includes/DifferenceEngine.php

   1 <?php
   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         /* private */ var $mOldComment, $mNewComment;
  10
  11         function DifferenceEngine( $old, $new )
  12         {
  13                 $this->mOldid = $old;
  14                 $this->mNewid = $new;
  15         }
  16
  17         function showDiffPage()
  18         {
  19                 global $wgUser, $wgTitle, $wgOut, $wgLang;
  20
  21                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  22                   "{$this->mNewid})";
  23                 $mtext = wfMsg( "missingarticle", $t );
  24
  25                 $wgOut->setArticleFlag( false );
  26                 if ( ! $this->loadText() ) {
  27                         $wgOut->setPagetitle( wfMsg( "errorpagetitle" ) );
  28                         $wgOut->addHTML( $mtext );
  29                         return;
  30                 }
  31                 $wgOut->suppressQuickbar();
  32                 $wgOut->setSubtitle( wfMsg( "difference" ) );
  33                 $wgOut->setRobotpolicy( "noindex,follow" );
  34
  35                 $sk = $wgUser->getSkin();
  36                 $talk = $wgLang->getNsText( NS_TALK );
  37                 $contribs = wfMsg( "contribslink" );
  38
  39
  40                 $this->mOldComment = $sk->formatComment($this->mOldComment);
  41                 $this->mNewComment = $sk->formatComment($this->mNewComment);
  42
  43
  44                 $oldUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mOldUser ), $this->mOldUser );
  45                 $newUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mNewUser ), $this->mNewUser );
  46                 $oldUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mOldUser ), $talk );
  47                 $newUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mNewUser ), $talk );
  48                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  49                         "target=" . urlencode($this->mOldUser) );
  50                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  51                         "target=" . urlencode($this->mNewUser) );
  52                 if ( !$this->mNewid && $wgUser->isSysop() ) {
  53                         $rollback = "&nbsp;&nbsp;&nbsp;<strong>[" . $sk->makeKnownLinkObj( $wgTitle, wfMsg( "rollbacklink" ),
  54                                 "action=rollback&from=" . urlencode($this->mNewUser) ) . "]</strong>";
  55                 } else {
  56                         $rollback = "";
  57                 }
  58
  59                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink ($oldUTLink | $oldContribs)<br />".
  60                   $this->mOldComment;
  61                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink ($newUTLink | $newContribs) $rollback<br />".
  62                   $this->mNewComment;
  63
  64                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
  65                   $oldHeader, $newHeader );
  66                 $wgOut->addHTML( "<hr /><h2>{$this->mNewtitle}</h2>\n" );
  67                 $wgOut->addWikiText( $this->mNewtext );
  68         }
  69
  70         function showDiff( $otext, $ntext, $otitle, $ntitle )
  71         {
  72                 global $wgOut;
  73
  74                 $ota = explode( "\n", str_replace( "\r\n", "\n",
  75                   htmlspecialchars( $otext ) ) );
  76                 $nta = explode( "\n", str_replace( "\r\n", "\n",
  77                   htmlspecialchars( $ntext ) ) );
  78
  79                 $wgOut->addHTML( "<table border='0' width='98%'
  80 cellpadding='0' cellspacing='4px' class='diff'><tr>
  81 <td colspan='2' width='50%' align='center' class='diff-otitle'>
  82 {$otitle}</td>
  83 <td colspan='2' width='50%' align='center' class='diff-ntitle'>
  84 {$ntitle}</td>
  85 </tr>\n" );
  86
  87                 $diffs = new Diff( $ota, $nta );
  88                 $formatter = new TableDiffFormatter();
  89                 $formatter->format( $diffs );
  90                 $wgOut->addHTML( "</table>\n" );
  91         }
  92
  93         # Load the text of the articles to compare.  If newid is 0, then compare
  94         # the old article in oldid to the current article; if oldid is 0, then
  95         # compare the current article to the immediately previous one (ignoring
  96         # the value of newid).
  97         #
  98         function loadText()
  99         {
 100                 global $wgTitle, $wgOut, $wgLang;
 101                 $fname = "DifferenceEngine::loadText";
 102
 103                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 104                         $wgOut->setArticleFlag( true );
 105                         $this->mNewtitle = wfMsg( "currentrev" );
 106                         $id = $wgTitle->getArticleID();
 107
 108                         $sql = "SELECT cur_text, cur_user_text, cur_comment FROM cur WHERE cur_id={$id}";
 109                         $res = wfQuery( $sql, DB_READ, $fname );
 110                         if ( 0 == wfNumRows( $res ) ) { return false; }
 111
 112                         $s = wfFetchObject( $res );
 113                         $this->mNewtext = $s->cur_text;
 114                         $this->mNewUser = $s->cur_user_text;
 115                         $this->mNewComment = $s->cur_comment;
 116                 } else {
 117                         $sql = "SELECT old_timestamp,old_text,old_flags,old_user_text,old_comment FROM old WHERE " .
 118                           "old_id={$this->mNewid}";
 119
 120                         $res = wfQuery( $sql, DB_READ, $fname );
 121                         if ( 0 == wfNumRows( $res ) ) { return false; }
 122
 123                         $s = wfFetchObject( $res );
 124                         $this->mNewtext = Article::getRevisionText( $s );
 125
 126                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 127                         $this->mNewtitle = wfMsg( "revisionasof", $t );
 128                         $this->mNewUser = $s->old_user_text;
 129                         $this->mNewComment = $s->old_comment;
 130                 }
 131                 if ( 0 == $this->mOldid ) {
 132                         $sql = "SELECT old_timestamp,old_text,old_flags,old_user_text,old_comment " .
 133                           "FROM old USE INDEX (name_title_timestamp) WHERE " .
 134                           "old_namespace=" . $wgTitle->getNamespace() . " AND " .
 135                           "old_title='" . wfStrencode( $wgTitle->getDBkey() ) .
 136                           "' ORDER BY inverse_timestamp LIMIT 1";
 137                         $res = wfQuery( $sql, DB_READ, $fname );
 138                 } else {
 139                         $sql = "SELECT old_timestamp,old_text,old_flags,old_user_text,old_comment FROM old WHERE " .
 140                           "old_id={$this->mOldid}";
 141                         $res = wfQuery( $sql, DB_READ, $fname );
 142                 }
 143                 if ( 0 == wfNumRows( $res ) ) { return false; }
 144
 145                 $s = wfFetchObject( $res );
 146                 $this->mOldtext = Article::getRevisionText( $s );
 147
 148                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 149                 $this->mOldtitle = wfMsg( "revisionasof", $t );
 150                 $this->mOldUser = $s->old_user_text;
 151                 $this->mOldComment = $s->old_comment;
 152
 153                 return true;
 154         }
 155 }
 156
 157 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 158 //
 159 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 160 // You may copy this code freely under the conditions of the GPL.
 161 //
 162
 163 define('USE_ASSERTS', function_exists('assert'));
 164
 165 class _DiffOp {
 166         var $type;
 167         var $orig;
 168         var $closing;
 169
 170         function reverse() {
 171                 trigger_error("pure virtual", E_USER_ERROR);
 172         }
 173
 174         function norig() {
 175                 return $this->orig ? sizeof($this->orig) : 0;
 176         }
 177
 178         function nclosing() {
 179                 return $this->closing ? sizeof($this->closing) : 0;
 180         }
 181 }
 182
 183 class _DiffOp_Copy extends _DiffOp {
 184         var $type = 'copy';
 185
 186         function _DiffOp_Copy ($orig, $closing = false) {
 187                 if (!is_array($closing))
 188                         $closing = $orig;
 189                 $this->orig = $orig;
 190                 $this->closing = $closing;
 191         }
 192
 193         function reverse() {
 194                 return new _DiffOp_Copy($this->closing, $this->orig);
 195         }
 196 }
 197
 198 class _DiffOp_Delete extends _DiffOp {
 199         var $type = 'delete';
 200
 201         function _DiffOp_Delete ($lines) {
 202                 $this->orig = $lines;
 203                 $this->closing = false;
 204         }
 205
 206         function reverse() {
 207                 return new _DiffOp_Add($this->orig);
 208         }
 209 }
 210
 211 class _DiffOp_Add extends _DiffOp {
 212         var $type = 'add';
 213
 214         function _DiffOp_Add ($lines) {
 215                 $this->closing = $lines;
 216                 $this->orig = false;
 217         }
 218
 219         function reverse() {
 220                 return new _DiffOp_Delete($this->closing);
 221         }
 222 }
 223
 224 class _DiffOp_Change extends _DiffOp {
 225         var $type = 'change';
 226
 227         function _DiffOp_Change ($orig, $closing) {
 228                 $this->orig = $orig;
 229                 $this->closing = $closing;
 230         }
 231
 232         function reverse() {
 233                 return new _DiffOp_Change($this->closing, $this->orig);
 234         }
 235 }
 236
 237
 238 /**
 239  * Class used internally by Diff to actually compute the diffs.
 240  *
 241  * The algorithm used here is mostly lifted from the perl module
 242  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 243  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 244  *
 245  * More ideas are taken from:
 246  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 247  *
 248  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 249  * diffutils-2.7, which can be found at:
 250  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 251  *
 252  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 253  * are my own.
 254  *
 255  * @author Geoffrey T. Dairiki
 256  * @access private
 257  */
 258 class _DiffEngine
 259 {
 260         function diff ($from_lines, $to_lines) {
 261                 $n_from = sizeof($from_lines);
 262                 $n_to = sizeof($to_lines);
 263
 264                 $this->xchanged = $this->ychanged = array();
 265                 $this->xv = $this->yv = array();
 266                 $this->xind = $this->yind = array();
 267                 unset($this->seq);
 268                 unset($this->in_seq);
 269                 unset($this->lcs);
 270
 271                 // Skip leading common lines.
 272                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 273                         if ($from_lines[$skip] != $to_lines[$skip])
 274                                 break;
 275                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 276                 }
 277                 // Skip trailing common lines.
 278                 $xi = $n_from; $yi = $n_to;
 279                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 280                         if ($from_lines[$xi] != $to_lines[$yi])
 281                                 break;
 282                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 283                 }
 284
 285                 // Ignore lines which do not exist in both files.
 286                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 287                         $xhash[$from_lines[$xi]] = 1;
 288                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 289                         $line = $to_lines[$yi];
 290                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 291                                 continue;
 292                         $yhash[$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($yhash[$line])) )
 299                                 continue;
 300                         $this->xv[] = $line;
 301                         $this->xind[] = $xi;
 302                 }
 303
 304                 // Find the LCS.
 305                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 306
 307                 // Merge edits when possible
 308                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 309                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 310
 311                 // Compute the edit operations.
 312                 $edits = array();
 313                 $xi = $yi = 0;
 314                 while ($xi < $n_from || $yi < $n_to) {
 315                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 316                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 317
 318                         // Skip matching "snake".
 319                         $copy = array();
 320                         while ( $xi < $n_from && $yi < $n_to
 321                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 322                                 $copy[] = $from_lines[$xi++];
 323                                 ++$yi;
 324                         }
 325                         if ($copy)
 326                                 $edits[] = new _DiffOp_Copy($copy);
 327
 328                         // Find deletes & adds.
 329                         $delete = array();
 330                         while ($xi < $n_from && $this->xchanged[$xi])
 331                                 $delete[] = $from_lines[$xi++];
 332
 333                         $add = array();
 334                         while ($yi < $n_to && $this->ychanged[$yi])
 335                                 $add[] = $to_lines[$yi++];
 336
 337                         if ($delete && $add)
 338                                 $edits[] = new _DiffOp_Change($delete, $add);
 339                         elseif ($delete)
 340                                 $edits[] = new _DiffOp_Delete($delete);
 341                         elseif ($add)
 342                                 $edits[] = new _DiffOp_Add($add);
 343                 }
 344                 return $edits;
 345         }
 346
 347
 348         /* Divide the Largest Common Subsequence (LCS) of the sequences
 349          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 350          * sized segments.
 351          *
 352          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 353          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 354          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 355          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 356          * that (X0, Y0) == (XOFF, YOFF) and
 357          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 358          *
 359          * This function assumes that the first lines of the specified portions
 360          * of the two files do not match, and likewise that the last lines do not
 361          * match.  The caller must trim matching lines from the beginning and end
 362          * of the portions it is going to specify.
 363          */
 364         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 365         $flip = false;
 366
 367         if ($xlim - $xoff > $ylim - $yoff) {
 368                 // Things seems faster (I'm not sure I understand why)
 369                         // when the shortest sequence in X.
 370                         $flip = true;
 371                 list ($xoff, $xlim, $yoff, $ylim)
 372                 = array( $yoff, $ylim, $xoff, $xlim);
 373                 }
 374
 375         if ($flip)
 376                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 377                 $ymatches[$this->xv[$i]][] = $i;
 378         else
 379                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 380                 $ymatches[$this->yv[$i]][] = $i;
 381
 382         $this->lcs = 0;
 383         $this->seq[0]= $yoff - 1;
 384         $this->in_seq = array();
 385         $ymids[0] = array();
 386
 387         $numer = $xlim - $xoff + $nchunks - 1;
 388         $x = $xoff;
 389         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 390                 if ($chunk > 0)
 391                 for ($i = 0; $i <= $this->lcs; $i++)
 392                         $ymids[$i][$chunk-1] = $this->seq[$i];
 393
 394                 $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 395                 for ( ; $x < $x1; $x++) {
 396                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 397                                 if (empty($ymatches[$line]))
 398                         continue;
 399                 $matches = $ymatches[$line];
 400                                 reset($matches);
 401                 while (list ($junk, $y) = each($matches))
 402                         if (empty($this->in_seq[$y])) {
 403                         $k = $this->_lcs_pos($y);
 404                         USE_ASSERTS && assert($k > 0);
 405                         $ymids[$k] = $ymids[$k-1];
 406                         break;
 407                                         }
 408                 while (list ($junk, $y) = each($matches)) {
 409                         if ($y > $this->seq[$k-1]) {
 410                         USE_ASSERTS && assert($y < $this->seq[$k]);
 411                         // Optimization: this is a common case:
 412                         //      next match is just replacing previous match.
 413                         $this->in_seq[$this->seq[$k]] = false;
 414                         $this->seq[$k] = $y;
 415                         $this->in_seq[$y] = 1;
 416                                         }
 417                         else if (empty($this->in_seq[$y])) {
 418                         $k = $this->_lcs_pos($y);
 419                         USE_ASSERTS && assert($k > 0);
 420                         $ymids[$k] = $ymids[$k-1];
 421                                         }
 422                                 }
 423                         }
 424                 }
 425
 426         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 427         $ymid = $ymids[$this->lcs];
 428         for ($n = 0; $n < $nchunks - 1; $n++) {
 429                 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 430                 $y1 = $ymid[$n] + 1;
 431                 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 432                 }
 433         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 434
 435         return array($this->lcs, $seps);
 436         }
 437
 438         function _lcs_pos ($ypos) {
 439         $end = $this->lcs;
 440         if ($end == 0 || $ypos > $this->seq[$end]) {
 441                 $this->seq[++$this->lcs] = $ypos;
 442                 $this->in_seq[$ypos] = 1;
 443                 return $this->lcs;
 444                 }
 445
 446         $beg = 1;
 447         while ($beg < $end) {
 448                 $mid = (int)(($beg + $end) / 2);
 449                 if ( $ypos > $this->seq[$mid] )
 450                 $beg = $mid + 1;
 451                 else
 452                 $end = $mid;
 453                 }
 454
 455         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 456
 457         $this->in_seq[$this->seq[$end]] = false;
 458         $this->seq[$end] = $ypos;
 459         $this->in_seq[$ypos] = 1;
 460         return $end;
 461         }
 462
 463         /* Find LCS of two sequences.
 464          *
 465          * The results are recorded in the vectors $this->{x,y}changed[], by
 466          * storing a 1 in the element for each line that is an insertion
 467          * or deletion (ie. is not in the LCS).
 468          *
 469          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 470          *
 471          * Note that XLIM, YLIM are exclusive bounds.
 472          * All line numbers are origin-0 and discarded lines are not counted.
 473          */
 474         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 475         // Slide down the bottom initial diagonal.
 476         while ($xoff < $xlim && $yoff < $ylim
 477                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 478                 ++$xoff;
 479                 ++$yoff;
 480                 }
 481
 482         // Slide up the top initial diagonal.
 483         while ($xlim > $xoff && $ylim > $yoff
 484                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 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)
 497                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 498                 }
 499
 500         if ($lcs == 0) {
 501                 // X and Y sequences have no common subsequence:
 502                 // mark all changed.
 503                 while ($yoff < $ylim)
 504                 $this->ychanged[$this->yind[$yoff++]] = 1;
 505                 while ($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 = $pt2;
 515                         }
 516                 }
 517         }
 518
 519         /* Adjust inserts/deletes of identical lines to join changes
 520          * as much as possible.
 521          *
 522          * We do something when a run of changed lines include a
 523          * line at one end and has an excluded, identical line at the other.
 524          * We are free to choose which identical line is included.
 525          * `compareseq' usually chooses the one at the beginning,
 526          * but usually it is cleaner to consider the following identical line
 527          * to be the "change".
 528          *
 529          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 530          */
 531         function _shift_boundaries ($lines, &$changed, $other_changed) {
 532         $i = 0;
 533         $j = 0;
 534
 535         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 536         $len = sizeof($lines);
 537         $other_len = sizeof($other_changed);
 538
 539         while (1) {
 540                 /*
 541                  * Scan forwards to find beginning of another run of changes.
 542                  * Also keep track of the corresponding point in the other file.
 543                  *
 544                  * Throughout this code, $i and $j are adjusted together so that
 545                  * the first $i elements of $changed and the first $j elements
 546                  * of $other_changed both contain the same number of zeros
 547                  * (unchanged lines).
 548                  * Furthermore, $j is always kept so that $j == $other_len or
 549                  * $other_changed[$j] == false.
 550                  */
 551                 while ($j < $other_len && $other_changed[$j])
 552                 $j++;
 553
 554                 while ($i < $len && ! $changed[$i]) {
 555                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 556                 $i++; $j++;
 557                 while ($j < $other_len && $other_changed[$j])
 558                         $j++;
 559                         }
 560
 561                 if ($i == $len)
 562                 break;
 563
 564                 $start = $i;
 565
 566                 // Find the end of this run of changes.
 567                 while (++$i < $len && $changed[$i])
 568                 continue;
 569
 570                 do {
 571                 /*
 572                  * Record the length of this run of changes, so that
 573                  * we can later determine whether the run has grown.
 574                  */
 575                 $runlength = $i - $start;
 576
 577                 /*
 578                  * Move the changed region back, so long as the
 579                  * previous unchanged line matches the last changed one.
 580                  * This merges with previous changed regions.
 581                  */
 582                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 583                         $changed[--$start] = 1;
 584                         $changed[--$i] = false;
 585                         while ($start > 0 && $changed[$start - 1])
 586                         $start--;
 587                         USE_ASSERTS && assert('$j > 0');
 588                         while ($other_changed[--$j])
 589                         continue;
 590                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 591                                 }
 592
 593                 /*
 594                  * Set CORRESPONDING to the end of the changed run, at the last
 595                  * point where it corresponds to a changed run in the other file.
 596                  * CORRESPONDING == LEN means no such point has been found.
 597                  */
 598                 $corresponding = $j < $other_len ? $i : $len;
 599
 600                 /*
 601                  * Move the changed region forward, so long as the
 602                  * first changed line matches the following unchanged one.
 603                  * This merges with following changed regions.
 604                  * Do this second, so that if there are no merges,
 605                  * the changed region is moved forward as far as possible.
 606                  */
 607                 while ($i < $len && $lines[$start] == $lines[$i]) {
 608                         $changed[$start++] = false;
 609                         $changed[$i++] = 1;
 610                         while ($i < $len && $changed[$i])
 611                         $i++;
 612
 613                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 614                         $j++;
 615                         if ($j < $other_len && $other_changed[$j]) {
 616                         $corresponding = $i;
 617                         while ($j < $other_len && $other_changed[$j])
 618                                 $j++;
 619                                         }
 620                                 }
 621                         } while ($runlength != $i - $start);
 622
 623                 /*
 624                  * If possible, move the fully-merged run of changes
 625                  * back to a corresponding run in the other file.
 626                  */
 627                 while ($corresponding < $i) {
 628                 $changed[--$start] = 1;
 629                 $changed[--$i] = 0;
 630                 USE_ASSERTS && assert('$j > 0');
 631                 while ($other_changed[--$j])
 632                         continue;
 633                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 634                         }
 635                 }
 636         }
 637 }
 638
 639 /**
 640  * Class representing a 'diff' between two sequences of strings.
 641  */
 642 class Diff
 643 {
 644         var $edits;
 645
 646         /**
 647          * Constructor.
 648          * Computes diff between sequences of strings.
 649          *
 650          * @param $from_lines array An array of strings.
 651          *                (Typically these are lines from a file.)
 652          * @param $to_lines array An array of strings.
 653          */
 654         function Diff($from_lines, $to_lines) {
 655                 $eng = new _DiffEngine;
 656                 $this->edits = $eng->diff($from_lines, $to_lines);
 657                 //$this->_check($from_lines, $to_lines);
 658         }
 659
 660         /**
 661          * Compute reversed Diff.
 662          *
 663          * SYNOPSIS:
 664          *
 665          *      $diff = new Diff($lines1, $lines2);
 666          *      $rev = $diff->reverse();
 667          * @return object A Diff object representing the inverse of the
 668          *                                original diff.
 669          */
 670         function reverse () {
 671         $rev = $this;
 672                 $rev->edits = array();
 673                 foreach ($this->edits as $edit) {
 674                         $rev->edits[] = $edit->reverse();
 675                 }
 676         return $rev;
 677         }
 678
 679         /**
 680          * Check for empty diff.
 681          *
 682          * @return bool True iff two sequences were identical.
 683          */
 684         function isEmpty () {
 685                 foreach ($this->edits as $edit) {
 686                         if ($edit->type != 'copy')
 687                                 return false;
 688                 }
 689                 return true;
 690         }
 691
 692         /**
 693          * Compute the length of the Longest Common Subsequence (LCS).
 694          *
 695          * This is mostly for diagnostic purposed.
 696          *
 697          * @return int The length of the LCS.
 698          */
 699         function lcs () {
 700         $lcs = 0;
 701                 foreach ($this->edits as $edit) {
 702                         if ($edit->type == 'copy')
 703                                 $lcs += sizeof($edit->orig);
 704                 }
 705         return $lcs;
 706         }
 707
 708         /**
 709          * Get the original set of lines.
 710          *
 711          * This reconstructs the $from_lines parameter passed to the
 712          * constructor.
 713          *
 714          * @return array The original sequence of strings.
 715          */
 716         function orig() {
 717                 $lines = array();
 718
 719                 foreach ($this->edits as $edit) {
 720                         if ($edit->orig)
 721                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 722                 }
 723                 return $lines;
 724         }
 725
 726         /**
 727          * Get the closing set of lines.
 728          *
 729          * This reconstructs the $to_lines parameter passed to the
 730          * constructor.
 731          *
 732          * @return array The sequence of strings.
 733          */
 734         function closing() {
 735                 $lines = array();
 736
 737                 foreach ($this->edits as $edit) {
 738                         if ($edit->closing)
 739                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 740                 }
 741                 return $lines;
 742         }
 743
 744         /**
 745          * Check a Diff for validity.
 746          *
 747          * This is here only for debugging purposes.
 748          */
 749         function _check ($from_lines, $to_lines) {
 750                 if (serialize($from_lines) != serialize($this->orig()))
 751                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 752                 if (serialize($to_lines) != serialize($this->closing()))
 753                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 754
 755                 $rev = $this->reverse();
 756                 if (serialize($to_lines) != serialize($rev->orig()))
 757                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 758                 if (serialize($from_lines) != serialize($rev->closing()))
 759                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 760
 761
 762                 $prevtype = 'none';
 763                 foreach ($this->edits as $edit) {
 764                         if ( $prevtype == $edit->type )
 765                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 766                         $prevtype = $edit->type;
 767                 }
 768
 769                 $lcs = $this->lcs();
 770                 trigger_error("Diff okay: LCS = $lcs", E_USER_NOTICE);
 771         }
 772 }
 773
 774 /**
 775  * FIXME: bad name.
 776  */
 777 class MappedDiff
 778 extends Diff
 779 {
 780         /**
 781          * Constructor.
 782          *
 783          * Computes diff between sequences of strings.
 784          *
 785          * This can be used to compute things like
 786          * case-insensitve diffs, or diffs which ignore
 787          * changes in white-space.
 788          *
 789          * @param $from_lines array An array of strings.
 790          *      (Typically these are lines from a file.)
 791          *
 792          * @param $to_lines array An array of strings.
 793          *
 794          * @param $mapped_from_lines array This array should
 795          *      have the same size number of elements as $from_lines.
 796          *      The elements in $mapped_from_lines and
 797          *      $mapped_to_lines are what is actually compared
 798          *      when computing the diff.
 799          *
 800          * @param $mapped_to_lines array This array should
 801          *      have the same number of elements as $to_lines.
 802          */
 803         function MappedDiff($from_lines, $to_lines,
 804                                                 $mapped_from_lines, $mapped_to_lines) {
 805
 806                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 807                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 808
 809                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 810
 811                 $xi = $yi = 0;
 812                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 813                         $orig = &$this->edits[$i]->orig;
 814                         if (is_array($orig)) {
 815                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 816                                 $xi += sizeof($orig);
 817                         }
 818
 819                         $closing = &$this->edits[$i]->closing;
 820                         if (is_array($closing)) {
 821                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 822                                 $yi += sizeof($closing);
 823                         }
 824                 }
 825         }
 826 }
 827
 828 /**
 829  * A class to format Diffs
 830  *
 831  * This class formats the diff in classic diff format.
 832  * It is intended that this class be customized via inheritance,
 833  * to obtain fancier outputs.
 834  */
 835 class DiffFormatter
 836 {
 837         /**
 838          * Number of leading context "lines" to preserve.
 839          *
 840          * This should be left at zero for this class, but subclasses
 841          * may want to set this to other values.
 842          */
 843         var $leading_context_lines = 0;
 844
 845         /**
 846          * Number of trailing context "lines" to preserve.
 847          *
 848          * This should be left at zero for this class, but subclasses
 849          * may want to set this to other values.
 850          */
 851         var $trailing_context_lines = 0;
 852
 853         /**
 854          * Format a diff.
 855          *
 856          * @param $diff object A Diff object.
 857          * @return string The formatted output.
 858          */
 859         function format($diff) {
 860
 861                 $xi = $yi = 1;
 862                 $block = false;
 863                 $context = array();
 864
 865                 $nlead = $this->leading_context_lines;
 866                 $ntrail = $this->trailing_context_lines;
 867
 868                 $this->_start_diff();
 869
 870                 foreach ($diff->edits as $edit) {
 871                         if ($edit->type == 'copy') {
 872                                 if (is_array($block)) {
 873                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 874                                                 $block[] = $edit;
 875                                         }
 876                                         else{
 877                                                 if ($ntrail) {
 878                                                         $context = array_slice($edit->orig, 0, $ntrail);
 879                                                         $block[] = new _DiffOp_Copy($context);
 880                                                 }
 881                                                 $this->_block($x0, $ntrail + $xi - $x0,
 882                                                                           $y0, $ntrail + $yi - $y0,
 883                                                                           $block);
 884                                                 $block = false;
 885                                         }
 886                                 }
 887                                 $context = $edit->orig;
 888                         }
 889                         else {
 890                                 if (! is_array($block)) {
 891                                         $context = array_slice($context, sizeof($context) - $nlead);
 892                                         $x0 = $xi - sizeof($context);
 893                                         $y0 = $yi - sizeof($context);
 894                                         $block = array();
 895                                         if ($context)
 896                                                 $block[] = new _DiffOp_Copy($context);
 897                                 }
 898                                 $block[] = $edit;
 899                         }
 900
 901                         if ($edit->orig)
 902                                 $xi += sizeof($edit->orig);
 903                         if ($edit->closing)
 904                                 $yi += sizeof($edit->closing);
 905                 }
 906
 907                 if (is_array($block))
 908                         $this->_block($x0, $xi - $x0,
 909                                                   $y0, $yi - $y0,
 910                                                   $block);
 911
 912                 return $this->_end_diff();
 913         }
 914
 915         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 916                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 917                 foreach ($edits as $edit) {
 918                         if ($edit->type == 'copy')
 919                                 $this->_context($edit->orig);
 920                         elseif ($edit->type == 'add')
 921                                 $this->_added($edit->closing);
 922                         elseif ($edit->type == 'delete')
 923                                 $this->_deleted($edit->orig);
 924                         elseif ($edit->type == 'change')
 925                                 $this->_changed($edit->orig, $edit->closing);
 926                         else
 927                                 trigger_error("Unknown edit type", E_USER_ERROR);
 928                 }
 929                 $this->_end_block();
 930         }
 931
 932         function _start_diff() {
 933                 ob_start();
 934         }
 935
 936         function _end_diff() {
 937                 $val = ob_get_contents();
 938                 ob_end_clean();
 939                 return $val;
 940         }
 941
 942         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 943                 if ($xlen > 1)
 944                         $xbeg .= "," . ($xbeg + $xlen - 1);
 945                 if ($ylen > 1)
 946                         $ybeg .= "," . ($ybeg + $ylen - 1);
 947
 948                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 949         }
 950
 951         function _start_block($header) {
 952                 echo $header;
 953         }
 954
 955         function _end_block() {
 956         }
 957
 958         function _lines($lines, $prefix = ' ') {
 959                 foreach ($lines as $line)
 960                         echo "$prefix $line\n";
 961         }
 962
 963         function _context($lines) {
 964                 $this->_lines($lines);
 965         }
 966
 967         function _added($lines) {
 968                 $this->_lines($lines, ">");
 969         }
 970         function _deleted($lines) {
 971                 $this->_lines($lines, "<");
 972         }
 973
 974         function _changed($orig, $closing) {
 975                 $this->_deleted($orig);
 976                 echo "---\n";
 977                 $this->_added($closing);
 978         }
 979 }
 980
 981
 982 /**
 983  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
 984  *
 985  */
 986
 987 define('NBSP', "\xA0");                 // iso-8859-x non-breaking space.
 988
 989 class _HWLDF_WordAccumulator {
 990         function _HWLDF_WordAccumulator () {
 991                 $this->_lines = array();
 992                 $this->_line = '';
 993                 $this->_group = '';
 994                 $this->_tag = '';
 995         }
 996
 997         function _flushGroup ($new_tag) {
 998                 if ($this->_group !== '') {
 999           if ($this->_tag == 'mark')
1000                         $this->_line .= "<font color=\"red\">$this->_group</font>";
1001           else
1002                 $this->_line .= $this->_group;
1003         }
1004                 $this->_group = '';
1005                 $this->_tag = $new_tag;
1006         }
1007
1008         function _flushLine ($new_tag) {
1009                 $this->_flushGroup($new_tag);
1010                 if ($this->_line != '')
1011                         $this->_lines[] = $this->_line;
1012                 $this->_line = '';
1013         }
1014
1015         function addWords ($words, $tag = '') {
1016                 if ($tag != $this->_tag)
1017                         $this->_flushGroup($tag);
1018
1019                 foreach ($words as $word) {
1020                         // new-line should only come as first char of word.
1021                         if ($word == '')
1022                                 continue;
1023                         if ($word[0] == "\n") {
1024                                 $this->_group .= NBSP;
1025                                 $this->_flushLine($tag);
1026                                 $word = substr($word, 1);
1027                         }
1028                         assert(!strstr($word, "\n"));
1029                         $this->_group .= $word;
1030                 }
1031         }
1032
1033         function getLines() {
1034                 $this->_flushLine('~done');
1035                 return $this->_lines;
1036         }
1037 }
1038
1039 class WordLevelDiff extends MappedDiff
1040 {
1041         function WordLevelDiff ($orig_lines, $closing_lines) {
1042                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1043                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1044
1045
1046                 $this->MappedDiff($orig_words, $closing_words,
1047                                                   $orig_stripped, $closing_stripped);
1048         }
1049
1050         function _split($lines) {
1051                 // FIXME: fix POSIX char class.
1052 #                if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1053                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1054                                                         implode("\n", $lines),
1055                                                         $m)) {
1056                         return array(array(''), array(''));
1057                 }
1058                 return array($m[0], $m[1]);
1059         }
1060
1061         function orig () {
1062                 $orig = new _HWLDF_WordAccumulator;
1063
1064                 foreach ($this->edits as $edit) {
1065                         if ($edit->type == 'copy')
1066                                 $orig->addWords($edit->orig);
1067                         elseif ($edit->orig)
1068                                 $orig->addWords($edit->orig, 'mark');
1069                 }
1070                 return $orig->getLines();
1071         }
1072
1073         function closing () {
1074                 $closing = new _HWLDF_WordAccumulator;
1075
1076                 foreach ($this->edits as $edit) {
1077                         if ($edit->type == 'copy')
1078                                 $closing->addWords($edit->closing);
1079                         elseif ($edit->closing)
1080                                 $closing->addWords($edit->closing, 'mark');
1081                 }
1082                 return $closing->getLines();
1083         }
1084 }
1085
1086 /**
1087  *      Wikipedia Table style diff formatter.
1088  *
1089  */
1090 class TableDiffFormatter extends DiffFormatter
1091 {
1092         function TableDiffFormatter() {
1093                 $this->leading_context_lines = 2;
1094                 $this->trailing_context_lines = 2;
1095         }
1096
1097         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1098                 $l1 = wfMsg( "lineno", $xbeg );
1099                 $l2 = wfMsg( "lineno", $ybeg );
1100
1101                 $r = "<tr><td colspan='2' align='left'><strong>{$l1}</strong></td>\n" .
1102                   "<td colspan='2' align='left'><strong>{$l2}</strong></td></tr>\n";
1103                 return $r;
1104         }
1105
1106         function _start_block( $header ) {
1107                 global $wgOut;
1108                 $wgOut->addHTML( $header );
1109         }
1110
1111         function _end_block() {
1112         }
1113
1114         function _lines( $lines, $prefix=' ', $color="white" ) {
1115         }
1116
1117         function addedLine( $line ) {
1118                 return "<td>+</td><td class='diff-addedline'>" .
1119                   "<small>{$line}</small></td>";
1120         }
1121
1122         function deletedLine( $line ) {
1123                 return "<td>-</td><td class='diff-deletedline'" .
1124                   "<small>{$line}</small></td>";
1125         }
1126
1127         function emptyLine() {
1128                 return "<td colspan='2'>&nbsp;</td>";
1129         }
1130
1131         function contextLine( $line ) {
1132                 return "<td> </td><td class='diff-context'><small>{$line}</small></td>";
1133         }
1134
1135         function _added($lines) {
1136                 global $wgOut;
1137                 foreach ($lines as $line) {
1138                         $wgOut->addHTML( "<tr>" . $this->emptyLine() .
1139                           $this->addedLine( $line ) . "</tr>\n" );
1140                 }
1141         }
1142
1143         function _deleted($lines) {
1144                 global $wgOut;
1145                 foreach ($lines as $line) {
1146                         $wgOut->addHTML( "<tr>" . $this->deletedLine( $line ) .
1147                           $this->emptyLine() . "</tr>\n" );
1148                 }
1149         }
1150
1151         function _context( $lines ) {
1152                 global $wgOut;
1153                 foreach ($lines as $line) {
1154                         $wgOut->addHTML( "<tr>" . $this->contextLine( $line ) .
1155                           $this->contextLine( $line ) . "</tr>\n" );
1156                 }
1157         }
1158
1159         function _changed( $orig, $closing ) {
1160                 global $wgOut;
1161                 $diff = new WordLevelDiff( $orig, $closing );
1162                 $del = $diff->orig();
1163                 $add = $diff->closing();
1164
1165                 while ( $line = array_shift( $del ) ) {
1166                         $aline = array_shift( $add );
1167                         $wgOut->addHTML( "<tr>" . $this->deletedLine( $line ) .
1168                           $this->addedLine( $aline ) . "</tr>\n" );
1169                 }
1170                 $this->_added( $add ); # If any leftovers
1171         }
1172 }
1173
1174 ?>