do not match, and likewise that the last lines do not
 294          * match.  The caller must trim matching lines from the beginning and end
 295          * of the portions it is going to specify.
 296          */
 297         function _diag( $xoff, $xlim, $yoff, $ylim, $nchunks ) {
 298                 $flip = false;
 299
 300                 if ( $xlim - $xoff > $ylim - $yoff ) {
 301                         // Things seems faster (I'm not sure I understand why)
 302                         // when the shortest sequence in X.
 303                         $flip = true;
 304                         list( $xoff, $xlim, $yoff, $ylim ) = array( $yoff, $ylim, $xoff, $xlim );
 305                 }
 306
 307                 if ( $flip ) {
 308                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 309                                 $ymatches[$this->xv[$i]][] = $i;
 310                         }
 311                 } else {
 312                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 313                                 $ymatches[$this->yv[$i]][] = $i;
 314                         }
 315                 }
 316
 317                 $this->lcs = 0;
 318                 $this->seq[0] = $yoff - 1;
 319                 $this->in_seq = array();
 320                 $ymids[0] = array();
 321
 322                 $numer = $xlim - $xoff + $nchunks - 1;
 323                 $x = $xoff;
 324                 for ( $chunk = 0; $chunk < $nchunks; $chunk++ ) {
 325                         if ( $chunk > 0 ) {
 326                                 for ( $i = 0; $i <= $this->lcs; $i++ ) {
 327                                         $ymids[$i][$chunk -1] = $this->seq[$i];
 328                                 }
 329                         }
 330
 331                         $x1 = $xoff + (int)( ( $numer + ( $xlim -$xoff ) * $chunk ) / $nchunks );
 332                         for ( ; $x < $x1; $x++ ) {
 333                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 334                                 if ( empty( $ymatches[$line] ) ) {
 335                                         continue;
 336                                 }
 337                                 $matches = $ymatches[$line];
 338                                 reset( $matches );
 339                                 while ( list( , $y ) = each( $matches ) ) {
 340                                         if ( empty( $this->in_seq[$y] ) ) {
 341                                                 $k = $this->_lcs_pos( $y );
 342                                                 assert( $k > 0 );
 343                                                 $ymids[$k] = $ymids[$k -1];
 344                                                 break;
 345                                         }
 346                                 }
 347                                 while ( list ( , $y ) = each( $matches ) ) {
 348                                         if ( $y > $this->seq[$k -1] ) {
 349                                                 assert( $y < $this->seq[$k] );
 350                                                 // Optimization: this is a common case:
 351                                                 //      next match is just replacing previous match.
 352                                                 $this->in_seq[$this->seq[$k]] = false;
 353                                                 $this->seq[$k] = $y;
 354                                                 $this->in_seq[$y] = 1;
 355                                         } elseif ( empty( $this->in_seq[$y] ) ) {
 356                                                 $k = $this->_lcs_pos( $y );
 357                                                 assert( $k > 0 );
 358                                                 $ymids[$k] = $ymids[$k -1];
 359                                         }
 360                                 }
 361                         }
 362                 }
 363
 364                 $seps[] = $flip ? array( $yoff, $xoff ) : array( $xoff, $yoff );
 365                 $ymid = $ymids[$this->lcs];
 366                 for ( $n = 0; $n < $nchunks - 1; $n++ ) {
 367                         $x1 = $xoff + (int)( ( $numer + ( $xlim - $xoff ) * $n ) / $nchunks );
 368                         $y1 = $ymid[$n] + 1;
 369                         $seps[] = $flip ? array( $y1, $x1 ) : array( $x1, $y1 );
 370                 }
 371                 $seps[] = $flip ? array( $ylim, $xlim ) : array( $xlim, $ylim );
 372
 373                 return array( $this->lcs, $seps );
 374         }
 375
 376         function _lcs_pos( $ypos ) {
 377                 $end = $this->lcs;
 378                 if ( $end == 0 || $ypos > $this->seq[$end] ) {
 379                         $this->seq[++$this->lcs] = $ypos;
 380                         $this->in_seq[$ypos] = 1;
 381                         return $this->lcs;
 382                 }
 383
 384                 $beg = 1;
 385                 while ( $beg < $end ) {
 386                         $mid = (int)( ( $beg + $end ) / 2 );
 387                         if ( $ypos > $this->seq[$mid] ) {
 388                                 $beg = $mid + 1;
 389                         } else {
 390                                 $end = $mid;
 391                         }
 392                 }
 393
 394                 assert( $ypos != $this->seq[$end] );
 395
 396                 $this->in_seq[$this->seq[$end]] = false;
 397                 $this->seq[$end] = $ypos;
 398                 $this->in_seq[$ypos] = 1;
 399                 return $end;
 400         }
 401
 402         /**
 403          * Find LCS of two sequences.
 404          *
 405          * The results are recorded in the vectors $this->{x,y}changed[], by
 406          * storing a 1 in the element for each line that is an insertion
 407          * or deletion (ie. is not in the LCS).
 408          *
 409          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 410          *
 411          * Note that XLIM, YLIM are exclusive bounds.
 412          * All line numbers are origin-0 and discarded lines are not counted.
 413          */
 414         function _compareseq ( $xoff, $xlim, $yoff, $ylim ) {
 415                 // Slide down the bottom initial diagonal.
 416                 while ( $xoff < $xlim && $yoff < $ylim
 417                 && $this->xv[$xoff] == $this->yv[$yoff] ) {
 418                         ++$xoff;
 419                         ++$yoff;
 420                 }
 421
 422                 // Slide up the top initial diagonal.
 423                 while ( $xlim > $xoff && $ylim > $yoff
 424                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1] ) {
 425                         --$xlim;
 426                         --$ylim;
 427                 }
 428
 429                 if ( $xoff == $xlim || $yoff == $ylim ) {
 430                         $lcs = 0;
 431                 } else {
 432                         // This is ad hoc but seems to work well.
 433                         // $nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 434                         // $nchunks = max(2,min(8,(int)$nchunks));
 435                         $nchunks = min( 7, $xlim - $xoff, $ylim - $yoff ) + 1;
 436                         list ( $lcs, $seps )
 437                         = $this->_diag( $xoff, $xlim, $yoff, $ylim, $nchunks );
 438                 }
 439
 440                 if ( $lcs == 0 ) {
 441                         // X and Y sequences have no common subsequence:
 442                         // mark all changed.
 443                         while ( $yoff < $ylim ) {
 444                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 445                         }
 446                         while ( $xoff < $xlim ) {
 447                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 448                         }
 449                 } else {
 450                         // Use the partitions to split this problem into subproblems.
 451                         reset( $seps );
 452                         $pt1 = $seps[0];
 453                         while ( $pt2 = next( $seps ) ) {
 454                                 $this->_compareseq ( $pt1[0], $pt2[0], $pt1[1], $pt2[1] );
 455                                 $pt1 = $pt2;
 456                         }
 457                 }
 458         }
 459
 460         /**
 461          * Adjust inserts/deletes of identical lines to join changes
 462          * as much as possible.
 463          *
 464          * We do something when a run of changed lines include a
 465          * line at one end and has an excluded, identical line at the other.
 466          * We are free to choose which identical line is included.
 467          * `compareseq' usually chooses the one at the beginning,
 468          * but usually it is cleaner to consider the following identical line
 469          * to be the "change".
 470          *
 471          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 472          */
 473         function _shift_boundaries( $lines, &$changed, $other_changed ) {
 474                 wfProfileIn( __METHOD__ );
 475                 $i = 0;
 476                 $j = 0;
 477
 478                 assert( 'sizeof($lines) == sizeof($changed)' );
 479                 $len = sizeof( $lines );
 480                 $other_len = sizeof( $other_changed );
 481
 482                 while ( 1 ) {
 483                         /*
 484                          * Scan forwards to find beginning of another run of changes.
 485                          * Also keep track of the corresponding point in the other file.
 486                          *
 487                          * Throughout this code, $i and $j are adjusted together so that
 488                          * the first $i elements of $changed and the first $j elements
 489                          * of $other_changed both contain the same number of zeros
 490                          * (unchanged lines).
 491                          * Furthermore, $j is always kept so that $j == $other_len or
 492                          * $other_changed[$j] == false.
 493                          */
 494                         while ( $j < $other_len && $other_changed[$j] ) {
 495                                 $j++;
 496                         }
 497
 498                         while ( $i < $len && ! $changed[$i] ) {
 499                                 assert( '$j < $other_len && ! $other_changed[$j]' );
 500                                 $i++; $j++;
 501                                 while ( $j < $other_len && $other_changed[$j] )
 502                                 $j++;
 503                         }
 504
 505                         if ( $i == $len ) {
 506                                 break;
 507                         }
 508
 509                         $start = $i;
 510
 511                         // Find the end of this run of changes.
 512                         while ( ++$i < $len && $changed[$i] ) {
 513                                 continue;
 514                         }
 515
 516                         do {
 517                                 /*
 518                                  * Record the length of this run of changes, so that
 519                                  * we can later determine whether the run has grown.
 520                                  */
 521                                 $runlength = $i - $start;
 522
 523                                 /*
 524                                  * Move the changed region back, so long as the
 525                                  * previous unchanged line matches the last changed one.
 526                                  * This merges with previous changed regions.
 527                                  */
 528                                 while ( $start > 0 && $lines[$start - 1] == $lines[$i - 1] ) {
 529                                         $changed[--$start] = 1;
 530                                         $changed[--$i] = false;
 531                                         while ( $start > 0 && $changed[$start - 1] ) {
 532                                                 $start--;
 533                                         }
 534                                         assert( '$j > 0' );
 535                                         while ( $other_changed[--$j] ) {
 536                                                 continue;
 537                                         }
 538                                         assert( '$j >= 0 && !$other_changed[$j]' );
 539                                 }
 540
 541                                 /*
 542                                  * Set CORRESPONDING to the end of the changed run, at the last
 543                                  * point where it corresponds to a changed run in the other file.
 544                                  * CORRESPONDING == LEN means no such point has been found.
 545                                  */
 546                                 $corresponding = $j < $other_len ? $i : $len;
 547
 548                                 /*
 549                                  * Move the changed region forward, so long as the
 550                                  * first changed line matches the following unchanged one.
 551                                  * This merges with following changed regions.
 552                                  * Do this second, so that if there are no merges,
 553                                  * the changed region is moved forward as far as possible.
 554                                  */
 555                                 while ( $i < $len && $lines[$start] == $lines[$i] ) {
 556                                         $changed[$start++] = false;
 557                                         $changed[$i++] = 1;
 558                                         while ( $i < $len && $changed[$i] ) {
 559                                                 $i++;
 560                                         }
 561
 562                                         assert( '$j < $other_len && ! $other_changed[$j]' );
 563                                         $j++;
 564                                         if ( $j < $other_len && $other_changed[$j] ) {
 565                                                 $corresponding = $i;
 566                                                 while ( $j < $other_len && $other_changed[$j] ) {
 567                                                         $j++;
 568                                                 }
 569                                         }
 570                                 }
 571                         } while ( $runlength != $i - $start );
 572
 573                         /*
 574                          * If possible, move the fully-merged run of changes
 575                          * back to a corresponding run in the other file.
 576                          */
 577                         while ( $corresponding < $i ) {
 578                                 $changed[--$start] = 1;
 579                                 $changed[--$i] = 0;
 580                                 assert( '$j > 0' );
 581                                 while ( $other_changed[--$j] ) {
 582                                         continue;
 583                                 }
 584                                 assert( '$j >= 0 && !$other_changed[$j]' );
 585                         }
 586                 }
 587                 wfProfileOut( __METHOD__ );
 588         }
 589 }
 590
 591 /**
 592  * Class representing a 'diff' between two sequences of strings.
 593  * @todo document
 594  * @private
 595  * @ingroup DifferenceEngine
 596  */
 597 class Diff {
 598         var $edits;
 599
 600         /**
 601          * Constructor.
 602          * Computes diff between sequences of strings.
 603          *
 604          * @param $from_lines array An array of strings.
 605          *                (Typically these are lines from a file.)
 606          * @param $to_lines array An array of strings.
 607          */
 608         function __construct( $from_lines, $to_lines ) {
 609                 $eng = new _DiffEngine;
 610                 $this->edits = $eng->diff( $from_lines, $to_lines );
 611                 // $this->_check($from_lines, $to_lines);
 612         }
 613
 614         /**
 615          * Compute reversed Diff.
 616          *
 617          * SYNOPSIS:
 618          *
 619          *      $diff = new Diff($lines1, $lines2);
 620          *      $rev = $diff->reverse();
 621          * @return object A Diff object representing the inverse of the
 622          *                                original diff.
 623          */
 624         function reverse() {
 625                 $rev = $this;
 626                 $rev->edits = array();
 627                 foreach ( $this->edits as $edit ) {
 628                         $rev->edits[] = $edit->reverse();
 629                 }
 630                 return $rev;
 631         }
 632
 633         /**
 634          * Check for empty diff.
 635          *
 636          * @return bool True iff two sequences were identical.
 637          */
 638         function isEmpty() {
 639                 foreach ( $this->edits as $edit ) {
 640                         if ( $edit->type != 'copy' ) {
 641                                 return false;
 642                         }
 643                 }
 644                 return true;
 645         }
 646
 647         /**
 648          * Compute the length of the Longest Common Subsequence (LCS).
 649          *
 650          * This is mostly for diagnostic purposed.
 651          *
 652          * @return int The length of the LCS.
 653          */
 654         function lcs() {
 655                 $lcs = 0;
 656                 foreach ( $this->edits as $edit ) {
 657                         if ( $edit->type == 'copy' ) {
 658                                 $lcs += sizeof( $edit->orig );
 659                         }
 660                 }
 661                 return $lcs;
 662         }
 663
 664         /**
 665          * Get the original set of lines.
 666          *
 667          * This reconstructs the $from_lines parameter passed to the
 668          * constructor.
 669          *
 670          * @return array The original sequence of strings.
 671          */
 672         function orig() {
 673                 $lines = array();
 674
 675                 foreach ( $this->edits as $edit ) {
 676                         if ( $edit->orig ) {
 677                                 array_splice( $lines, sizeof( $lines ), 0, $edit->orig );
 678                         }
 679                 }
 680                 return $lines;
 681         }
 682
 683         /**
 684          * Get the closing set of lines.
 685          *
 686          * This reconstructs the $to_lines parameter passed to the
 687          * constructor.
 688          *
 689          * @return array The sequence of strings.
 690          */
 691         function closing() {
 692                 $lines = array();
 693
 694                 foreach ( $this->edits as $edit ) {
 695                         if ( $edit->closing ) {
 696                                 array_splice( $lines, sizeof( $lines ), 0, $edit->closing );
 697                         }
 698                 }
 699                 return $lines;
 700         }
 701
 702         /**
 703          * Check a Diff for validity.
 704          *
 705          * This is here only for debugging purposes.
 706          */
 707         function _check( $from_lines, $to_lines ) {
 708                 wfProfileIn( __METHOD__ );
 709                 if ( serialize( $from_lines ) != serialize( $this->orig() ) ) {
 710                         trigger_error( "Reconstructed original doesn't match", E_USER_ERROR );
 711                 }
 712                 if ( serialize( $to_lines ) != serialize( $this->closing() ) ) {
 713                         trigger_error( "Reconstructed closing doesn't match", E_USER_ERROR );
 714                 }
 715
 716                 $rev = $this->reverse();
 717                 if ( serialize( $to_lines ) != serialize( $rev->orig() ) ) {
 718                         trigger_error( "Reversed original doesn't match", E_USER_ERROR );
 719                 }
 720                 if ( serialize( $from_lines ) != serialize( $rev->closing() ) ) {
 721                         trigger_error( "Reversed closing doesn't match", E_USER_ERROR );
 722                 }
 723
 724
 725                 $prevtype = 'none';
 726                 foreach ( $this->edits as $edit ) {
 727                         if ( $prevtype == $edit->type ) {
 728                                 trigger_error( 'Edit sequence is non-optimal', E_USER_ERROR );
 729                         }
 730                         $prevtype = $edit->type;
 731                 }
 732
 733                 $lcs = $this->lcs();
 734                 trigger_error( 'Diff okay: LCS = ' . $lcs, E_USER_NOTICE );
 735                 wfProfileOut( __METHOD__ );
 736         }
 737 }
 738
 739 /**
 740  * @todo document, bad name.
 741  * @private
 742  * @ingroup DifferenceEngine
 743  */
 744 class MappedDiff extends Diff {
 745         /**
 746          * Constructor.
 747          *
 748          * Computes diff between sequences of strings.
 749          *
 750          * This can be used to compute things like
 751          * case-insensitve diffs, or diffs which ignore
 752          * changes in white-space.
 753          *
 754          * @param $from_lines array An array of strings.
 755          *      (Typically these are lines from a file.)
 756          *
 757          * @param $to_lines array An array of strings.
 758          *
 759          * @param $mapped_from_lines array This array should
 760          *      have the same size number of elements as $from_lines.
 761          *      The elements in $mapped_from_lines and
 762          *      $mapped_to_lines are what is actually compared
 763          *      when computing the diff.
 764          *
 765          * @param $mapped_to_lines array This array should
 766          *      have the same number of elements as $to_lines.
 767          */
 768         function __construct( $from_lines, $to_lines,
 769                 $mapped_from_lines, $mapped_to_lines ) {
 770                 wfProfileIn( __METHOD__ );
 771
 772                 assert( sizeof( $from_lines ) == sizeof( $mapped_from_lines ) );
 773                 assert( sizeof( $to_lines ) == sizeof( $mapped_to_lines ) );
 774
 775                 parent::__construct( $mapped_from_lines, $mapped_to_lines );
 776
 777                 $xi = $yi = 0;
 778                 for ( $i = 0; $i < sizeof( $this->edits ); $i++ ) {
 779                         $orig = &$this->edits[$i]->orig;
 780                         if ( is_array( $orig ) ) {
 781                                 $orig = array_slice( $from_lines, $xi, sizeof( $orig ) );
 782                                 $xi += sizeof( $orig );
 783                         }
 784
 785                         $closing = &$this->edits[$i]->closing;
 786                         if ( is_array( $closing ) ) {
 787                                 $closing = array_slice( $to_lines, $yi, sizeof( $closing ) );
 788                                 $yi += sizeof( $closing );
 789                         }
 790                 }
 791                 wfProfileOut( __METHOD__ );
 792         }
 793 }
 794
 795 /**
 796  * A class to format Diffs
 797  *
 798  * This class formats the diff in classic diff format.
 799  * It is intended that this class be customized via inheritance,
 800  * to obtain fancier outputs.
 801  * @todo document
 802  * @private
 803  * @ingroup DifferenceEngine
 804  */
 805 class DiffFormatter {
 806         /**
 807          * Number of leading context "lines" to preserve.
 808          *
 809          * This should be left at zero for this class, but subclasses
 810          * may want to set this to other values.
 811          */
 812         var $leading_context_lines = 0;
 813
 814         /**
 815          * Number of trailing context "lines" to preserve.
 816          *
 817          * This should be left at zero for this class, but subclasses
 818          * may want to set this to other values.
 819          */
 820         var $trailing_context_lines = 0;
 821
 822         /**
 823          * Format a diff.
 824          *
 825          * @param $diff Diff A Diff object.
 826          * @return string The formatted output.
 827          */
 828         function format( $diff ) {
 829                 wfProfileIn( __METHOD__ );
 830
 831                 $xi = $yi = 1;
 832                 $block = false;
 833                 $context = array();
 834
 835                 $nlead = $this->leading_context_lines;
 836                 $ntrail = $this->trailing_context_lines;
 837
 838                 $this->_start_diff();
 839
 840                 foreach ( $diff->edits as $edit ) {
 841                         if ( $edit->type == 'copy' ) {
 842                                 if ( is_array( $block ) ) {
 843                                         if ( sizeof( $edit->orig ) <= $nlead + $ntrail ) {
 844                                                 $block[] = $edit;
 845                                         } else {
 846                                                 if ( $ntrail ) {
 847                                                         $context = array_slice( $edit->orig, 0, $ntrail );
 848                                                         $block[] = new _DiffOp_Copy( $context );
 849                                                 }
 850                                                 $this->_block( $x0, $ntrail + $xi - $x0,
 851                                                         $y0, $ntrail + $yi - $y0,
 852                                                         $block );
 853                                                 $block = false;
 854                                         }
 855                                 }
 856                                 $context = $edit->orig;
 857                         } else {
 858                                 if ( !is_array( $block ) ) {
 859                                         $context = array_slice( $context, sizeof( $context ) - $nlead );
 860                                         $x0 = $xi - sizeof( $context );
 861                                         $y0 = $yi - sizeof( $context );
 862                                         $block = array();
 863                                         if ( $context ) {
 864                                                 $block[] = new _DiffOp_Copy( $context );
 865                                         }
 866                                 }
 867                                 $block[] = $edit;
 868                         }
 869
 870                         if ( $edit->orig ) {
 871                                 $xi += sizeof( $edit->orig );
 872                         }
 873                         if ( $edit->closing ) {
 874                                 $yi += sizeof( $edit->closing );
 875                         }
 876                 }
 877
 878                 if ( is_array( $block ) ) {
 879                         $this->_block( $x0, $xi - $x0,
 880                                 $y0, $yi - $y0,
 881                                 $block );
 882                 }
 883
 884                 $end = $this->_end_diff();
 885                 wfProfileOut( __METHOD__ );
 886                 return $end;
 887         }
 888
 889         function _block( $xbeg, $xlen, $ybeg, $ylen, &$edits ) {
 890                 wfProfileIn( __METHOD__ );
 891                 $this->_start_block( $this->_block_header( $xbeg, $xlen, $ybeg, $ylen ) );
 892                 foreach ( $edits as $edit ) {
 893                         if ( $edit->type == 'copy' ) {
 894                                 $this->_context( $edit->orig );
 895                         } elseif ( $edit->type == 'add' ) {
 896                                 $this->_added( $edit->closing );
 897                         } elseif ( $edit->type == 'delete' ) {
 898                                 $this->_deleted( $edit->orig );
 899                         } elseif ( $edit->type == 'change' ) {
 900                                 $this->_changed( $edit->orig, $edit->closing );
 901                         } else {
 902                                 trigger_error( 'Unknown edit type', E_USER_ERROR );
 903                         }
 904                 }
 905                 $this->_end_block();
 906                 wfProfileOut( __METHOD__ );
 907         }
 908
 909         function _start_diff() {
 910                 ob_start();
 911         }
 912
 913         function _end_diff() {
 914                 $val = ob_get_contents();
 915                 ob_end_clean();
 916                 return $val;
 917         }
 918
 919         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
 920                 if ( $xlen > 1 ) {
 921                         $xbeg .= ',' . ( $xbeg + $xlen - 1 );
 922                 }
 923                 if ( $ylen > 1 ) {
 924                         $ybeg .= ',' . ( $ybeg + $ylen - 1 );
 925                 }
 926
 927                 return $xbeg . ( $xlen ? ( $ylen ? 'c' : 'd' ) : 'a' ) . $ybeg;
 928         }
 929
 930         function _start_block( $header ) {
 931                 echo $header . "\n";
 932         }
 933
 934         function _end_block() {
 935         }
 936
 937         function _lines( $lines, $prefix = ' ' ) {
 938                 foreach ( $lines as $line ) {
 939                         echo "$prefix $line\n";
 940                 }
 941         }
 942
 943         function _context( $lines ) {
 944                 $this->_lines( $lines );
 945         }
 946
 947         function _added( $lines ) {
 948                 $this->_lines( $lines, '>' );
 949         }
 950         function _deleted( $lines ) {
 951                 $this->_lines( $lines, '<' );
 952         }
 953
 954         function _changed( $orig, $closing ) {
 955                 $this->_deleted( $orig );
 956                 echo "---\n";
 957                 $this->_added( $closing );
 958         }
 959 }
 960
 961 /**
 962  * A formatter that outputs unified diffs
 963  * @ingroup DifferenceEngine
 964  */
 965 class UnifiedDiffFormatter extends DiffFormatter {
 966         var $leading_context_lines = 2;
 967         var $trailing_context_lines = 2;
 968
 969         function _added( $lines ) {
 970                 $this->_lines( $lines, '+' );
 971         }
 972         function _deleted( $lines ) {
 973                 $this->_lines( $lines, '-' );
 974         }
 975         function _changed( $orig, $closing ) {
 976                 $this->_deleted( $orig );
 977                 $this->_added( $closing );
 978         }
 979         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
 980                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
 981         }
 982 }
 983
 984 /**
 985  * A pseudo-formatter that just passes along the Diff::$edits array
 986  * @ingroup DifferenceEngine
 987  */
 988 class ArrayDiffFormatter extends DiffFormatter {
 989         function format( $diff ) {
 990                 $oldline = 1;
 991                 $newline = 1;
 992                 $retval = array();
 993                 foreach ( $diff->edits as $edit ) {
 994                         switch( $edit->type ) {
 995                                 case 'add':
 996                                         foreach ( $edit->closing as $l ) {
 997                                                 $retval[] = array(
 998                                                         'action' => 'add',
 999                                                         'new' => $l,
1000                                                         'newline' => $newline++
1001                                                 );
1002                                         }
1003                                         break;
1004                                 case 'delete':
1005                                         foreach ( $edit->orig as $l ) {
1006                                                 $retval[] = array(
1007                                                         'action' => 'delete',
1008                                                         'old' => $l,
1009                                                         'oldline' => $oldline++,
1010                                                 );
1011                                         }
1012                                         break;
1013                                 case 'change':
1014                                         foreach ( $edit->orig as $i => $l ) {
1015                                                 $retval[] = array(
1016                                                         'action' => 'change',
1017                                                         'old' => $l,
1018                                                         'new' => isset( $edit->closing[$i] ) ? $edit->closing[$i] : null,
1019                                                         'oldline' => $oldline++,
1020                                                         'newline' => $newline++,
1021                                                 );
1022                                         }
1023                                         break;
1024                                 case 'copy':
1025                                         $oldline += count( $edit->orig );
1026                                         $newline += count( $edit->orig );
1027                         }
1028                 }
1029                 return $retval;
1030         }
1031 }
1032
1033 /**
1034  * Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1035  */
1036
1037 define( 'NBSP', '&#160;' ); // iso-8859-x non-breaking space.
1038
1039 /**
1040  * @todo document
1041  * @private
1042  * @ingroup DifferenceEngine
1043  */
1044 class _HWLDF_WordAccumulator {
1045         function __construct() {
1046                 $this->_lines = array();
1047                 $this->_line = '';
1048                 $this->_group = '';
1049                 $this->_tag = '';
1050         }
1051
1052         function _flushGroup( $new_tag ) {
1053                 if ( $this->_group !== '' ) {
1054                         if ( $this->_tag == 'ins' ) {
1055                                 $this->_line .= '<ins class="diffchange diffchange-inline">' .
1056                                                 htmlspecialchars( $this->_group ) . '</ins>';
1057                         } elseif ( $this->_tag == 'del' ) {
1058                                 $this->_line .= '<del class="diffchange diffchange-inline">' .
1059                                                 htmlspecialchars( $this->_group ) . '</del>';
1060                         } else {
1061                                 $this->_line .= htmlspecialchars( $this->_group );
1062                         }
1063                 }
1064                 $this->_group = '';
1065                 $this->_tag = $new_tag;
1066         }
1067
1068         function _flushLine( $new_tag ) {
1069                 $this->_flushGroup( $new_tag );
1070                 if ( $this->_line != '' ) {
1071                         array_push( $this->_lines, $this->_line );
1072                 } else {
1073                         # make empty lines visible by inserting an NBSP
1074                         array_push( $this->_lines, NBSP );
1075                 }
1076                 $this->_line = '';
1077         }
1078
1079         function addWords ( $words, $tag = '' ) {
1080                 if ( $tag != $this->_tag ) {
1081                         $this->_flushGroup( $tag );
1082                 }
1083
1084                 foreach ( $words as $word ) {
1085                         // new-line should only come as first char of word.
1086                         if ( $word == '' ) {
1087                                 continue;
1088                         }
1089                         if ( $word[0] == "\n" ) {
1090                                 $this->_flushLine( $tag );
1091                                 $word = substr( $word, 1 );
1092                         }
1093                         assert( !strstr( $word, "\n" ) );
1094                         $this->_group .= $word;
1095                 }
1096         }
1097
1098         function getLines() {
1099                 $this->_flushLine( '~done' );
1100                 return $this->_lines;
1101         }
1102 }
1103
1104 /**
1105  * @todo document
1106  * @private
1107  * @ingroup DifferenceEngine
1108  */
1109 class WordLevelDiff extends MappedDiff {
1110         const MAX_LINE_LENGTH = 10000;
1111
1112         function __construct ( $orig_lines, $closing_lines ) {
1113                 wfProfileIn( __METHOD__ );
1114
1115                 list( $orig_words, $orig_stripped ) = $this->_split( $orig_lines );
1116                 list( $closing_words, $closing_stripped ) = $this->_split( $closing_lines );
1117
1118                 parent::__construct( $orig_words, $closing_words,
1119                 $orig_stripped, $closing_stripped );
1120                 wfProfileOut( __METHOD__ );
1121         }
1122
1123         function _split( $lines ) {
1124                 wfProfileIn( __METHOD__ );
1125
1126                 $words = array();
1127                 $stripped = array();
1128                 $first = true;
1129                 foreach ( $lines as $line ) {
1130                         # If the line is too long, just pretend the entire line is one big word
1131                         # This prevents resource exhaustion problems
1132                         if ( $first ) {
1133                                 $first = false;
1134                         } else {
1135                                 $words[] = "\n";
1136                                 $stripped[] = "\n";
1137                         }
1138                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1139                                 $words[] = $line;
1140                                 $stripped[] = $line;
1141                         } else {
1142                                 $m = array();
1143                                 if ( preg_match_all( '/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1144                                         $line, $m ) )
1145                                 {
1146                                         $words = array_merge( $words, $m[0] );
1147                                         $stripped = array_merge( $stripped, $m[1] );
1148                                 }
1149                         }
1150                 }
1151                 wfProfileOut( __METHOD__ );
1152                 return array( $words, $stripped );
1153         }
1154
1155         function orig() {
1156                 wfProfileIn( __METHOD__ );
1157                 $orig = new _HWLDF_WordAccumulator;
1158
1159                 foreach ( $this->edits as $edit ) {
1160                         if ( $edit->type == 'copy' ) {
1161                                 $orig->addWords( $edit->orig );
1162                         } elseif ( $edit->orig ) {
1163                                 $orig->addWords( $edit->orig, 'del' );
1164                         }
1165                 }
1166                 $lines = $orig->getLines();
1167                 wfProfileOut( __METHOD__ );
1168                 return $lines;
1169         }
1170
1171         function closing() {
1172                 wfProfileIn( __METHOD__ );
1173                 $closing = new _HWLDF_WordAccumulator;
1174
1175                 foreach ( $this->edits as $edit ) {
1176                         if ( $edit->type == 'copy' ) {
1177                                 $closing->addWords( $edit->closing );
1178                         } elseif ( $edit->closing ) {
1179                                 $closing->addWords( $edit->closing, 'ins' );
1180                         }
1181                 }
1182                 $lines = $closing->getLines();
1183                 wfProfileOut( __METHOD__ );
1184                 return $lines;
1185         }
1186 }
1187
1188 /**
1189  * Wikipedia Table style diff formatter.
1190  * @todo document
1191  * @private
1192  * @ingroup DifferenceEngine
1193  */
1194 class TableDiffFormatter extends DiffFormatter {
1195         function __construct() {
1196                 $this->leading_context_lines = 2;
1197                 $this->trailing_context_lines = 2;
1198         }
1199
1200         public static function escapeWhiteSpace( $msg ) {
1201                 $msg = preg_replace( '/^ /m', '&#160; ', $msg );
1202                 $msg = preg_replace( '/ $/m', ' &#160;', $msg );
1203                 $msg = preg_replace( '/  /', '&#160; ', $msg );
1204                 return $msg;
1205         }
1206
1207         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1208                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE ' . $xbeg . "--></td>\n" .
1209                   '<td colspan="2" class="diff-lineno"><!--LINE ' . $ybeg . "--></td></tr>\n";
1210                 return $r;
1211         }
1212
1213         function _start_block( $header ) {
1214                 echo $header;
1215         }
1216
1217         function _end_block() {
1218         }
1219
1220         function _lines( $lines, $prefix = ' ', $color = 'white' ) {
1221         }
1222
1223         # HTML-escape parameter before calling this
1224         function addedLine( $line ) {
1225                 return $this->wrapLine( '+', 'diff-addedline', $line );
1226         }
1227
1228         # HTML-escape parameter before calling this
1229         function deletedLine( $line ) {
1230                 return $this->wrapLine( '−', 'diff-deletedline', $line );
1231         }
1232
1233         # HTML-escape parameter before calling this
1234         function contextLine( $line ) {
1235                 return $this->wrapLine( '&#160;', 'diff-context', $line );
1236         }
1237
1238         private function wrapLine( $marker, $class, $line ) {
1239                 if ( $line !== '' ) {
1240                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1241                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1242                 }
1243                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1244         }
1245
1246         function emptyLine() {
1247                 return '<td colspan="2">&#160;</td>';
1248         }
1249
1250         function _added( $lines ) {
1251                 foreach ( $lines as $line ) {
1252                         echo '<tr>' . $this->emptyLine() .
1253                         $this->addedLine( '<ins class="diffchange">' .
1254                         htmlspecialchars( $line ) . '</ins>' ) . "</tr>\n";
1255                 }
1256         }
1257
1258         function _deleted( $lines ) {
1259                 foreach ( $lines as $line ) {
1260                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1261                         htmlspecialchars( $line ) . '</del>' ) .
1262                         $this->emptyLine() . "</tr>\n";
1263                 }
1264         }
1265
1266         function _context( $lines ) {
1267                 foreach ( $lines as $line ) {
1268                         echo '<tr>' .
1269                         $this->contextLine( htmlspecialchars( $line ) ) .
1270                         $this->contextLine( htmlspecialchars( $line ) ) . "</tr>\n";
1271                 }
1272         }
1273
1274         function _changed( $orig, $closing ) {
1275                 wfProfileIn( __METHOD__ );
1276
1277                 $diff = new WordLevelDiff( $orig, $closing );
1278                 $del = $diff->orig();
1279                 $add = $diff->closing();
1280
1281                 # Notice that WordLevelDiff returns HTML-escaped output.
1282                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1283
1284                 while ( $line = array_shift( $del ) ) {
1285                         $aline = array_shift( $add );
1286                         echo '<tr>' . $this->deletedLine( $line ) .
1287                         $this->addedLine( $aline ) . "</tr>\n";
1288                 }
1289                 foreach ( $add as $line ) {     # If any leftovers
1290                         echo '<tr>' . $this->emptyLine() .
1291                         $this->addedLine( $line ) . "</tr>\n";
1292                 }
1293                 wfProfileOut( __METHOD__ );
1294         }
1295 }