includes/DifferenceEngine.php

   1 <?
   2 # See diff.doc
   3
   4 class DifferenceEngine {
   5         /* private */ var $mOldid, $mNewid;
   6         /* private */ var $mOldtitle, $mNewtitle;
   7         /* private */ var $mOldtext, $mNewtext;
   8         /* private */ var $mOldUser, $mNewUser;
   9
  10         function DifferenceEngine( $old, $new )
  11         {
  12                 $this->mOldid = $old;
  13                 $this->mNewid = $new;
  14         }
  15
  16         function showDiffPage()
  17         {
  18                 global $wgUser, $wgTitle, $wgOut, $wgLang;
  19
  20                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  21                   "{$this->mNewid})";
  22                 $mtext = wfMsg( "missingarticle", $t );
  23
  24                 $wgOut->setArticleFlag( false );
  25                 if ( ! $this->loadText() ) {
  26                         $wgOut->setPagetitle( wfMsg( "errorpagetitle" ) );
  27                         $wgOut->addHTML( $mtext );
  28                         return;
  29                 }
  30                 $wgOut->supressQuickbar();
  31                 $wgOut->setSubtitle( wfMsg( "difference" ) );
  32                 $wgOut->setRobotpolicy( "noindex,follow" );
  33
  34                 $sk = $wgUser->getSkin();
  35                 $oldUserTitle = Title::makeTitle( NS_USER, $this->mOldUser );
  36                 $newUserTitle = Title::makeTitle( NS_USER, $this->mNewUser );
  37                 $oldUserLink = $sk->makeLinkObj( $oldUserTitle, $this->mOldUser );
  38                 $newUserLink = $sk->makeLinkObj( $newUserTitle, $this->mNewUser );
  39                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br>$oldUserLink";
  40                 $newHeader = "<strong>{$this->mNewtitle}</strong><br>$newUserLink";
  41
  42                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
  43                   $oldHeader, $newHeader );
  44                 $wgOut->addHTML( "<hr><h2>{$this->mNewtitle}</h2>\n" );
  45                 $wgOut->addWikiText( $this->mNewtext );
  46         }
  47
  48         function showDiff( $otext, $ntext, $otitle, $ntitle )
  49         {
  50                 global $wgOut;
  51
  52                 $ota = explode( "\n", str_replace( "\r\n", "\n",
  53                   htmlspecialchars( $otext ) ) );
  54                 $nta = explode( "\n", str_replace( "\r\n", "\n",
  55                   htmlspecialchars( $ntext ) ) );
  56
  57                 $wgOut->addHTML( "<table width='98%' border=0
  58 cellpadding=0 cellspacing='4px'><tr>
  59 <td colspan=2 width='50%' align=center bgcolor='#cccccc'>
  60 {$otitle}</td>
  61 <td colspan=2 width='50%' align=center bgcolor='#cccccc'>
  62 {$ntitle}</td>
  63 </tr>\n" );
  64
  65                 $diffs = new Diff( $ota, $nta );
  66                 $formatter = new TableDiffFormatter();
  67                 $formatter->format( $diffs );
  68                 $wgOut->addHTML( "</table>\n" );
  69         }
  70
  71         # Load the text of the articles to compare.  If newid is 0, then compare
  72         # the old article in oldid to the current article; if oldid is 0, then
  73         # compare the current article to the immediately previous one (ignoring
  74         # the value of newid).
  75         #
  76         function loadText()
  77         {
  78                 global $wgTitle, $wgOut, $wgLang;
  79                 $fname = "DifferenceEngine::loadText";
  80
  81                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
  82                         $wgOut->setArticleFlag( true );
  83                         $this->mNewtitle = wfMsg( "currentrev" );
  84                         $id = $wgTitle->getArticleID();
  85
  86                         $sql = "SELECT cur_text, cur_user_text FROM cur WHERE cur_id={$id}";
  87                         $res = wfQuery( $sql, DB_READ, $fname );
  88                         if ( 0 == wfNumRows( $res ) ) { return false; }
  89
  90                         $s = wfFetchObject( $res );
  91                         $this->mNewtext = $s->cur_text;
  92                         $this->mNewUser = $s->cur_user_text;
  93                 } else {
  94                         $sql = "SELECT old_timestamp,old_text,old_flags,old_user_text FROM old WHERE " .
  95                           "old_id={$this->mNewid}";
  96
  97                         $res = wfQuery( $sql, DB_READ, $fname );
  98                         if ( 0 == wfNumRows( $res ) ) { return false; }
  99
 100                         $s = wfFetchObject( $res );
 101                         $this->mNewtext = Article::getRevisionText( $s );
 102
 103                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 104                         $this->mNewtitle = wfMsg( "revisionasof", $t );
 105                         $this->mNewUser = $s->old_user_text;
 106                 }
 107                 if ( 0 == $this->mOldid ) {
 108                         $sql = "SELECT old_timestamp,old_text,old_flags,old_user_text " .
 109                           "FROM old USE INDEX (name_title_timestamp) WHERE " .
 110                           "old_namespace=" . $wgTitle->getNamespace() . " AND " .
 111                           "old_title='" . wfStrencode( $wgTitle->getDBkey() ) .
 112                           "' ORDER BY inverse_timestamp LIMIT 1";
 113                         $res = wfQuery( $sql, DB_READ, $fname );
 114                 } else {
 115                         $sql = "SELECT old_timestamp,old_text,old_flags,old_user_text FROM old WHERE " .
 116                           "old_id={$this->mOldid}";
 117                         $res = wfQuery( $sql, DB_READ, $fname );
 118                 }
 119                 if ( 0 == wfNumRows( $res ) ) { return false; }
 120
 121                 $s = wfFetchObject( $res );
 122                 $this->mOldtext = Article::getRevisionText( $s );
 123
 124                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 125                 $this->mOldtitle = wfMsg( "revisionasof", $t );
 126                 $this->mOldUser = $s->old_user_text;
 127
 128                 return true;
 129         }
 130 }
 131
 132 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 133 //
 134 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 135 // You may copy this code freely under the conditions of the GPL.
 136 //
 137
 138 define('USE_ASSERTS', function_exists('assert'));
 139
 140 class _DiffOp {
 141     var $type;
 142     var $orig;
 143     var $closing;
 144
 145     function reverse() {
 146         trigger_error("pure virtual", E_USER_ERROR);
 147     }
 148
 149     function norig() {
 150         return $this->orig ? sizeof($this->orig) : 0;
 151     }
 152
 153     function nclosing() {
 154         return $this->closing ? sizeof($this->closing) : 0;
 155     }
 156 }
 157
 158 class _DiffOp_Copy extends _DiffOp {
 159     var $type = 'copy';
 160
 161     function _DiffOp_Copy ($orig, $closing = false) {
 162         if (!is_array($closing))
 163             $closing = $orig;
 164         $this->orig = $orig;
 165         $this->closing = $closing;
 166     }
 167
 168     function reverse() {
 169         return new _DiffOp_Copy($this->closing, $this->orig);
 170     }
 171 }
 172
 173 class _DiffOp_Delete extends _DiffOp {
 174     var $type = 'delete';
 175
 176     function _DiffOp_Delete ($lines) {
 177         $this->orig = $lines;
 178         $this->closing = false;
 179     }
 180
 181     function reverse() {
 182         return new _DiffOp_Add($this->orig);
 183     }
 184 }
 185
 186 class _DiffOp_Add extends _DiffOp {
 187     var $type = 'add';
 188
 189     function _DiffOp_Add ($lines) {
 190         $this->closing = $lines;
 191         $this->orig = false;
 192     }
 193
 194     function reverse() {
 195         return new _DiffOp_Delete($this->closing);
 196     }
 197 }
 198
 199 class _DiffOp_Change extends _DiffOp {
 200     var $type = 'change';
 201
 202     function _DiffOp_Change ($orig, $closing) {
 203         $this->orig = $orig;
 204         $this->closing = $closing;
 205     }
 206
 207     function reverse() {
 208         return new _DiffOp_Change($this->closing, $this->orig);
 209     }
 210 }
 211
 212
 213 /**
 214  * Class used internally by Diff to actually compute the diffs.
 215  *
 216  * The algorithm used here is mostly lifted from the perl module
 217  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 218  *   http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 219  *
 220  * More ideas are taken from:
 221  *   http://www.ics.uci.edu/~eppstein/161/960229.html
 222  *
 223  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 224  * diffutils-2.7, which can be found at:
 225  *   ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 226  *
 227  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 228  * are my own.
 229  *
 230  * @author Geoffrey T. Dairiki
 231  * @access private
 232  */
 233 class _DiffEngine
 234 {
 235     function diff ($from_lines, $to_lines) {
 236         $n_from = sizeof($from_lines);
 237         $n_to = sizeof($to_lines);
 238
 239         $this->xchanged = $this->ychanged = array();
 240         $this->xv = $this->yv = array();
 241         $this->xind = $this->yind = array();
 242         unset($this->seq);
 243         unset($this->in_seq);
 244         unset($this->lcs);
 245
 246         // Skip leading common lines.
 247         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 248             if ($from_lines[$skip] != $to_lines[$skip])
 249                 break;
 250             $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 251         }
 252         // Skip trailing common lines.
 253         $xi = $n_from; $yi = $n_to;
 254         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 255             if ($from_lines[$xi] != $to_lines[$yi])
 256                 break;
 257             $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 258         }
 259
 260         // Ignore lines which do not exist in both files.
 261         for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 262             $xhash[$from_lines[$xi]] = 1;
 263         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 264             $line = $to_lines[$yi];
 265             if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 266                 continue;
 267             $yhash[$line] = 1;
 268             $this->yv[] = $line;
 269             $this->yind[] = $yi;
 270         }
 271         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 272             $line = $from_lines[$xi];
 273             if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 274                 continue;
 275             $this->xv[] = $line;
 276             $this->xind[] = $xi;
 277         }
 278
 279         // Find the LCS.
 280         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 281
 282         // Merge edits when possible
 283         $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 284         $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 285
 286         // Compute the edit operations.
 287         $edits = array();
 288         $xi = $yi = 0;
 289         while ($xi < $n_from || $yi < $n_to) {
 290             USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 291             USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 292
 293             // Skip matching "snake".
 294             $copy = array();
 295             while ( $xi < $n_from && $yi < $n_to
 296                     && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 297                 $copy[] = $from_lines[$xi++];
 298                 ++$yi;
 299             }
 300             if ($copy)
 301                 $edits[] = new _DiffOp_Copy($copy);
 302
 303             // Find deletes & adds.
 304             $delete = array();
 305             while ($xi < $n_from && $this->xchanged[$xi])
 306                 $delete[] = $from_lines[$xi++];
 307
 308             $add = array();
 309             while ($yi < $n_to && $this->ychanged[$yi])
 310                 $add[] = $to_lines[$yi++];
 311
 312             if ($delete && $add)
 313                 $edits[] = new _DiffOp_Change($delete, $add);
 314             elseif ($delete)
 315                 $edits[] = new _DiffOp_Delete($delete);
 316             elseif ($add)
 317                 $edits[] = new _DiffOp_Add($add);
 318         }
 319         return $edits;
 320     }
 321
 322
 323     /* Divide the Largest Common Subsequence (LCS) of the sequences
 324      * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 325      * sized segments.
 326      *
 327      * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 328      * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 329      * sub sequences.  The first sub-sequence is contained in [X0, X1),
 330      * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 331      * that (X0, Y0) == (XOFF, YOFF) and
 332      * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 333      *
 334      * This function assumes that the first lines of the specified portions
 335      * of the two files do not match, and likewise that the last lines do not
 336      * match.  The caller must trim matching lines from the beginning and end
 337      * of the portions it is going to specify.
 338      */
 339     function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 340         $flip = false;
 341
 342         if ($xlim - $xoff > $ylim - $yoff) {
 343             // Things seems faster (I'm not sure I understand why)
 344             // when the shortest sequence in X.
 345             $flip = true;
 346             list ($xoff, $xlim, $yoff, $ylim)
 347                 = array( $yoff, $ylim, $xoff, $xlim);
 348         }
 349
 350         if ($flip)
 351             for ($i = $ylim - 1; $i >= $yoff; $i--)
 352                 $ymatches[$this->xv[$i]][] = $i;
 353         else
 354             for ($i = $ylim - 1; $i >= $yoff; $i--)
 355                 $ymatches[$this->yv[$i]][] = $i;
 356
 357         $this->lcs = 0;
 358         $this->seq[0]= $yoff - 1;
 359         $this->in_seq = array();
 360         $ymids[0] = array();
 361
 362         $numer = $xlim - $xoff + $nchunks - 1;
 363         $x = $xoff;
 364         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 365             if ($chunk > 0)
 366                 for ($i = 0; $i <= $this->lcs; $i++)
 367                     $ymids[$i][$chunk-1] = $this->seq[$i];
 368
 369             $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 370             for ( ; $x < $x1; $x++) {
 371                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 372                 if (empty($ymatches[$line]))
 373                     continue;
 374                 $matches = $ymatches[$line];
 375                 reset($matches);
 376                 while (list ($junk, $y) = each($matches))
 377                     if (empty($this->in_seq[$y])) {
 378                         $k = $this->_lcs_pos($y);
 379                         USE_ASSERTS && assert($k > 0);
 380                         $ymids[$k] = $ymids[$k-1];
 381                         break;
 382                     }
 383                 while (list ($junk, $y) = each($matches)) {
 384                     if ($y > $this->seq[$k-1]) {
 385                         USE_ASSERTS && assert($y < $this->seq[$k]);
 386                         // Optimization: this is a common case:
 387                         //  next match is just replacing previous match.
 388                         $this->in_seq[$this->seq[$k]] = false;
 389                         $this->seq[$k] = $y;
 390                         $this->in_seq[$y] = 1;
 391                     }
 392                     else if (empty($this->in_seq[$y])) {
 393                         $k = $this->_lcs_pos($y);
 394                         USE_ASSERTS && assert($k > 0);
 395                         $ymids[$k] = $ymids[$k-1];
 396                     }
 397                 }
 398             }
 399         }
 400
 401         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 402         $ymid = $ymids[$this->lcs];
 403         for ($n = 0; $n < $nchunks - 1; $n++) {
 404             $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 405             $y1 = $ymid[$n] + 1;
 406             $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 407         }
 408         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 409
 410         return array($this->lcs, $seps);
 411     }
 412
 413     function _lcs_pos ($ypos) {
 414         $end = $this->lcs;
 415         if ($end == 0 || $ypos > $this->seq[$end]) {
 416             $this->seq[++$this->lcs] = $ypos;
 417             $this->in_seq[$ypos] = 1;
 418             return $this->lcs;
 419         }
 420
 421         $beg = 1;
 422         while ($beg < $end) {
 423             $mid = (int)(($beg + $end) / 2);
 424             if ( $ypos > $this->seq[$mid] )
 425                 $beg = $mid + 1;
 426             else
 427                 $end = $mid;
 428         }
 429
 430         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 431
 432         $this->in_seq[$this->seq[$end]] = false;
 433         $this->seq[$end] = $ypos;
 434         $this->in_seq[$ypos] = 1;
 435         return $end;
 436     }
 437
 438     /* Find LCS of two sequences.
 439      *
 440      * The results are recorded in the vectors $this->{x,y}changed[], by
 441      * storing a 1 in the element for each line that is an insertion
 442      * or deletion (ie. is not in the LCS).
 443      *
 444      * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 445      *
 446      * Note that XLIM, YLIM are exclusive bounds.
 447      * All line numbers are origin-0 and discarded lines are not counted.
 448      */
 449     function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 450         // Slide down the bottom initial diagonal.
 451         while ($xoff < $xlim && $yoff < $ylim
 452                && $this->xv[$xoff] == $this->yv[$yoff]) {
 453             ++$xoff;
 454             ++$yoff;
 455         }
 456
 457         // Slide up the top initial diagonal.
 458         while ($xlim > $xoff && $ylim > $yoff
 459                && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 460             --$xlim;
 461             --$ylim;
 462         }
 463
 464         if ($xoff == $xlim || $yoff == $ylim)
 465             $lcs = 0;
 466         else {
 467             // This is ad hoc but seems to work well.
 468             //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 469             //$nchunks = max(2,min(8,(int)$nchunks));
 470             $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 471             list ($lcs, $seps)
 472                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 473         }
 474
 475         if ($lcs == 0) {
 476             // X and Y sequences have no common subsequence:
 477             // mark all changed.
 478             while ($yoff < $ylim)
 479                 $this->ychanged[$this->yind[$yoff++]] = 1;
 480             while ($xoff < $xlim)
 481                 $this->xchanged[$this->xind[$xoff++]] = 1;
 482         }
 483         else {
 484             // Use the partitions to split this problem into subproblems.
 485             reset($seps);
 486             $pt1 = $seps[0];
 487             while ($pt2 = next($seps)) {
 488                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 489                 $pt1 = $pt2;
 490             }
 491         }
 492     }
 493
 494     /* Adjust inserts/deletes of identical lines to join changes
 495      * as much as possible.
 496      *
 497      * We do something when a run of changed lines include a
 498      * line at one end and has an excluded, identical line at the other.
 499      * We are free to choose which identical line is included.
 500      * `compareseq' usually chooses the one at the beginning,
 501      * but usually it is cleaner to consider the following identical line
 502      * to be the "change".
 503      *
 504      * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 505      */
 506     function _shift_boundaries ($lines, &$changed, $other_changed) {
 507         $i = 0;
 508         $j = 0;
 509
 510         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 511         $len = sizeof($lines);
 512         $other_len = sizeof($other_changed);
 513
 514         while (1) {
 515             /*
 516              * Scan forwards to find beginning of another run of changes.
 517              * Also keep track of the corresponding point in the other file.
 518              *
 519              * Throughout this code, $i and $j are adjusted together so that
 520              * the first $i elements of $changed and the first $j elements
 521              * of $other_changed both contain the same number of zeros
 522              * (unchanged lines).
 523              * Furthermore, $j is always kept so that $j == $other_len or
 524              * $other_changed[$j] == false.
 525              */
 526             while ($j < $other_len && $other_changed[$j])
 527                 $j++;
 528
 529             while ($i < $len && ! $changed[$i]) {
 530                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 531                 $i++; $j++;
 532                 while ($j < $other_len && $other_changed[$j])
 533                     $j++;
 534             }
 535
 536             if ($i == $len)
 537                 break;
 538
 539             $start = $i;
 540
 541             // Find the end of this run of changes.
 542             while (++$i < $len && $changed[$i])
 543                 continue;
 544
 545             do {
 546                 /*
 547                  * Record the length of this run of changes, so that
 548                  * we can later determine whether the run has grown.
 549                  */
 550                 $runlength = $i - $start;
 551
 552                 /*
 553                  * Move the changed region back, so long as the
 554                  * previous unchanged line matches the last changed one.
 555                  * This merges with previous changed regions.
 556                  */
 557                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 558                     $changed[--$start] = 1;
 559                     $changed[--$i] = false;
 560                     while ($start > 0 && $changed[$start - 1])
 561                         $start--;
 562                     USE_ASSERTS && assert('$j > 0');
 563                     while ($other_changed[--$j])
 564                         continue;
 565                     USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 566                 }
 567
 568                 /*
 569                  * Set CORRESPONDING to the end of the changed run, at the last
 570                  * point where it corresponds to a changed run in the other file.
 571                  * CORRESPONDING == LEN means no such point has been found.
 572                  */
 573                 $corresponding = $j < $other_len ? $i : $len;
 574
 575                 /*
 576                  * Move the changed region forward, so long as the
 577                  * first changed line matches the following unchanged one.
 578                  * This merges with following changed regions.
 579                  * Do this second, so that if there are no merges,
 580                  * the changed region is moved forward as far as possible.
 581                  */
 582                 while ($i < $len && $lines[$start] == $lines[$i]) {
 583                     $changed[$start++] = false;
 584                     $changed[$i++] = 1;
 585                     while ($i < $len && $changed[$i])
 586                         $i++;
 587
 588                     USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 589                     $j++;
 590                     if ($j < $other_len && $other_changed[$j]) {
 591                         $corresponding = $i;
 592                         while ($j < $other_len && $other_changed[$j])
 593                             $j++;
 594                     }
 595                 }
 596             } while ($runlength != $i - $start);
 597
 598             /*
 599              * If possible, move the fully-merged run of changes
 600              * back to a corresponding run in the other file.
 601              */
 602             while ($corresponding < $i) {
 603                 $changed[--$start] = 1;
 604                 $changed[--$i] = 0;
 605                 USE_ASSERTS && assert('$j > 0');
 606                 while ($other_changed[--$j])
 607                     continue;
 608                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 609             }
 610         }
 611     }
 612 }
 613
 614 /**
 615  * Class representing a 'diff' between two sequences of strings.
 616  */
 617 class Diff
 618 {
 619     var $edits;
 620
 621     /**
 622      * Constructor.
 623      * Computes diff between sequences of strings.
 624      *
 625      * @param $from_lines array An array of strings.
 626      *        (Typically these are lines from a file.)
 627      * @param $to_lines array An array of strings.
 628      */
 629     function Diff($from_lines, $to_lines) {
 630         $eng = new _DiffEngine;
 631         $this->edits = $eng->diff($from_lines, $to_lines);
 632         //$this->_check($from_lines, $to_lines);
 633     }
 634
 635     /**
 636      * Compute reversed Diff.
 637      *
 638      * SYNOPSIS:
 639      *
 640      *  $diff = new Diff($lines1, $lines2);
 641      *  $rev = $diff->reverse();
 642      * @return object A Diff object representing the inverse of the
 643      *                original diff.
 644      */
 645     function reverse () {
 646         $rev = $this;
 647         $rev->edits = array();
 648         foreach ($this->edits as $edit) {
 649             $rev->edits[] = $edit->reverse();
 650         }
 651         return $rev;
 652     }
 653
 654     /**
 655      * Check for empty diff.
 656      *
 657      * @return bool True iff two sequences were identical.
 658      */
 659     function isEmpty () {
 660         foreach ($this->edits as $edit) {
 661             if ($edit->type != 'copy')
 662                 return false;
 663         }
 664         return true;
 665     }
 666
 667     /**
 668      * Compute the length of the Longest Common Subsequence (LCS).
 669      *
 670      * This is mostly for diagnostic purposed.
 671      *
 672      * @return int The length of the LCS.
 673      */
 674     function lcs () {
 675         $lcs = 0;
 676         foreach ($this->edits as $edit) {
 677             if ($edit->type == 'copy')
 678                 $lcs += sizeof($edit->orig);
 679         }
 680         return $lcs;
 681     }
 682
 683     /**
 684      * Get the original set of lines.
 685      *
 686      * This reconstructs the $from_lines parameter passed to the
 687      * constructor.
 688      *
 689      * @return array The original sequence of strings.
 690      */
 691     function orig() {
 692         $lines = array();
 693
 694         foreach ($this->edits as $edit) {
 695             if ($edit->orig)
 696                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 697         }
 698         return $lines;
 699     }
 700
 701     /**
 702      * Get the closing set of lines.
 703      *
 704      * This reconstructs the $to_lines parameter passed to the
 705      * constructor.
 706      *
 707      * @return array The sequence of strings.
 708      */
 709     function closing() {
 710         $lines = array();
 711
 712         foreach ($this->edits as $edit) {
 713             if ($edit->closing)
 714                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 715         }
 716         return $lines;
 717     }
 718
 719     /**
 720      * Check a Diff for validity.
 721      *
 722      * This is here only for debugging purposes.
 723      */
 724     function _check ($from_lines, $to_lines) {
 725         if (serialize($from_lines) != serialize($this->orig()))
 726             trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 727         if (serialize($to_lines) != serialize($this->closing()))
 728             trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 729
 730         $rev = $this->reverse();
 731         if (serialize($to_lines) != serialize($rev->orig()))
 732             trigger_error("Reversed original doesn't match", E_USER_ERROR);
 733         if (serialize($from_lines) != serialize($rev->closing()))
 734             trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 735
 736
 737         $prevtype = 'none';
 738         foreach ($this->edits as $edit) {
 739             if ( $prevtype == $edit->type )
 740                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 741             $prevtype = $edit->type;
 742         }
 743
 744         $lcs = $this->lcs();
 745         trigger_error("Diff okay: LCS = $lcs", E_USER_NOTICE);
 746     }
 747 }
 748
 749 /**
 750  * FIXME: bad name.
 751  */
 752 class MappedDiff
 753 extends Diff
 754 {
 755     /**
 756      * Constructor.
 757      *
 758      * Computes diff between sequences of strings.
 759      *
 760      * This can be used to compute things like
 761      * case-insensitve diffs, or diffs which ignore
 762      * changes in white-space.
 763      *
 764      * @param $from_lines array An array of strings.
 765      *  (Typically these are lines from a file.)
 766      *
 767      * @param $to_lines array An array of strings.
 768      *
 769      * @param $mapped_from_lines array This array should
 770      *  have the same size number of elements as $from_lines.
 771      *  The elements in $mapped_from_lines and
 772      *  $mapped_to_lines are what is actually compared
 773      *  when computing the diff.
 774      *
 775      * @param $mapped_to_lines array This array should
 776      *  have the same number of elements as $to_lines.
 777      */
 778     function MappedDiff($from_lines, $to_lines,
 779                         $mapped_from_lines, $mapped_to_lines) {
 780
 781         assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 782         assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 783
 784         $this->Diff($mapped_from_lines, $mapped_to_lines);
 785
 786         $xi = $yi = 0;
 787         for ($i = 0; $i < sizeof($this->edits); $i++) {
 788             $orig = &$this->edits[$i]->orig;
 789             if (is_array($orig)) {
 790                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 791                 $xi += sizeof($orig);
 792             }
 793
 794             $closing = &$this->edits[$i]->closing;
 795             if (is_array($closing)) {
 796                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 797                 $yi += sizeof($closing);
 798             }
 799         }
 800     }
 801 }
 802
 803 /**
 804  * A class to format Diffs
 805  *
 806  * This class formats the diff in classic diff format.
 807  * It is intended that this class be customized via inheritance,
 808  * to obtain fancier outputs.
 809  */
 810 class DiffFormatter
 811 {
 812     /**
 813      * Number of leading context "lines" to preserve.
 814      *
 815      * This should be left at zero for this class, but subclasses
 816      * may want to set this to other values.
 817      */
 818     var $leading_context_lines = 0;
 819
 820     /**
 821      * Number of trailing context "lines" to preserve.
 822      *
 823      * This should be left at zero for this class, but subclasses
 824      * may want to set this to other values.
 825      */
 826     var $trailing_context_lines = 0;
 827
 828     /**
 829      * Format a diff.
 830      *
 831      * @param $diff object A Diff object.
 832      * @return string The formatted output.
 833      */
 834     function format($diff) {
 835
 836         $xi = $yi = 1;
 837         $block = false;
 838         $context = array();
 839
 840         $nlead = $this->leading_context_lines;
 841         $ntrail = $this->trailing_context_lines;
 842
 843         $this->_start_diff();
 844
 845         foreach ($diff->edits as $edit) {
 846             if ($edit->type == 'copy') {
 847                 if (is_array($block)) {
 848                     if (sizeof($edit->orig) <= $nlead + $ntrail) {
 849                         $block[] = $edit;
 850                     }
 851                     else{
 852                         if ($ntrail) {
 853                             $context = array_slice($edit->orig, 0, $ntrail);
 854                             $block[] = new _DiffOp_Copy($context);
 855                         }
 856                         $this->_block($x0, $ntrail + $xi - $x0,
 857                                       $y0, $ntrail + $yi - $y0,
 858                                       $block);
 859                         $block = false;
 860                     }
 861                 }
 862                 $context = $edit->orig;
 863             }
 864             else {
 865                 if (! is_array($block)) {
 866                     $context = array_slice($context, sizeof($context) - $nlead);
 867                     $x0 = $xi - sizeof($context);
 868                     $y0 = $yi - sizeof($context);
 869                     $block = array();
 870                     if ($context)
 871                         $block[] = new _DiffOp_Copy($context);
 872                 }
 873                 $block[] = $edit;
 874             }
 875
 876             if ($edit->orig)
 877                 $xi += sizeof($edit->orig);
 878             if ($edit->closing)
 879                 $yi += sizeof($edit->closing);
 880         }
 881
 882         if (is_array($block))
 883             $this->_block($x0, $xi - $x0,
 884                           $y0, $yi - $y0,
 885                           $block);
 886
 887         return $this->_end_diff();
 888     }
 889
 890     function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 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         $this->_end_block();
 905     }
 906
 907     function _start_diff() {
 908         ob_start();
 909     }
 910
 911     function _end_diff() {
 912         $val = ob_get_contents();
 913         ob_end_clean();
 914         return $val;
 915     }
 916
 917     function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 918         if ($xlen > 1)
 919             $xbeg .= "," . ($xbeg + $xlen - 1);
 920         if ($ylen > 1)
 921             $ybeg .= "," . ($ybeg + $ylen - 1);
 922
 923         return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 924     }
 925
 926     function _start_block($header) {
 927         echo $header;
 928     }
 929
 930     function _end_block() {
 931     }
 932
 933     function _lines($lines, $prefix = ' ') {
 934         foreach ($lines as $line)
 935             echo "$prefix $line\n";
 936     }
 937
 938     function _context($lines) {
 939         $this->_lines($lines);
 940     }
 941
 942     function _added($lines) {
 943         $this->_lines($lines, ">");
 944     }
 945     function _deleted($lines) {
 946         $this->_lines($lines, "<");
 947     }
 948
 949     function _changed($orig, $closing) {
 950         $this->_deleted($orig);
 951         echo "---\n";
 952         $this->_added($closing);
 953     }
 954 }
 955
 956
 957 /**
 958  *  Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
 959  *
 960  */
 961
 962 define('NBSP', "\xA0");         // iso-8859-x non-breaking space.
 963
 964 class _HWLDF_WordAccumulator {
 965     function _HWLDF_WordAccumulator () {
 966         $this->_lines = array();
 967         $this->_line = '';
 968         $this->_group = '';
 969         $this->_tag = '';
 970     }
 971
 972     function _flushGroup ($new_tag) {
 973         if ($this->_group !== '') {
 974           if ($this->_tag == 'mark')
 975             $this->_line .= "<font color=\"red\">$this->_group</font>";
 976           else
 977             $this->_line .= $this->_group;
 978         }
 979         $this->_group = '';
 980         $this->_tag = $new_tag;
 981     }
 982
 983     function _flushLine ($new_tag) {
 984         $this->_flushGroup($new_tag);
 985         if ($this->_line != '')
 986             $this->_lines[] = $this->_line;
 987         $this->_line = '';
 988     }
 989
 990     function addWords ($words, $tag = '') {
 991         if ($tag != $this->_tag)
 992             $this->_flushGroup($tag);
 993
 994         foreach ($words as $word) {
 995             // new-line should only come as first char of word.
 996             if ($word == '')
 997                 continue;
 998             if ($word[0] == "\n") {
 999                 $this->_group .= NBSP;
1000                 $this->_flushLine($tag);
1001                 $word = substr($word, 1);
1002             }
1003             assert(!strstr($word, "\n"));
1004             $this->_group .= $word;
1005         }
1006     }
1007
1008     function getLines() {
1009         $this->_flushLine('~done');
1010         return $this->_lines;
1011     }
1012 }
1013
1014 class WordLevelDiff extends MappedDiff
1015 {
1016     function WordLevelDiff ($orig_lines, $closing_lines) {
1017         list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1018         list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1019
1020
1021         $this->MappedDiff($orig_words, $closing_words,
1022                           $orig_stripped, $closing_stripped);
1023     }
1024
1025     function _split($lines) {
1026         // FIXME: fix POSIX char class.
1027 #        if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1028         if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1029                             implode("\n", $lines),
1030                             $m)) {
1031             return array(array(''), array(''));
1032         }
1033         return array($m[0], $m[1]);
1034     }
1035
1036     function orig () {
1037         $orig = new _HWLDF_WordAccumulator;
1038
1039         foreach ($this->edits as $edit) {
1040             if ($edit->type == 'copy')
1041                 $orig->addWords($edit->orig);
1042             elseif ($edit->orig)
1043                 $orig->addWords($edit->orig, 'mark');
1044         }
1045         return $orig->getLines();
1046     }
1047
1048     function closing () {
1049         $closing = new _HWLDF_WordAccumulator;
1050
1051         foreach ($this->edits as $edit) {
1052             if ($edit->type == 'copy')
1053                 $closing->addWords($edit->closing);
1054             elseif ($edit->closing)
1055                 $closing->addWords($edit->closing, 'mark');
1056         }
1057         return $closing->getLines();
1058     }
1059 }
1060
1061 /**
1062  *  Wikipedia Table style diff formatter.
1063  *
1064  */
1065 class TableDiffFormatter extends DiffFormatter
1066 {
1067         function TableDiffFormatter() {
1068                 $this->leading_context_lines = 2;
1069                 $this->trailing_context_lines = 2;
1070         }
1071
1072         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1073                 $l1 = wfMsg( "lineno", $xbeg );
1074                 $l2 = wfMsg( "lineno", $ybeg );
1075
1076                 $r = "<tr><td colspan=2 align=left><strong>{$l1}</strong></td>\n" .
1077                   "<td colspan=2 align=left><strong>{$l2}</strong></td></tr>\n";
1078                 return $r;
1079         }
1080
1081         function _start_block( $header ) {
1082                 global $wgOut;
1083                 $wgOut->addHTML( $header );
1084         }
1085
1086         function _end_block() {
1087         }
1088
1089         function _lines( $lines, $prefix=' ', $color="white" ) {
1090         }
1091
1092         function addedLine( $line ) {
1093                 return "<td>+</td><td bgcolor='#ccffcc'>" .
1094                   "<small>{$line}</small></td>";
1095         }
1096
1097         function deletedLine( $line ) {
1098                 return "<td>-</td><td bgcolor='#ffffaa'>" .
1099                   "<small>{$line}</small></td>";
1100         }
1101
1102         function emptyLine() {
1103                 return "<td colspan=2>&nbsp;</td>";
1104         }
1105
1106         function contextLine( $line ) {
1107                 return "<td> </td><td bgcolor='white'><small>{$line}</small></td>";
1108         }
1109
1110     function _added($lines) {
1111                 global $wgOut;
1112                 foreach ($lines as $line) {
1113                         $wgOut->addHTML( "<tr>" . $this->emptyLine() .
1114                           $this->addedLine( $line ) . "</tr>\n" );
1115                 }
1116         }
1117
1118         function _deleted($lines) {
1119                 global $wgOut;
1120                 foreach ($lines as $line) {
1121                         $wgOut->addHTML( "<tr>" . $this->deletedLine( $line ) .
1122                           $this->emptyLine() . "</tr>\n" );
1123                 }
1124         }
1125
1126         function _context( $lines ) {
1127                 global $wgOut;
1128                 foreach ($lines as $line) {
1129                         $wgOut->addHTML( "<tr>" . $this->contextLine( $line ) .
1130                           $this->contextLine( $line ) . "</tr>\n" );
1131                 }
1132         }
1133
1134         function _changed( $orig, $closing ) {
1135                 global $wgOut;
1136         $diff = new WordLevelDiff( $orig, $closing );
1137                 $del = $diff->orig();
1138                 $add = $diff->closing();
1139
1140                 while ( $line = array_shift( $del ) ) {
1141                         $aline = array_shift( $add );
1142                         $wgOut->addHTML( "<tr>" . $this->deletedLine( $line ) .
1143                           $this->addedLine( $aline ) . "</tr>\n" );
1144                 }
1145                 $this->_added( $add ); # If any leftovers
1146     }
1147 }
1148
1149 ?>