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  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup DifferenceEngine
  25  * @defgroup DifferenceEngine DifferenceEngine
  26  */
  27
  28 /**
  29  * @todo document
  30  * @private
  31  * @ingroup DifferenceEngine
  32  */
  33 abstract class DiffOp {
  34         public $type;
  35         public $orig;
  36         public $closing;
  37
  38         public abstract function reverse();
  39
  40         /**
  41          * @return int
  42          */
  43         function norig() {
  44                 return $this->orig ? count( $this->orig ) : 0;
  45         }
  46
  47         /**
  48          * @return int
  49          */
  50         function nclosing() {
  51                 return $this->closing ? count( $this->closing ) : 0;
  52         }
  53 }
  54
  55 /**
  56  * @todo document
  57  * @private
  58  * @ingroup DifferenceEngine
  59  */
  60 class DiffOp_Copy extends DiffOp {
  61         public $type = 'copy';
  62
  63         function __construct( $orig, $closing = false ) {
  64                 if ( !is_array( $closing ) ) {
  65                         $closing = $orig;
  66                 }
  67                 $this->orig = $orig;
  68                 $this->closing = $closing;
  69         }
  70
  71         /**
  72          * @return DiffOp_Copy
  73          */
  74         function reverse() {
  75                 return new DiffOp_Copy( $this->closing, $this->orig );
  76         }
  77 }
  78
  79 /**
  80  * @todo document
  81  * @private
  82  * @ingroup DifferenceEngine
  83  */
  84 class DiffOp_Delete extends DiffOp {
  85         public $type = 'delete';
  86
  87         function __construct( $lines ) {
  88                 $this->orig = $lines;
  89                 $this->closing = false;
  90         }
  91
  92         /**
  93          * @return DiffOp_Add
  94          */
  95         function reverse() {
  96                 return new DiffOp_Add( $this->orig );
  97         }
  98 }
  99
 100 /**
 101  * @todo document
 102  * @private
 103  * @ingroup DifferenceEngine
 104  */
 105 class DiffOp_Add extends DiffOp {
 106         public $type = 'add';
 107
 108         function __construct( $lines ) {
 109                 $this->closing = $lines;
 110                 $this->orig = false;
 111         }
 112
 113         /**
 114          * @return DiffOp_Delete
 115          */
 116         function reverse() {
 117                 return new DiffOp_Delete( $this->closing );
 118         }
 119 }
 120
 121 /**
 122  * @todo document
 123  * @private
 124  * @ingroup DifferenceEngine
 125  */
 126 class DiffOp_Change extends DiffOp {
 127         public $type = 'change';
 128
 129         function __construct( $orig, $closing ) {
 130                 $this->orig = $orig;
 131                 $this->closing = $closing;
 132         }
 133
 134         /**
 135          * @return DiffOp_Change
 136          */
 137         function reverse() {
 138                 return new DiffOp_Change( $this->closing, $this->orig );
 139         }
 140 }
 141
 142 /**
 143  * Class used internally by Diff to actually compute the diffs.
 144  *
 145  * The algorithm used here is mostly lifted from the perl module
 146  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 147  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 148  *
 149  * More ideas are taken from:
 150  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 151  *
 152  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 153  * diffutils-2.7, which can be found at:
 154  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 155  *
 156  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 157  * are my own.
 158  *
 159  * Line length limits for robustness added by Tim Starling, 2005-08-31
 160  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
 161  *
 162  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
 163  * @private
 164  * @ingroup DifferenceEngine
 165  */
 166 class DiffEngine {
 167
 168         const MAX_XREF_LENGTH = 10000;
 169
 170         protected $xchanged, $ychanged;
 171
 172         protected $xv = array(), $yv = array();
 173         protected $xind = array(), $yind = array();
 174
 175         protected $seq = array(), $in_seq = array();
 176
 177         protected $lcs = 0;
 178
 179         /**
 180          * @param $from_lines
 181          * @param $to_lines
 182          * @return array
 183          */
 184         function diff( $from_lines, $to_lines ) {
 185                 wfProfileIn( __METHOD__ );
 186
 187                 // Diff and store locally
 188                 $this->diffLocal( $from_lines, $to_lines );
 189
 190                 // Merge edits when possible
 191                 $this->shiftBoundaries( $from_lines, $this->xchanged, $this->ychanged );
 192                 $this->shiftBoundaries( $to_lines, $this->ychanged, $this->xchanged );
 193
 194                 // Compute the edit operations.
 195                 $n_from = count( $from_lines );
 196                 $n_to = count( $to_lines );
 197
 198                 $edits = array();
 199                 $xi = $yi = 0;
 200                 while ( $xi < $n_from || $yi < $n_to ) {
 201                         assert( '$yi < $n_to || $this->xchanged[$xi]' );
 202                         assert( '$xi < $n_from || $this->ychanged[$yi]' );
 203
 204                         // Skip matching "snake".
 205                         $copy = array();
 206                         while ( $xi < $n_from && $yi < $n_to
 207                         && !$this->xchanged[$xi] && !$this->ychanged[$yi] ) {
 208                                 $copy[] = $from_lines[$xi++];
 209                                 ++$yi;
 210                         }
 211                         if ( $copy ) {
 212                                 $edits[] = new DiffOp_Copy( $copy );
 213                         }
 214
 215                         // Find deletes & adds.
 216                         $delete = array();
 217                         while ( $xi < $n_from && $this->xchanged[$xi] ) {
 218                                 $delete[] = $from_lines[$xi++];
 219                         }
 220
 221                         $add = array();
 222                         while ( $yi < $n_to && $this->ychanged[$yi] ) {
 223                                 $add[] = $to_lines[$yi++];
 224                         }
 225
 226                         if ( $delete && $add ) {
 227                                 $edits[] = new DiffOp_Change( $delete, $add );
 228                         } elseif ( $delete ) {
 229                                 $edits[] = new DiffOp_Delete( $delete );
 230                         } elseif ( $add ) {
 231                                 $edits[] = new DiffOp_Add( $add );
 232                         }
 233                 }
 234                 wfProfileOut( __METHOD__ );
 235                 return $edits;
 236         }
 237
 238         /**
 239          * @param $from_lines
 240          * @param $to_lines
 241          */
 242         private function diffLocal( $from_lines, $to_lines ) {
 243                 global $wgExternalDiffEngine;
 244                 wfProfileIn( __METHOD__ );
 245
 246                 if ( $wgExternalDiffEngine == 'wikidiff3' ) {
 247                         // wikidiff3
 248                         $wikidiff3 = new WikiDiff3();
 249                         $wikidiff3->diff( $from_lines, $to_lines );
 250                         $this->xchanged = $wikidiff3->removed;
 251                         $this->ychanged = $wikidiff3->added;
 252                         unset( $wikidiff3 );
 253                 } else {
 254                         // old diff
 255                         $n_from = count( $from_lines );
 256                         $n_to = count( $to_lines );
 257                         $this->xchanged = $this->ychanged = array();
 258                         $this->xv = $this->yv = array();
 259                         $this->xind = $this->yind = array();
 260                         unset( $this->seq );
 261                         unset( $this->in_seq );
 262                         unset( $this->lcs );
 263
 264                         // Skip leading common lines.
 265                         for ( $skip = 0; $skip < $n_from && $skip < $n_to; $skip++ ) {
 266                                 if ( $from_lines[$skip] !== $to_lines[$skip] ) {
 267                                         break;
 268                                 }
 269                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 270                         }
 271                         // Skip trailing common lines.
 272                         $xi = $n_from;
 273                         $yi = $n_to;
 274                         for ( $endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++ ) {
 275                                 if ( $from_lines[$xi] !== $to_lines[$yi] ) {
 276                                         break;
 277                                 }
 278                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 279                         }
 280
 281                         // Ignore lines which do not exist in both files.
 282                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 283                                 $xhash[$this->lineHash( $from_lines[$xi] )] = 1;
 284                         }
 285
 286                         for ( $yi = $skip; $yi < $n_to - $endskip; $yi++ ) {
 287                                 $line = $to_lines[$yi];
 288                                 if ( ( $this->ychanged[$yi] = empty( $xhash[$this->lineHash( $line )] ) ) ) {
 289                                         continue;
 290                                 }
 291                                 $yhash[$this->lineHash( $line )] = 1;
 292                                 $this->yv[] = $line;
 293                                 $this->yind[] = $yi;
 294                         }
 295                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 296                                 $line = $from_lines[$xi];
 297                                 if ( ( $this->xchanged[$xi] = empty( $yhash[$this->lineHash( $line )] ) ) ) {
 298                                         continue;
 299                                 }
 300                                 $this->xv[] = $line;
 301                                 $this->xind[] = $xi;
 302                         }
 303
 304                         // Find the LCS.
 305                         $this->compareSeq( 0, count( $this->xv ), 0, count( $this->yv ) );
 306                 }
 307                 wfProfileOut( __METHOD__ );
 308         }
 309
 310         /**
 311          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 312          * @param $line string
 313          * @return string
 314          */
 315         private function lineHash( $line ) {
 316                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 317                         return md5( $line );
 318                 } else {
 319                         return $line;
 320                 }
 321         }
 322
 323         /**
 324          * Divide the Largest Common Subsequence (LCS) of the sequences
 325          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 326          * sized segments.
 327          *
 328          * Returns (LCS, PTS). LCS is the length of the LCS. PTS is an
 329          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 330          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 331          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 332          * that (X0, Y0) == (XOFF, YOFF) and
 333          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 334          *
 335          * This function assumes that the first lines of the specified portions
 336          * of the two files do not match, and likewise that the last lines do not
 337          * match.  The caller must trim matching lines from the beginning and end
 338          * of the portions it is going to specify.
 339          * @param $xoff
 340          * @param $xlim
 341          * @param $yoff
 342          * @param $ylim
 343          * @param $nchunks
 344          * @return array
 345          */
 346         private function diag( $xoff, $xlim, $yoff, $ylim, $nchunks ) {
 347                 $flip = false;
 348
 349                 if ( $xlim - $xoff > $ylim - $yoff ) {
 350                         // Things seems faster (I'm not sure I understand why)
 351                         // when the shortest sequence in X.
 352                         $flip = true;
 353                         list( $xoff, $xlim, $yoff, $ylim ) = array( $yoff, $ylim, $xoff, $xlim );
 354                 }
 355
 356                 if ( $flip ) {
 357                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 358                                 $ymatches[$this->xv[$i]][] = $i;
 359                         }
 360                 } else {
 361                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 362                                 $ymatches[$this->yv[$i]][] = $i;
 363                         }
 364                 }
 365
 366                 $this->lcs = 0;
 367                 $this->seq[0] = $yoff - 1;
 368                 $this->in_seq = array();
 369                 $ymids[0] = array();
 370
 371                 $numer = $xlim - $xoff + $nchunks - 1;
 372                 $x = $xoff;
 373                 for ( $chunk = 0; $chunk < $nchunks; $chunk++ ) {
 374                         if ( $chunk > 0 ) {
 375                                 for ( $i = 0; $i <= $this->lcs; $i++ ) {
 376                                         $ymids[$i][$chunk -1] = $this->seq[$i];
 377                                 }
 378                         }
 379
 380                         $x1 = $xoff + (int)( ( $numer + ( $xlim -$xoff ) * $chunk ) / $nchunks );
 381                         for ( ; $x < $x1; $x++ ) {
 382                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 383                                 if ( empty( $ymatches[$line] ) ) {
 384                                         continue;
 385                                 }
 386                                 $matches = $ymatches[$line];
 387                                 reset( $matches );
 388                                 while ( list( , $y ) = each( $matches ) ) {
 389                                         if ( empty( $this->in_seq[$y] ) ) {
 390                                                 $k = $this->lcsPos( $y );
 391                                                 assert( '$k > 0' );
 392                                                 $ymids[$k] = $ymids[$k -1];
 393                                                 break;
 394                                         }
 395                                 }
 396                                 while ( list( , $y ) = each( $matches ) ) {
 397                                         if ( $y > $this->seq[$k -1] ) {
 398                                                 assert( '$y < $this->seq[$k]' );
 399                                                 // Optimization: this is a common case:
 400                                                 //      next match is just replacing previous match.
 401                                                 $this->in_seq[$this->seq[$k]] = false;
 402                                                 $this->seq[$k] = $y;
 403                                                 $this->in_seq[$y] = 1;
 404                                         } elseif ( empty( $this->in_seq[$y] ) ) {
 405                                                 $k = $this->lcsPos( $y );
 406                                                 assert( '$k > 0' );
 407                                                 $ymids[$k] = $ymids[$k -1];
 408                                         }
 409                                 }
 410                         }
 411                 }
 412
 413                 $seps[] = $flip ? array( $yoff, $xoff ) : array( $xoff, $yoff );
 414                 $ymid = $ymids[$this->lcs];
 415                 for ( $n = 0; $n < $nchunks - 1; $n++ ) {
 416                         $x1 = $xoff + (int)( ( $numer + ( $xlim - $xoff ) * $n ) / $nchunks );
 417                         $y1 = $ymid[$n] + 1;
 418                         $seps[] = $flip ? array( $y1, $x1 ) : array( $x1, $y1 );
 419                 }
 420                 $seps[] = $flip ? array( $ylim, $xlim ) : array( $xlim, $ylim );
 421
 422                 return array( $this->lcs, $seps );
 423         }
 424
 425         /**
 426          * @param $ypos
 427          * @return int
 428          */
 429         private function lcsPos( $ypos ) {
 430                 $end = $this->lcs;
 431                 if ( $end == 0 || $ypos > $this->seq[$end] ) {
 432                         $this->seq[++$this->lcs] = $ypos;
 433                         $this->in_seq[$ypos] = 1;
 434                         return $this->lcs;
 435                 }
 436
 437                 $beg = 1;
 438                 while ( $beg < $end ) {
 439                         $mid = (int)( ( $beg + $end ) / 2 );
 440                         if ( $ypos > $this->seq[$mid] ) {
 441                                 $beg = $mid + 1;
 442                         } else {
 443                                 $end = $mid;
 444                         }
 445                 }
 446
 447                 assert( '$ypos != $this->seq[$end]' );
 448
 449                 $this->in_seq[$this->seq[$end]] = false;
 450                 $this->seq[$end] = $ypos;
 451                 $this->in_seq[$ypos] = 1;
 452                 return $end;
 453         }
 454
 455         /**
 456          * Find LCS of two sequences.
 457          *
 458          * The results are recorded in the vectors $this->{x,y}changed[], by
 459          * storing a 1 in the element for each line that is an insertion
 460          * or deletion (ie. is not in the LCS).
 461          *
 462          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 463          *
 464          * Note that XLIM, YLIM are exclusive bounds.
 465          * All line numbers are origin-0 and discarded lines are not counted.
 466          * @param $xoff
 467          * @param $xlim
 468          * @param $yoff
 469          * @param $ylim
 470          */
 471         private function compareSeq( $xoff, $xlim, $yoff, $ylim ) {
 472                 // Slide down the bottom initial diagonal.
 473                 while ( $xoff < $xlim && $yoff < $ylim && $this->xv[$xoff] == $this->yv[$yoff] ) {
 474                         ++$xoff;
 475                         ++$yoff;
 476                 }
 477
 478                 // Slide up the top initial diagonal.
 479                 while ( $xlim > $xoff && $ylim > $yoff
 480                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1] ) {
 481                         --$xlim;
 482                         --$ylim;
 483                 }
 484
 485                 if ( $xoff == $xlim || $yoff == $ylim ) {
 486                         $lcs = 0;
 487                 } else {
 488                         // This is ad hoc but seems to work well.
 489                         // $nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 490                         // $nchunks = max(2,min(8,(int)$nchunks));
 491                         $nchunks = min( 7, $xlim - $xoff, $ylim - $yoff ) + 1;
 492                         list( $lcs, $seps ) = $this->diag( $xoff, $xlim, $yoff, $ylim, $nchunks );
 493                 }
 494
 495                 if ( $lcs == 0 ) {
 496                         // X and Y sequences have no common subsequence:
 497                         // mark all changed.
 498                         while ( $yoff < $ylim ) {
 499                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 500                         }
 501                         while ( $xoff < $xlim ) {
 502                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 503                         }
 504                 } else {
 505                         // Use the partitions to split this problem into subproblems.
 506                         reset( $seps );
 507                         $pt1 = $seps[0];
 508                         while ( $pt2 = next( $seps ) ) {
 509                                 $this->compareSeq( $pt1[0], $pt2[0], $pt1[1], $pt2[1] );
 510                                 $pt1 = $pt2;
 511                         }
 512                 }
 513         }
 514
 515         /**
 516          * Adjust inserts/deletes of identical lines to join changes
 517          * as much as possible.
 518          *
 519          * We do something when a run of changed lines include a
 520          * line at one end and has an excluded, identical line at the other.
 521          * We are free to choose which identical line is included.
 522          * `compareseq' usually chooses the one at the beginning,
 523          * but usually it is cleaner to consider the following identical line
 524          * to be the "change".
 525          *
 526          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 527          */
 528         private function shiftBoundaries( $lines, &$changed, $other_changed ) {
 529                 wfProfileIn( __METHOD__ );
 530                 $i = 0;
 531                 $j = 0;
 532
 533                 assert( 'count($lines) == count($changed)' );
 534                 $len = count( $lines );
 535                 $other_len = count( $other_changed );
 536
 537                 while ( 1 ) {
 538                         /*
 539                          * Scan forwards to find beginning of another run of changes.
 540                          * Also keep track of the corresponding point in the other file.
 541                          *
 542                          * Throughout this code, $i and $j are adjusted together so that
 543                          * the first $i elements of $changed and the first $j elements
 544                          * of $other_changed both contain the same number of zeros
 545                          * (unchanged lines).
 546                          * Furthermore, $j is always kept so that $j == $other_len or
 547                          * $other_changed[$j] == false.
 548                          */
 549                         while ( $j < $other_len && $other_changed[$j] ) {
 550                                 $j++;
 551                         }
 552
 553                         while ( $i < $len && ! $changed[$i] ) {
 554                                 assert( '$j < $other_len && ! $other_changed[$j]' );
 555                                 $i++;
 556                                 $j++;
 557                                 while ( $j < $other_len && $other_changed[$j] ) {
 558                                         $j++;
 559                                 }
 560                         }
 561
 562                         if ( $i == $len ) {
 563                                 break;
 564                         }
 565
 566                         $start = $i;
 567
 568                         // Find the end of this run of changes.
 569                         while ( ++$i < $len && $changed[$i] ) {
 570                                 continue;
 571                         }
 572
 573                         do {
 574                                 /*
 575                                  * Record the length of this run of changes, so that
 576                                  * we can later determine whether the run has grown.
 577                                  */
 578                                 $runlength = $i - $start;
 579
 580                                 /*
 581                                  * Move the changed region back, so long as the
 582                                  * previous unchanged line matches the last changed one.
 583                                  * This merges with previous changed regions.
 584                                  */
 585                                 while ( $start > 0 && $lines[$start - 1] == $lines[$i - 1] ) {
 586                                         $changed[--$start] = 1;
 587                                         $changed[--$i] = false;
 588                                         while ( $start > 0 && $changed[$start - 1] ) {
 589                                                 $start--;
 590                                         }
 591                                         assert( '$j > 0' );
 592                                         while ( $other_changed[--$j] ) {
 593                                                 continue;
 594                                         }
 595                                         assert( '$j >= 0 && !$other_changed[$j]' );
 596                                 }
 597
 598                                 /*
 599                                  * Set CORRESPONDING to the end of the changed run, at the last
 600                                  * point where it corresponds to a changed run in the other file.
 601                                  * CORRESPONDING == LEN means no such point has been found.
 602                                  */
 603                                 $corresponding = $j < $other_len ? $i : $len;
 604
 605                                 /*
 606                                  * Move the changed region forward, so long as the
 607                                  * first changed line matches the following unchanged one.
 608                                  * This merges with following changed regions.
 609                                  * Do this second, so that if there are no merges,
 610                                  * the changed region is moved forward as far as possible.
 611                                  */
 612                                 while ( $i < $len && $lines[$start] == $lines[$i] ) {
 613                                         $changed[$start++] = false;
 614                                         $changed[$i++] = 1;
 615                                         while ( $i < $len && $changed[$i] ) {
 616                                                 $i++;
 617                                         }
 618
 619                                         assert( '$j < $other_len && ! $other_changed[$j]' );
 620                                         $j++;
 621                                         if ( $j < $other_len && $other_changed[$j] ) {
 622                                                 $corresponding = $i;
 623                                                 while ( $j < $other_len && $other_changed[$j] ) {
 624                                                         $j++;
 625                                                 }
 626                                         }
 627                                 }
 628                         } while ( $runlength != $i - $start );
 629
 630                         /*
 631                          * If possible, move the fully-merged run of changes
 632                          * back to a corresponding run in the other file.
 633                          */
 634                         while ( $corresponding < $i ) {
 635                                 $changed[--$start] = 1;
 636                                 $changed[--$i] = 0;
 637                                 assert( '$j > 0' );
 638                                 while ( $other_changed[--$j] ) {
 639                                         continue;
 640                                 }
 641                                 assert( '$j >= 0 && !$other_changed[$j]' );
 642                         }
 643                 }
 644                 wfProfileOut( __METHOD__ );
 645         }
 646 }
 647
 648 /**
 649  * Class representing a 'diff' between two sequences of strings.
 650  * @todo document
 651  * @private
 652  * @ingroup DifferenceEngine
 653  */
 654 class Diff {
 655         public $edits;
 656
 657         /**
 658          * Constructor.
 659          * Computes diff between sequences of strings.
 660          *
 661          * @param $from_lines array An array of strings.
 662          *                (Typically these are lines from a file.)
 663          * @param $to_lines array An array of strings.
 664          */
 665         function __construct( $from_lines, $to_lines ) {
 666                 $eng = new DiffEngine;
 667                 $this->edits = $eng->diff( $from_lines, $to_lines );
 668                 // $this->check($from_lines, $to_lines);
 669         }
 670
 671         /**
 672          * Compute reversed Diff.
 673          *
 674          * SYNOPSIS:
 675          *
 676          *      $diff = new Diff($lines1, $lines2);
 677          *      $rev = $diff->reverse();
 678          * @return Object A Diff object representing the inverse of the
 679          *                                original diff.
 680          */
 681         function reverse() {
 682                 $rev = $this;
 683                 $rev->edits = array();
 684                 /** @var DiffOp $edit */
 685                 foreach ( $this->edits as $edit ) {
 686                         $rev->edits[] = $edit->reverse();
 687                 }
 688                 return $rev;
 689         }
 690
 691         /**
 692          * Check for empty diff.
 693          *
 694          * @return bool True if two sequences were identical.
 695          */
 696         function isEmpty() {
 697                 foreach ( $this->edits as $edit ) {
 698                         if ( $edit->type != 'copy' ) {
 699                                 return false;
 700                         }
 701                 }
 702                 return true;
 703         }
 704
 705         /**
 706          * Compute the length of the Longest Common Subsequence (LCS).
 707          *
 708          * This is mostly for diagnostic purposed.
 709          *
 710          * @return int The length of the LCS.
 711          */
 712         function lcs() {
 713                 $lcs = 0;
 714                 foreach ( $this->edits as $edit ) {
 715                         if ( $edit->type == 'copy' ) {
 716                                 $lcs += count( $edit->orig );
 717                         }
 718                 }
 719                 return $lcs;
 720         }
 721
 722         /**
 723          * Get the original set of lines.
 724          *
 725          * This reconstructs the $from_lines parameter passed to the
 726          * constructor.
 727          *
 728          * @return array The original sequence of strings.
 729          */
 730         function orig() {
 731                 $lines = array();
 732
 733                 foreach ( $this->edits as $edit ) {
 734                         if ( $edit->orig ) {
 735                                 array_splice( $lines, count( $lines ), 0, $edit->orig );
 736                         }
 737                 }
 738                 return $lines;
 739         }
 740
 741         /**
 742          * Get the closing set of lines.
 743          *
 744          * This reconstructs the $to_lines parameter passed to the
 745          * constructor.
 746          *
 747          * @return array The sequence of strings.
 748          */
 749         function closing() {
 750                 $lines = array();
 751
 752                 foreach ( $this->edits as $edit ) {
 753                         if ( $edit->closing ) {
 754                                 array_splice( $lines, count( $lines ), 0, $edit->closing );
 755                         }
 756                 }
 757                 return $lines;
 758         }
 759
 760         /**
 761          * Check a Diff for validity.
 762          *
 763          * This is here only for debugging purposes.
 764          * @param $from_lines
 765          * @param $to_lines
 766          */
 767         private function check( $from_lines, $to_lines ) {
 768                 wfProfileIn( __METHOD__ );
 769                 if ( serialize( $from_lines ) != serialize( $this->orig() ) ) {
 770                         trigger_error( "Reconstructed original doesn't match", E_USER_ERROR );
 771                 }
 772                 if ( serialize( $to_lines ) != serialize( $this->closing() ) ) {
 773                         trigger_error( "Reconstructed closing doesn't match", E_USER_ERROR );
 774                 }
 775
 776                 $rev = $this->reverse();
 777                 if ( serialize( $to_lines ) != serialize( $rev->orig() ) ) {
 778                         trigger_error( "Reversed original doesn't match", E_USER_ERROR );
 779                 }
 780                 if ( serialize( $from_lines ) != serialize( $rev->closing() ) ) {
 781                         trigger_error( "Reversed closing doesn't match", E_USER_ERROR );
 782                 }
 783
 784                 $prevtype = 'none';
 785                 foreach ( $this->edits as $edit ) {
 786                         if ( $prevtype == $edit->type ) {
 787                                 trigger_error( 'Edit sequence is non-optimal', E_USER_ERROR );
 788                         }
 789                         $prevtype = $edit->type;
 790                 }
 791
 792                 $lcs = $this->lcs();
 793                 trigger_error( 'Diff okay: LCS = ' . $lcs, E_USER_NOTICE );
 794                 wfProfileOut( __METHOD__ );
 795         }
 796 }
 797
 798 /**
 799  * @todo document, bad name.
 800  * @private
 801  * @ingroup DifferenceEngine
 802  */
 803 class MappedDiff extends Diff {
 804         /**
 805          * Constructor.
 806          *
 807          * Computes diff between sequences of strings.
 808          *
 809          * This can be used to compute things like
 810          * case-insensitve diffs, or diffs which ignore
 811          * changes in white-space.
 812          *
 813          * @param $from_lines array An array of strings.
 814          *      (Typically these are lines from a file.)
 815          *
 816          * @param $to_lines array An array of strings.
 817          *
 818          * @param $mapped_from_lines array This array should
 819          *      have the same size number of elements as $from_lines.
 820          *      The elements in $mapped_from_lines and
 821          *      $mapped_to_lines are what is actually compared
 822          *      when computing the diff.
 823          *
 824          * @param $mapped_to_lines array This array should
 825          *      have the same number of elements as $to_lines.
 826          */
 827         function __construct( $from_lines, $to_lines,
 828                 $mapped_from_lines, $mapped_to_lines ) {
 829                 wfProfileIn( __METHOD__ );
 830
 831                 assert( 'count( $from_lines ) == count( $mapped_from_lines )' );
 832                 assert( 'count( $to_lines ) == count( $mapped_to_lines )' );
 833
 834                 parent::__construct( $mapped_from_lines, $mapped_to_lines );
 835
 836                 $xi = $yi = 0;
 837                 for ( $i = 0; $i < count( $this->edits ); $i++ ) {
 838                         $orig = &$this->edits[$i]->orig;
 839                         if ( is_array( $orig ) ) {
 840                                 $orig = array_slice( $from_lines, $xi, count( $orig ) );
 841                                 $xi += count( $orig );
 842                         }
 843
 844                         $closing = &$this->edits[$i]->closing;
 845                         if ( is_array( $closing ) ) {
 846                                 $closing = array_slice( $to_lines, $yi, count( $closing ) );
 847                                 $yi += count( $closing );
 848                         }
 849                 }
 850                 wfProfileOut( __METHOD__ );
 851         }
 852 }
 853
 854 /**
 855  * Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
 856  */
 857
 858 /**
 859  * @todo document
 860  * @private
 861  * @ingroup DifferenceEngine
 862  */
 863 class HWLDF_WordAccumulator {
 864         private $lines = array();
 865         private $line = '';
 866         private $group = '';
 867         private $tag = '';
 868
 869         /**
 870          * @param $new_tag
 871          */
 872         private function flushGroup( $new_tag ) {
 873                 if ( $this->group !== '' ) {
 874                         if ( $this->tag == 'ins' ) {
 875                                 $this->line .= '<ins class="diffchange diffchange-inline">' .
 876                                                 htmlspecialchars( $this->group ) . '</ins>';
 877                         } elseif ( $this->tag == 'del' ) {
 878                                 $this->line .= '<del class="diffchange diffchange-inline">' .
 879                                                 htmlspecialchars( $this->group ) . '</del>';
 880                         } else {
 881                                 $this->line .= htmlspecialchars( $this->group );
 882                         }
 883                 }
 884                 $this->group = '';
 885                 $this->tag = $new_tag;
 886         }
 887
 888         /**
 889          * @param $new_tag
 890          */
 891         private function flushLine( $new_tag ) {
 892                 $this->flushGroup( $new_tag );
 893                 if ( $this->line != '' ) {
 894                         array_push( $this->lines, $this->line );
 895                 } else {
 896                         # make empty lines visible by inserting an NBSP
 897                         array_push( $this->lines, '&#160;' );
 898                 }
 899                 $this->line = '';
 900         }
 901
 902         /**
 903          * @param $words
 904          * @param $tag string
 905          */
 906         public function addWords( $words, $tag = '' ) {
 907                 if ( $tag != $this->tag ) {
 908                         $this->flushGroup( $tag );
 909                 }
 910
 911                 foreach ( $words as $word ) {
 912                         // new-line should only come as first char of word.
 913                         if ( $word == '' ) {
 914                                 continue;
 915                         }
 916                         if ( $word[0] == "\n" ) {
 917                                 $this->flushLine( $tag );
 918                                 $word = substr( $word, 1 );
 919                         }
 920                         assert( '!strstr( $word, "\n" )' );
 921                         $this->group .= $word;
 922                 }
 923         }
 924
 925         /**
 926          * @return array
 927          */
 928         public function getLines() {
 929                 $this->flushLine( '~done' );
 930                 return $this->lines;
 931         }
 932 }
 933
 934 /**
 935  * @todo document
 936  * @private
 937  * @ingroup DifferenceEngine
 938  */
 939 class WordLevelDiff extends MappedDiff {
 940         const MAX_LINE_LENGTH = 10000;
 941
 942         /**
 943          * @param $orig_lines
 944          * @param $closing_lines
 945          */
 946         function __construct( $orig_lines, $closing_lines ) {
 947                 wfProfileIn( __METHOD__ );
 948
 949                 list( $orig_words, $orig_stripped ) = $this->split( $orig_lines );
 950                 list( $closing_words, $closing_stripped ) = $this->split( $closing_lines );
 951
 952                 parent::__construct( $orig_words, $closing_words,
 953                 $orig_stripped, $closing_stripped );
 954                 wfProfileOut( __METHOD__ );
 955         }
 956
 957         /**
 958          * @param $lines
 959          * @return array
 960          */
 961         private function split( $lines ) {
 962                 wfProfileIn( __METHOD__ );
 963
 964                 $words = array();
 965                 $stripped = array();
 966                 $first = true;
 967                 foreach ( $lines as $line ) {
 968                         # If the line is too long, just pretend the entire line is one big word
 969                         # This prevents resource exhaustion problems
 970                         if ( $first ) {
 971                                 $first = false;
 972                         } else {
 973                                 $words[] = "\n";
 974                                 $stripped[] = "\n";
 975                         }
 976                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
 977                                 $words[] = $line;
 978                                 $stripped[] = $line;
 979                         } else {
 980                                 $m = array();
 981                                 if ( preg_match_all( '/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
 982                                         $line, $m ) )
 983                                 {
 984                                         foreach ( $m[0] as $word ) {
 985                                                 $words[] = $word;
 986                                         }
 987                                         foreach ( $m[1] as $stripped_word ) {
 988                                                 $stripped[] = $stripped_word;
 989                                         }
 990                                 }
 991                         }
 992                 }
 993                 wfProfileOut( __METHOD__ );
 994                 return array( $words, $stripped );
 995         }
 996
 997         /**
 998          * @return array
 999          */
1000         public function orig() {
1001                 wfProfileIn( __METHOD__ );
1002                 $orig = new HWLDF_WordAccumulator;
1003
1004                 foreach ( $this->edits as $edit ) {
1005                         if ( $edit->type == 'copy' ) {
1006                                 $orig->addWords( $edit->orig );
1007                         } elseif ( $edit->orig ) {
1008                                 $orig->addWords( $edit->orig, 'del' );
1009                         }
1010                 }
1011                 $lines = $orig->getLines();
1012                 wfProfileOut( __METHOD__ );
1013                 return $lines;
1014         }
1015
1016         /**
1017          * @return array
1018          */
1019         public function closing() {
1020                 wfProfileIn( __METHOD__ );
1021                 $closing = new HWLDF_WordAccumulator;
1022
1023                 foreach ( $this->edits as $edit ) {
1024                         if ( $edit->type == 'copy' ) {
1025                                 $closing->addWords( $edit->closing );
1026                         } elseif ( $edit->closing ) {
1027                                 $closing->addWords( $edit->closing, 'ins' );
1028                         }
1029                 }
1030                 $lines = $closing->getLines();
1031                 wfProfileOut( __METHOD__ );
1032                 return $lines;
1033         }
1034 }