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         const MAX_XREF_LENGTH = 10000;
 168
 169         protected $xchanged, $ychanged;
 170
 171         protected $xv = array(), $yv = array();
 172         protected $xind = array(), $yind = array();
 173
 174         protected $seq = array(), $in_seq = array();
 175
 176         protected $lcs = 0;
 177
 178         /**
 179          * @param $from_lines
 180          * @param $to_lines
 181          * @return array
 182          */
 183         function diff( $from_lines, $to_lines ) {
 184                 wfProfileIn( __METHOD__ );
 185
 186                 // Diff and store locally
 187                 $this->diffLocal( $from_lines, $to_lines );
 188
 189                 // Merge edits when possible
 190                 $this->shiftBoundaries( $from_lines, $this->xchanged, $this->ychanged );
 191                 $this->shiftBoundaries( $to_lines, $this->ychanged, $this->xchanged );
 192
 193                 // Compute the edit operations.
 194                 $n_from = count( $from_lines );
 195                 $n_to = count( $to_lines );
 196
 197                 $edits = array();
 198                 $xi = $yi = 0;
 199                 while ( $xi < $n_from || $yi < $n_to ) {
 200                         assert( '$yi < $n_to || $this->xchanged[$xi]' );
 201                         assert( '$xi < $n_from || $this->ychanged[$yi]' );
 202
 203                         // Skip matching "snake".
 204                         $copy = array();
 205                         while ( $xi < $n_from && $yi < $n_to
 206                                 && !$this->xchanged[$xi] && !$this->ychanged[$yi]
 207                         ) {
 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
 236                 return $edits;
 237         }
 238
 239         /**
 240          * @param $from_lines
 241          * @param $to_lines
 242          */
 243         private function diffLocal( $from_lines, $to_lines ) {
 244                 global $wgExternalDiffEngine;
 245                 wfProfileIn( __METHOD__ );
 246
 247                 if ( $wgExternalDiffEngine == 'wikidiff3' ) {
 248                         // wikidiff3
 249                         $wikidiff3 = new WikiDiff3();
 250                         $wikidiff3->diff( $from_lines, $to_lines );
 251                         $this->xchanged = $wikidiff3->removed;
 252                         $this->ychanged = $wikidiff3->added;
 253                         unset( $wikidiff3 );
 254                 } else {
 255                         // old diff
 256                         $n_from = count( $from_lines );
 257                         $n_to = count( $to_lines );
 258                         $this->xchanged = $this->ychanged = array();
 259                         $this->xv = $this->yv = array();
 260                         $this->xind = $this->yind = array();
 261                         unset( $this->seq );
 262                         unset( $this->in_seq );
 263                         unset( $this->lcs );
 264
 265                         // Skip leading common lines.
 266                         for ( $skip = 0; $skip < $n_from && $skip < $n_to; $skip++ ) {
 267                                 if ( $from_lines[$skip] !== $to_lines[$skip] ) {
 268                                         break;
 269                                 }
 270                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 271                         }
 272                         // Skip trailing common lines.
 273                         $xi = $n_from;
 274                         $yi = $n_to;
 275                         for ( $endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++ ) {
 276                                 if ( $from_lines[$xi] !== $to_lines[$yi] ) {
 277                                         break;
 278                                 }
 279                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 280                         }
 281
 282                         // Ignore lines which do not exist in both files.
 283                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 284                                 $xhash[$this->lineHash( $from_lines[$xi] )] = 1;
 285                         }
 286
 287                         for ( $yi = $skip; $yi < $n_to - $endskip; $yi++ ) {
 288                                 $line = $to_lines[$yi];
 289                                 if ( ( $this->ychanged[$yi] = empty( $xhash[$this->lineHash( $line )] ) ) ) {
 290                                         continue;
 291                                 }
 292                                 $yhash[$this->lineHash( $line )] = 1;
 293                                 $this->yv[] = $line;
 294                                 $this->yind[] = $yi;
 295                         }
 296                         for ( $xi = $skip; $xi < $n_from - $endskip; $xi++ ) {
 297                                 $line = $from_lines[$xi];
 298                                 if ( ( $this->xchanged[$xi] = empty( $yhash[$this->lineHash( $line )] ) ) ) {
 299                                         continue;
 300                                 }
 301                                 $this->xv[] = $line;
 302                                 $this->xind[] = $xi;
 303                         }
 304
 305                         // Find the LCS.
 306                         $this->compareSeq( 0, count( $this->xv ), 0, count( $this->yv ) );
 307                 }
 308                 wfProfileOut( __METHOD__ );
 309         }
 310
 311         /**
 312          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 313          * @param $line string
 314          * @return string
 315          */
 316         private function lineHash( $line ) {
 317                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 318                         return md5( $line );
 319                 } else {
 320                         return $line;
 321                 }
 322         }
 323
 324         /**
 325          * Divide the Largest Common Subsequence (LCS) of the sequences
 326          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 327          * sized segments.
 328          *
 329          * Returns (LCS, PTS). LCS is the length of the LCS. PTS is an
 330          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 331          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 332          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 333          * that (X0, Y0) == (XOFF, YOFF) and
 334          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 335          *
 336          * This function assumes that the first lines of the specified portions
 337          * of the two files do not match, and likewise that the last lines do not
 338          * match.  The caller must trim matching lines from the beginning and end
 339          * of the portions it is going to specify.
 340          * @param $xoff
 341          * @param $xlim
 342          * @param $yoff
 343          * @param $ylim
 344          * @param $nchunks
 345          * @return array
 346          */
 347         private function diag( $xoff, $xlim, $yoff, $ylim, $nchunks ) {
 348                 $flip = false;
 349
 350                 if ( $xlim - $xoff > $ylim - $yoff ) {
 351                         // Things seems faster (I'm not sure I understand why)
 352                         // when the shortest sequence in X.
 353                         $flip = true;
 354                         list( $xoff, $xlim, $yoff, $ylim ) = array( $yoff, $ylim, $xoff, $xlim );
 355                 }
 356
 357                 if ( $flip ) {
 358                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 359                                 $ymatches[$this->xv[$i]][] = $i;
 360                         }
 361                 } else {
 362                         for ( $i = $ylim - 1; $i >= $yoff; $i-- ) {
 363                                 $ymatches[$this->yv[$i]][] = $i;
 364                         }
 365                 }
 366
 367                 $this->lcs = 0;
 368                 $this->seq[0] = $yoff - 1;
 369                 $this->in_seq = array();
 370                 $ymids[0] = array();
 371
 372                 $numer = $xlim - $xoff + $nchunks - 1;
 373                 $x = $xoff;
 374                 for ( $chunk = 0; $chunk < $nchunks; $chunk++ ) {
 375                         if ( $chunk > 0 ) {
 376                                 for ( $i = 0; $i <= $this->lcs; $i++ ) {
 377                                         $ymids[$i][$chunk - 1] = $this->seq[$i];
 378                                 }
 379                         }
 380
 381                         $x1 = $xoff + (int)( ( $numer + ( $xlim - $xoff ) * $chunk ) / $nchunks );
 382                         for ( ; $x < $x1; $x++ ) {
 383                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 384                                 if ( empty( $ymatches[$line] ) ) {
 385                                         continue;
 386                                 }
 387                                 $matches = $ymatches[$line];
 388                                 reset( $matches );
 389                                 while ( list( , $y ) = each( $matches ) ) {
 390                                         if ( empty( $this->in_seq[$y] ) ) {
 391                                                 $k = $this->lcsPos( $y );
 392                                                 assert( '$k > 0' );
 393                                                 $ymids[$k] = $ymids[$k - 1];
 394                                                 break;
 395                                         }
 396                                 }
 397                                 while ( list( , $y ) = each( $matches ) ) {
 398                                         if ( $y > $this->seq[$k - 1] ) {
 399                                                 assert( '$y < $this->seq[$k]' );
 400                                                 // Optimization: this is a common case:
 401                                                 //      next match is just replacing previous match.
 402                                                 $this->in_seq[$this->seq[$k]] = false;
 403                                                 $this->seq[$k] = $y;
 404                                                 $this->in_seq[$y] = 1;
 405                                         } elseif ( empty( $this->in_seq[$y] ) ) {
 406                                                 $k = $this->lcsPos( $y );
 407                                                 assert( '$k > 0' );
 408                                                 $ymids[$k] = $ymids[$k - 1];
 409                                         }
 410                                 }
 411                         }
 412                 }
 413
 414                 $seps[] = $flip ? array( $yoff, $xoff ) : array( $xoff, $yoff );
 415                 $ymid = $ymids[$this->lcs];
 416                 for ( $n = 0; $n < $nchunks - 1; $n++ ) {
 417                         $x1 = $xoff + (int)( ( $numer + ( $xlim - $xoff ) * $n ) / $nchunks );
 418                         $y1 = $ymid[$n] + 1;
 419                         $seps[] = $flip ? array( $y1, $x1 ) : array( $x1, $y1 );
 420                 }
 421                 $seps[] = $flip ? array( $ylim, $xlim ) : array( $xlim, $ylim );
 422
 423                 return array( $this->lcs, $seps );
 424         }
 425
 426         /**
 427          * @param $ypos
 428          * @return int
 429          */
 430         private function lcsPos( $ypos ) {
 431                 $end = $this->lcs;
 432                 if ( $end == 0 || $ypos > $this->seq[$end] ) {
 433                         $this->seq[++$this->lcs] = $ypos;
 434                         $this->in_seq[$ypos] = 1;
 435
 436                         return $this->lcs;
 437                 }
 438
 439                 $beg = 1;
 440                 while ( $beg < $end ) {
 441                         $mid = (int)( ( $beg + $end ) / 2 );
 442                         if ( $ypos > $this->seq[$mid] ) {
 443                                 $beg = $mid + 1;
 444                         } else {
 445                                 $end = $mid;
 446                         }
 447                 }
 448
 449                 assert( '$ypos != $this->seq[$end]' );
 450
 451                 $this->in_seq[$this->seq[$end]] = false;
 452                 $this->seq[$end] = $ypos;
 453                 $this->in_seq[$ypos] = 1;
 454
 455                 return $end;
 456         }
 457
 458         /**
 459          * Find LCS of two sequences.
 460          *
 461          * The results are recorded in the vectors $this->{x,y}changed[], by
 462          * storing a 1 in the element for each line that is an insertion
 463          * or deletion (ie. is not in the LCS).
 464          *
 465          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 466          *
 467          * Note that XLIM, YLIM are exclusive bounds.
 468          * All line numbers are origin-0 and discarded lines are not counted.
 469          * @param $xoff
 470          * @param $xlim
 471          * @param $yoff
 472          * @param $ylim
 473          */
 474         private function compareSeq( $xoff, $xlim, $yoff, $ylim ) {
 475                 // Slide down the bottom initial diagonal.
 476                 while ( $xoff < $xlim && $yoff < $ylim && $this->xv[$xoff] == $this->yv[$yoff] ) {
 477                         ++$xoff;
 478                         ++$yoff;
 479                 }
 480
 481                 // Slide up the top initial diagonal.
 482                 while ( $xlim > $xoff && $ylim > $yoff
 483                         && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]
 484                 ) {
 485                         --$xlim;
 486                         --$ylim;
 487                 }
 488
 489                 if ( $xoff == $xlim || $yoff == $ylim ) {
 490                         $lcs = 0;
 491                 } else {
 492                         // This is ad hoc but seems to work well.
 493                         // $nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 494                         // $nchunks = max(2,min(8,(int)$nchunks));
 495                         $nchunks = min( 7, $xlim - $xoff, $ylim - $yoff ) + 1;
 496                         list( $lcs, $seps ) = $this->diag( $xoff, $xlim, $yoff, $ylim, $nchunks );
 497                 }
 498
 499                 if ( $lcs == 0 ) {
 500                         // X and Y sequences have no common subsequence:
 501                         // mark all changed.
 502                         while ( $yoff < $ylim ) {
 503                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 504                         }
 505                         while ( $xoff < $xlim ) {
 506                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 507                         }
 508                 } else {
 509                         // Use the partitions to split this problem into subproblems.
 510                         reset( $seps );
 511                         $pt1 = $seps[0];
 512                         while ( $pt2 = next( $seps ) ) {
 513                                 $this->compareSeq( $pt1[0], $pt2[0], $pt1[1], $pt2[1] );
 514                                 $pt1 = $pt2;
 515                         }
 516                 }
 517         }
 518
 519         /**
 520          * Adjust inserts/deletes of identical lines to join changes
 521          * as much as possible.
 522          *
 523          * We do something when a run of changed lines include a
 524          * line at one end and has an excluded, identical line at the other.
 525          * We are free to choose which identical line is included.
 526          * `compareseq' usually chooses the one at the beginning,
 527          * but usually it is cleaner to consider the following identical line
 528          * to be the "change".
 529          *
 530          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 531          */
 532         private function shiftBoundaries( $lines, &$changed, $other_changed ) {
 533                 wfProfileIn( __METHOD__ );
 534                 $i = 0;
 535                 $j = 0;
 536
 537                 assert( 'count($lines) == count($changed)' );
 538                 $len = count( $lines );
 539                 $other_len = count( $other_changed );
 540
 541                 while ( 1 ) {
 542                         /*
 543                          * Scan forwards to find beginning of another run of changes.
 544                          * Also keep track of the corresponding point in the other file.
 545                          *
 546                          * Throughout this code, $i and $j are adjusted together so that
 547                          * the first $i elements of $changed and the first $j elements
 548                          * of $other_changed both contain the same number of zeros
 549                          * (unchanged lines).
 550                          * Furthermore, $j is always kept so that $j == $other_len or
 551                          * $other_changed[$j] == false.
 552                          */
 553                         while ( $j < $other_len && $other_changed[$j] ) {
 554                                 $j++;
 555                         }
 556
 557                         while ( $i < $len && !$changed[$i] ) {
 558                                 assert( '$j < $other_len && ! $other_changed[$j]' );
 559                                 $i++;
 560                                 $j++;
 561                                 while ( $j < $other_len && $other_changed[$j] ) {
 562                                         $j++;
 563                                 }
 564                         }
 565
 566                         if ( $i == $len ) {
 567                                 break;
 568                         }
 569
 570                         $start = $i;
 571
 572                         // Find the end of this run of changes.
 573                         while ( ++$i < $len && $changed[$i] ) {
 574                                 continue;
 575                         }
 576
 577                         do {
 578                                 /*
 579                                  * Record the length of this run of changes, so that
 580                                  * we can later determine whether the run has grown.
 581                                  */
 582                                 $runlength = $i - $start;
 583
 584                                 /*
 585                                  * Move the changed region back, so long as the
 586                                  * previous unchanged line matches the last changed one.
 587                                  * This merges with previous changed regions.
 588                                  */
 589                                 while ( $start > 0 && $lines[$start - 1] == $lines[$i - 1] ) {
 590                                         $changed[--$start] = 1;
 591                                         $changed[--$i] = false;
 592                                         while ( $start > 0 && $changed[$start - 1] ) {
 593                                                 $start--;
 594                                         }
 595                                         assert( '$j > 0' );
 596                                         while ( $other_changed[--$j] ) {
 597                                                 continue;
 598                                         }
 599                                         assert( '$j >= 0 && !$other_changed[$j]' );
 600                                 }
 601
 602                                 /*
 603                                  * Set CORRESPONDING to the end of the changed run, at the last
 604                                  * point where it corresponds to a changed run in the other file.
 605                                  * CORRESPONDING == LEN means no such point has been found.
 606                                  */
 607                                 $corresponding = $j < $other_len ? $i : $len;
 608
 609                                 /*
 610                                  * Move the changed region forward, so long as the
 611                                  * first changed line matches the following unchanged one.
 612                                  * This merges with following changed regions.
 613                                  * Do this second, so that if there are no merges,
 614                                  * the changed region is moved forward as far as possible.
 615                                  */
 616                                 while ( $i < $len && $lines[$start] == $lines[$i] ) {
 617                                         $changed[$start++] = false;
 618                                         $changed[$i++] = 1;
 619                                         while ( $i < $len && $changed[$i] ) {
 620                                                 $i++;
 621                                         }
 622
 623                                         assert( '$j < $other_len && ! $other_changed[$j]' );
 624                                         $j++;
 625                                         if ( $j < $other_len && $other_changed[$j] ) {
 626                                                 $corresponding = $i;
 627                                                 while ( $j < $other_len && $other_changed[$j] ) {
 628                                                         $j++;
 629                                                 }
 630                                         }
 631                                 }
 632                         } while ( $runlength != $i - $start );
 633
 634                         /*
 635                          * If possible, move the fully-merged run of changes
 636                          * back to a corresponding run in the other file.
 637                          */
 638                         while ( $corresponding < $i ) {
 639                                 $changed[--$start] = 1;
 640                                 $changed[--$i] = 0;
 641                                 assert( '$j > 0' );
 642                                 while ( $other_changed[--$j] ) {
 643                                         continue;
 644                                 }
 645                                 assert( '$j >= 0 && !$other_changed[$j]' );
 646                         }
 647                 }
 648                 wfProfileOut( __METHOD__ );
 649         }
 650 }
 651
 652 /**
 653  * Class representing a 'diff' between two sequences of strings.
 654  * @todo document
 655  * @private
 656  * @ingroup DifferenceEngine
 657  */
 658 class Diff {
 659         public $edits;
 660
 661         /**
 662          * Constructor.
 663          * Computes diff between sequences of strings.
 664          *
 665          * @param $from_lines array An array of strings.
 666          *   Typically these are lines from a file.
 667          * @param $to_lines array An array of strings.
 668          */
 669         function __construct( $from_lines, $to_lines ) {
 670                 $eng = new DiffEngine;
 671                 $this->edits = $eng->diff( $from_lines, $to_lines );
 672                 // $this->check($from_lines, $to_lines);
 673         }
 674
 675         /**
 676          * Compute reversed Diff.
 677          *
 678          * SYNOPSIS:
 679          *
 680          *    $diff = new Diff($lines1, $lines2);
 681          *    $rev = $diff->reverse();
 682          * @return Object A Diff object representing the inverse of the
 683          *   original diff.
 684          */
 685         function reverse() {
 686                 $rev = $this;
 687                 $rev->edits = array();
 688                 /** @var DiffOp $edit */
 689                 foreach ( $this->edits as $edit ) {
 690                         $rev->edits[] = $edit->reverse();
 691                 }
 692
 693                 return $rev;
 694         }
 695
 696         /**
 697          * Check for empty diff.
 698          *
 699          * @return bool True if two sequences were identical.
 700          */
 701         function isEmpty() {
 702                 foreach ( $this->edits as $edit ) {
 703                         if ( $edit->type != 'copy' ) {
 704                                 return false;
 705                         }
 706                 }
 707
 708                 return true;
 709         }
 710
 711         /**
 712          * Compute the length of the Longest Common Subsequence (LCS).
 713          *
 714          * This is mostly for diagnostic purposed.
 715          *
 716          * @return int The length of the LCS.
 717          */
 718         function lcs() {
 719                 $lcs = 0;
 720                 foreach ( $this->edits as $edit ) {
 721                         if ( $edit->type == 'copy' ) {
 722                                 $lcs += count( $edit->orig );
 723                         }
 724                 }
 725
 726                 return $lcs;
 727         }
 728
 729         /**
 730          * Get the original set of lines.
 731          *
 732          * This reconstructs the $from_lines parameter passed to the
 733          * constructor.
 734          *
 735          * @return array The original sequence of strings.
 736          */
 737         function orig() {
 738                 $lines = array();
 739
 740                 foreach ( $this->edits as $edit ) {
 741                         if ( $edit->orig ) {
 742                                 array_splice( $lines, count( $lines ), 0, $edit->orig );
 743                         }
 744                 }
 745
 746                 return $lines;
 747         }
 748
 749         /**
 750          * Get the closing set of lines.
 751          *
 752          * This reconstructs the $to_lines parameter passed to the
 753          * constructor.
 754          *
 755          * @return array The sequence of strings.
 756          */
 757         function closing() {
 758                 $lines = array();
 759
 760                 foreach ( $this->edits as $edit ) {
 761                         if ( $edit->closing ) {
 762                                 array_splice( $lines, count( $lines ), 0, $edit->closing );
 763                         }
 764                 }
 765
 766                 return $lines;
 767         }
 768
 769         /**
 770          * Check a Diff for validity.
 771          *
 772          * This is here only for debugging purposes.
 773          * @param $from_lines
 774          * @param $to_lines
 775          */
 776         private function check( $from_lines, $to_lines ) {
 777                 wfProfileIn( __METHOD__ );
 778                 if ( serialize( $from_lines ) != serialize( $this->orig() ) ) {
 779                         trigger_error( "Reconstructed original doesn't match", E_USER_ERROR );
 780                 }
 781                 if ( serialize( $to_lines ) != serialize( $this->closing() ) ) {
 782                         trigger_error( "Reconstructed closing doesn't match", E_USER_ERROR );
 783                 }
 784
 785                 $rev = $this->reverse();
 786                 if ( serialize( $to_lines ) != serialize( $rev->orig() ) ) {
 787                         trigger_error( "Reversed original doesn't match", E_USER_ERROR );
 788                 }
 789                 if ( serialize( $from_lines ) != serialize( $rev->closing() ) ) {
 790                         trigger_error( "Reversed closing doesn't match", E_USER_ERROR );
 791                 }
 792
 793                 $prevtype = 'none';
 794                 foreach ( $this->edits as $edit ) {
 795                         if ( $prevtype == $edit->type ) {
 796                                 trigger_error( 'Edit sequence is non-optimal', E_USER_ERROR );
 797                         }
 798                         $prevtype = $edit->type;
 799                 }
 800
 801                 $lcs = $this->lcs();
 802                 trigger_error( 'Diff okay: LCS = ' . $lcs, E_USER_NOTICE );
 803                 wfProfileOut( __METHOD__ );
 804         }
 805 }
 806
 807 /**
 808  * @todo document, bad name.
 809  * @private
 810  * @ingroup DifferenceEngine
 811  */
 812 class MappedDiff extends Diff {
 813         /**
 814          * Constructor.
 815          *
 816          * Computes diff between sequences of strings.
 817          *
 818          * This can be used to compute things like
 819          * case-insensitve diffs, or diffs which ignore
 820          * changes in white-space.
 821          *
 822          * @param $from_lines array An array of strings.
 823          *   Typically these are lines from a file.
 824          *
 825          * @param $to_lines array An array of strings.
 826          *
 827          * @param $mapped_from_lines array This array should
 828          *   have the same size number of elements as $from_lines.
 829          *   The elements in $mapped_from_lines and
 830          *   $mapped_to_lines are what is actually compared
 831          *   when computing the diff.
 832          *
 833          * @param $mapped_to_lines array This array should
 834          *   have the same number of elements as $to_lines.
 835          */
 836         function __construct( $from_lines, $to_lines,
 837                 $mapped_from_lines, $mapped_to_lines ) {
 838                 wfProfileIn( __METHOD__ );
 839
 840                 assert( 'count( $from_lines ) == count( $mapped_from_lines )' );
 841                 assert( 'count( $to_lines ) == count( $mapped_to_lines )' );
 842
 843                 parent::__construct( $mapped_from_lines, $mapped_to_lines );
 844
 845                 $xi = $yi = 0;
 846                 for ( $i = 0; $i < count( $this->edits ); $i++ ) {
 847                         $orig = &$this->edits[$i]->orig;
 848                         if ( is_array( $orig ) ) {
 849                                 $orig = array_slice( $from_lines, $xi, count( $orig ) );
 850                                 $xi += count( $orig );
 851                         }
 852
 853                         $closing = &$this->edits[$i]->closing;
 854                         if ( is_array( $closing ) ) {
 855                                 $closing = array_slice( $to_lines, $yi, count( $closing ) );
 856                                 $yi += count( $closing );
 857                         }
 858                 }
 859                 wfProfileOut( __METHOD__ );
 860         }
 861 }
 862
 863 /**
 864  * Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
 865  */
 866
 867 /**
 868  * @todo document
 869  * @private
 870  * @ingroup DifferenceEngine
 871  */
 872 class HWLDF_WordAccumulator {
 873         private $lines = array();
 874         private $line = '';
 875         private $group = '';
 876         private $tag = '';
 877
 878         /**
 879          * @param $new_tag
 880          */
 881         private function flushGroup( $new_tag ) {
 882                 if ( $this->group !== '' ) {
 883                         if ( $this->tag == 'ins' ) {
 884                                 $this->line .= '<ins class="diffchange diffchange-inline">' .
 885                                         htmlspecialchars( $this->group ) . '</ins>';
 886                         } elseif ( $this->tag == 'del' ) {
 887                                 $this->line .= '<del class="diffchange diffchange-inline">' .
 888                                         htmlspecialchars( $this->group ) . '</del>';
 889                         } else {
 890                                 $this->line .= htmlspecialchars( $this->group );
 891                         }
 892                 }
 893                 $this->group = '';
 894                 $this->tag = $new_tag;
 895         }
 896
 897         /**
 898          * @param $new_tag
 899          */
 900         private function flushLine( $new_tag ) {
 901                 $this->flushGroup( $new_tag );
 902                 if ( $this->line != '' ) {
 903                         array_push( $this->lines, $this->line );
 904                 } else {
 905                         # make empty lines visible by inserting an NBSP
 906                         array_push( $this->lines, '&#160;' );
 907                 }
 908                 $this->line = '';
 909         }
 910
 911         /**
 912          * @param $words
 913          * @param $tag string
 914          */
 915         public function addWords( $words, $tag = '' ) {
 916                 if ( $tag != $this->tag ) {
 917                         $this->flushGroup( $tag );
 918                 }
 919
 920                 foreach ( $words as $word ) {
 921                         // new-line should only come as first char of word.
 922                         if ( $word == '' ) {
 923                                 continue;
 924                         }
 925                         if ( $word[0] == "\n" ) {
 926                                 $this->flushLine( $tag );
 927                                 $word = substr( $word, 1 );
 928                         }
 929                         assert( '!strstr( $word, "\n" )' );
 930                         $this->group .= $word;
 931                 }
 932         }
 933
 934         /**
 935          * @return array
 936          */
 937         public function getLines() {
 938                 $this->flushLine( '~done' );
 939
 940                 return $this->lines;
 941         }
 942 }
 943
 944 /**
 945  * @todo document
 946  * @private
 947  * @ingroup DifferenceEngine
 948  */
 949 class WordLevelDiff extends MappedDiff {
 950         const MAX_LINE_LENGTH = 10000;
 951
 952         /**
 953          * @param $orig_lines
 954          * @param $closing_lines
 955          */
 956         function __construct( $orig_lines, $closing_lines ) {
 957                 wfProfileIn( __METHOD__ );
 958
 959                 list( $orig_words, $orig_stripped ) = $this->split( $orig_lines );
 960                 list( $closing_words, $closing_stripped ) = $this->split( $closing_lines );
 961
 962                 parent::__construct( $orig_words, $closing_words,
 963                         $orig_stripped, $closing_stripped );
 964                 wfProfileOut( __METHOD__ );
 965         }
 966
 967         /**
 968          * @param $lines
 969          * @return array
 970          */
 971         private function split( $lines ) {
 972                 wfProfileIn( __METHOD__ );
 973
 974                 $words = array();
 975                 $stripped = array();
 976                 $first = true;
 977                 foreach ( $lines as $line ) {
 978                         # If the line is too long, just pretend the entire line is one big word
 979                         # This prevents resource exhaustion problems
 980                         if ( $first ) {
 981                                 $first = false;
 982                         } else {
 983                                 $words[] = "\n";
 984                                 $stripped[] = "\n";
 985                         }
 986                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
 987                                 $words[] = $line;
 988                                 $stripped[] = $line;
 989                         } else {
 990                                 $m = array();
 991                                 if ( preg_match_all( '/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
 992                                         $line, $m )
 993                                 ) {
 994                                         foreach ( $m[0] as $word ) {
 995                                                 $words[] = $word;
 996                                         }
 997                                         foreach ( $m[1] as $stripped_word ) {
 998                                                 $stripped[] = $stripped_word;
 999                                         }
1000                                 }
1001                         }
1002                 }
1003                 wfProfileOut( __METHOD__ );
1004
1005                 return array( $words, $stripped );
1006         }
1007
1008         /**
1009          * @return array
1010          */
1011         public function orig() {
1012                 wfProfileIn( __METHOD__ );
1013                 $orig = new HWLDF_WordAccumulator;
1014
1015                 foreach ( $this->edits as $edit ) {
1016                         if ( $edit->type == 'copy' ) {
1017                                 $orig->addWords( $edit->orig );
1018                         } elseif ( $edit->orig ) {
1019                                 $orig->addWords( $edit->orig, 'del' );
1020                         }
1021                 }
1022                 $lines = $orig->getLines();
1023                 wfProfileOut( __METHOD__ );
1024
1025                 return $lines;
1026         }
1027
1028         /**
1029          * @return array
1030          */
1031         public function closing() {
1032                 wfProfileIn( __METHOD__ );
1033                 $closing = new HWLDF_WordAccumulator;
1034
1035                 foreach ( $this->edits as $edit ) {
1036                         if ( $edit->type == 'copy' ) {
1037                                 $closing->addWords( $edit->closing );
1038                         } elseif ( $edit->closing ) {
1039                                 $closing->addWords( $edit->closing, 'ins' );
1040                         }
1041                 }
1042                 $lines = $closing->getLines();
1043                 wfProfileOut( __METHOD__ );
1044
1045                 return $lines;
1046         }
1047 }