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         /* private */ var $mOldPage, $mNewPage;
  11
  12         function DifferenceEngine( $old, $new )
  13         {
  14                 $this->mOldid = $old;
  15                 $this->mNewid = $new;
  16         }
  17
  18         function showDiffPage()
  19         {
  20                 global $wgUser, $wgTitle, $wgOut, $wgLang;
  21                 $fname = "DifferenceEngine::showDiffPage";
  22                 wfProfileIn( $fname );
  23
  24                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  25                   "{$this->mNewid})";
  26                 $mtext = wfMsg( "missingarticle", $t );
  27
  28                 $wgOut->setArticleFlag( false );
  29                 if ( ! $this->loadText() ) {
  30                         $wgOut->setPagetitle( wfMsg( "errorpagetitle" ) );
  31                         $wgOut->addHTML( $mtext );
  32                         wfProfileOut( $fname );
  33                         return;
  34                 }
  35                 $wgOut->suppressQuickbar();
  36
  37                 $oldTitle = $this->mOldPage->getPrefixedText();
  38                 $newTitle = $this->mNewPage->getPrefixedText();
  39                 if( $oldTitle == $newTitle ) {
  40                         $wgOut->setPageTitle( $newTitle );
  41                 } else {
  42                         $wgOut->setPageTitle( $oldTitle . ", " . $newTitle );
  43                 }
  44                 $wgOut->setSubtitle( wfMsg( "difference" ) );
  45                 $wgOut->setRobotpolicy( "noindex,follow" );
  46
  47                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  48                         $wgOut->loginToUse();
  49                         $wgOut->output();
  50                         wfProfileOut( $fname );
  51                         exit;
  52                 }
  53
  54                 $sk = $wgUser->getSkin();
  55                 $talk = $wgLang->getNsText( NS_TALK );
  56                 $contribs = wfMsg( "contribslink" );
  57
  58
  59                 $this->mOldComment = $sk->formatComment($this->mOldComment);
  60                 $this->mNewComment = $sk->formatComment($this->mNewComment);
  61
  62
  63                 $oldUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mOldUser ), $this->mOldUser );
  64                 $newUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mNewUser ), $this->mNewUser );
  65                 $oldUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mOldUser ), $talk );
  66                 $newUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mNewUser ), $talk );
  67                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  68                         "target=" . urlencode($this->mOldUser) );
  69                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  70                         "target=" . urlencode($this->mNewUser) );
  71                 if ( !$this->mNewid && $wgUser->isSysop() ) {
  72                         $rollback = "&nbsp;&nbsp;&nbsp;<strong>[" . $sk->makeKnownLinkObj( $wgTitle, wfMsg( "rollbacklink" ),
  73                                 "action=rollback&from=" . urlencode($this->mNewUser) ) . "]</strong>";
  74                 } else {
  75                         $rollback = "";
  76                 }
  77
  78                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink ($oldUTLink | $oldContribs)<br />".
  79                   $this->mOldComment;
  80                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink ($newUTLink | $newContribs) $rollback<br />".
  81                   $this->mNewComment;
  82
  83                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
  84                   $oldHeader, $newHeader );
  85                 $wgOut->addHTML( "<hr /><h2>{$this->mNewtitle}</h2>\n" );
  86                 $wgOut->addWikiText( $this->mNewtext );
  87
  88                 wfProfileOut( $fname );
  89         }
  90
  91         function showDiff( $otext, $ntext, $otitle, $ntitle )
  92         {
  93                 global $wgOut;
  94
  95                 $ota = explode( "\n", str_replace( "\r\n", "\n",
  96                   htmlspecialchars( $otext ) ) );
  97                 $nta = explode( "\n", str_replace( "\r\n", "\n",
  98                   htmlspecialchars( $ntext ) ) );
  99
 100                 $wgOut->addHTML( "<table border='0' width='98%'
 101 cellpadding='0' cellspacing='4px' class='diff'><tr>
 102 <td colspan='2' width='50%' align='center' class='diff-otitle'>
 103 {$otitle}</td>
 104 <td colspan='2' width='50%' align='center' class='diff-ntitle'>
 105 {$ntitle}</td>
 106 </tr>\n" );
 107
 108                 $diffs = new Diff( $ota, $nta );
 109                 $formatter = new TableDiffFormatter();
 110                 $formatter->format( $diffs );
 111                 $wgOut->addHTML( "</table>\n" );
 112         }
 113
 114         # Load the text of the articles to compare.  If newid is 0, then compare
 115         # the old article in oldid to the current article; if oldid is 0, then
 116         # compare the current article to the immediately previous one (ignoring
 117         # the value of newid).
 118         #
 119         function loadText()
 120         {
 121                 global $wgTitle, $wgOut, $wgLang, $wgIsMySQL, $wgIsPg;
 122                 $fname = "DifferenceEngine::loadText";
 123
 124                 $oldtable=$wgIsPg?'"old"':'old';
 125                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 126                         $wgOut->setArticleFlag( true );
 127                         $this->mNewtitle = wfMsg( "currentrev" );
 128                         $id = $wgTitle->getArticleID();
 129
 130                         $sql = "SELECT cur_text, cur_user_text, cur_comment FROM cur WHERE cur_id={$id}";
 131                         $res = wfQuery( $sql, DB_READ, $fname );
 132                         if ( 0 == wfNumRows( $res ) ) { return false; }
 133
 134                         $s = wfFetchObject( $res );
 135                         $this->mNewPage = &$wgTitle;
 136                         $this->mNewtext = $s->cur_text;
 137                         $this->mNewUser = $s->cur_user_text;
 138                         $this->mNewComment = $s->cur_comment;
 139                 } else {
 140                         $sql = "SELECT old_namespace,old_title,old_timestamp,old_text,old_flags,old_user_text,old_comment FROM $oldtable WHERE " .
 141                           "old_id={$this->mNewid}";
 142
 143                         $res = wfQuery( $sql, DB_READ, $fname );
 144                         if ( 0 == wfNumRows( $res ) ) { return false; }
 145
 146                         $s = wfFetchObject( $res );
 147                         $this->mNewtext = Article::getRevisionText( $s );
 148
 149                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 150                         $this->mNewPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 151                         $this->mNewtitle = wfMsg( "revisionasof", $t );
 152                         $this->mNewUser = $s->old_user_text;
 153                         $this->mNewComment = $s->old_comment;
 154                 }
 155                 if ( 0 == $this->mOldid ) {
 156                         $use_index=$wgIsMySQL?"USE INDEX (name_title_timestamp)":"";
 157                         $sql = "SELECT old_namespace,old_title,old_timestamp,old_text,old_flags,old_user_text,old_comment " .
 158                           "FROM $oldtable $use_index WHERE " .
 159                           "old_namespace=" . $this->mNewPage->getNamespace() . " AND " .
 160                           "old_title='" . wfStrencode( $this->mNewPage->getDBkey() ) .
 161                           "' ORDER BY inverse_timestamp LIMIT 1";
 162                         $res = wfQuery( $sql, DB_READ, $fname );
 163                 } else {
 164                         $sql = "SELECT old_namespace,old_title,old_timestamp,old_text,old_flags,old_user_text,old_comment FROM $oldtable WHERE " .
 165                           "old_id={$this->mOldid}";
 166                         $res = wfQuery( $sql, DB_READ, $fname );
 167                 }
 168                 if ( 0 == wfNumRows( $res ) ) { return false; }
 169
 170                 $s = wfFetchObject( $res );
 171                 $this->mOldPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 172                 $this->mOldtext = Article::getRevisionText( $s );
 173
 174                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 175                 $this->mOldtitle = wfMsg( "revisionasof", $t );
 176                 $this->mOldUser = $s->old_user_text;
 177                 $this->mOldComment = $s->old_comment;
 178
 179                 return true;
 180         }
 181 }
 182
 183 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 184 //
 185 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 186 // You may copy this code freely under the conditions of the GPL.
 187 //
 188
 189 define('USE_ASSERTS', function_exists('assert'));
 190
 191 class _DiffOp {
 192         var $type;
 193         var $orig;
 194         var $closing;
 195
 196         function reverse() {
 197                 trigger_error("pure virtual", E_USER_ERROR);
 198         }
 199
 200         function norig() {
 201                 return $this->orig ? sizeof($this->orig) : 0;
 202         }
 203
 204         function nclosing() {
 205                 return $this->closing ? sizeof($this->closing) : 0;
 206         }
 207 }
 208
 209 class _DiffOp_Copy extends _DiffOp {
 210         var $type = 'copy';
 211
 212         function _DiffOp_Copy ($orig, $closing = false) {
 213                 if (!is_array($closing))
 214                         $closing = $orig;
 215                 $this->orig = $orig;
 216                 $this->closing = $closing;
 217         }
 218
 219         function reverse() {
 220                 return new _DiffOp_Copy($this->closing, $this->orig);
 221         }
 222 }
 223
 224 class _DiffOp_Delete extends _DiffOp {
 225         var $type = 'delete';
 226
 227         function _DiffOp_Delete ($lines) {
 228                 $this->orig = $lines;
 229                 $this->closing = false;
 230         }
 231
 232         function reverse() {
 233                 return new _DiffOp_Add($this->orig);
 234         }
 235 }
 236
 237 class _DiffOp_Add extends _DiffOp {
 238         var $type = 'add';
 239
 240         function _DiffOp_Add ($lines) {
 241                 $this->closing = $lines;
 242                 $this->orig = false;
 243         }
 244
 245         function reverse() {
 246                 return new _DiffOp_Delete($this->closing);
 247         }
 248 }
 249
 250 class _DiffOp_Change extends _DiffOp {
 251         var $type = 'change';
 252
 253         function _DiffOp_Change ($orig, $closing) {
 254                 $this->orig = $orig;
 255                 $this->closing = $closing;
 256         }
 257
 258         function reverse() {
 259                 return new _DiffOp_Change($this->closing, $this->orig);
 260         }
 261 }
 262
 263
 264 /**
 265  * Class used internally by Diff to actually compute the diffs.
 266  *
 267  * The algorithm used here is mostly lifted from the perl module
 268  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 269  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 270  *
 271  * More ideas are taken from:
 272  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 273  *
 274  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 275  * diffutils-2.7, which can be found at:
 276  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 277  *
 278  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 279  * are my own.
 280  *
 281  * @author Geoffrey T. Dairiki
 282  * @access private
 283  */
 284 class _DiffEngine
 285 {
 286         function diff ($from_lines, $to_lines) {
 287                 $n_from = sizeof($from_lines);
 288                 $n_to = sizeof($to_lines);
 289
 290                 $this->xchanged = $this->ychanged = array();
 291                 $this->xv = $this->yv = array();
 292                 $this->xind = $this->yind = array();
 293                 unset($this->seq);
 294                 unset($this->in_seq);
 295                 unset($this->lcs);
 296
 297                 // Skip leading common lines.
 298                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 299                         if ($from_lines[$skip] != $to_lines[$skip])
 300                                 break;
 301                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 302                 }
 303                 // Skip trailing common lines.
 304                 $xi = $n_from; $yi = $n_to;
 305                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 306                         if ($from_lines[$xi] != $to_lines[$yi])
 307                                 break;
 308                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 309                 }
 310
 311                 // Ignore lines which do not exist in both files.
 312                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 313                         $xhash[$from_lines[$xi]] = 1;
 314                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 315                         $line = $to_lines[$yi];
 316                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 317                                 continue;
 318                         $yhash[$line] = 1;
 319                         $this->yv[] = $line;
 320                         $this->yind[] = $yi;
 321                 }
 322                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 323                         $line = $from_lines[$xi];
 324                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 325                                 continue;
 326                         $this->xv[] = $line;
 327                         $this->xind[] = $xi;
 328                 }
 329
 330                 // Find the LCS.
 331                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 332
 333                 // Merge edits when possible
 334                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 335                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 336
 337                 // Compute the edit operations.
 338                 $edits = array();
 339                 $xi = $yi = 0;
 340                 while ($xi < $n_from || $yi < $n_to) {
 341                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 342                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 343
 344                         // Skip matching "snake".
 345                         $copy = array();
 346                         while ( $xi < $n_from && $yi < $n_to
 347                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 348                                 $copy[] = $from_lines[$xi++];
 349                                 ++$yi;
 350                         }
 351                         if ($copy)
 352                                 $edits[] = new _DiffOp_Copy($copy);
 353
 354                         // Find deletes & adds.
 355                         $delete = array();
 356                         while ($xi < $n_from && $this->xchanged[$xi])
 357                                 $delete[] = $from_lines[$xi++];
 358
 359                         $add = array();
 360                         while ($yi < $n_to && $this->ychanged[$yi])
 361                                 $add[] = $to_lines[$yi++];
 362
 363                         if ($delete && $add)
 364                                 $edits[] = new _DiffOp_Change($delete, $add);
 365                         elseif ($delete)
 366                                 $edits[] = new _DiffOp_Delete($delete);
 367                         elseif ($add)
 368                                 $edits[] = new _DiffOp_Add($add);
 369                 }
 370                 return $edits;
 371         }
 372
 373
 374         /* Divide the Largest Common Subsequence (LCS) of the sequences
 375          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 376          * sized segments.
 377          *
 378          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 379          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 380          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 381          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 382          * that (X0, Y0) == (XOFF, YOFF) and
 383          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 384          *
 385          * This function assumes that the first lines of the specified portions
 386          * of the two files do not match, and likewise that the last lines do not
 387          * match.  The caller must trim matching lines from the beginning and end
 388          * of the portions it is going to specify.
 389          */
 390         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 391         $flip = false;
 392
 393         if ($xlim - $xoff > $ylim - $yoff) {
 394                 // Things seems faster (I'm not sure I understand why)
 395                         // when the shortest sequence in X.
 396                         $flip = true;
 397                 list ($xoff, $xlim, $yoff, $ylim)
 398                 = array( $yoff, $ylim, $xoff, $xlim);
 399                 }
 400
 401         if ($flip)
 402                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 403                 $ymatches[$this->xv[$i]][] = $i;
 404         else
 405                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 406                 $ymatches[$this->yv[$i]][] = $i;
 407
 408         $this->lcs = 0;
 409         $this->seq[0]= $yoff - 1;
 410         $this->in_seq = array();
 411         $ymids[0] = array();
 412
 413         $numer = $xlim - $xoff + $nchunks - 1;
 414         $x = $xoff;
 415         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 416                 if ($chunk > 0)
 417                 for ($i = 0; $i <= $this->lcs; $i++)
 418                         $ymids[$i][$chunk-1] = $this->seq[$i];
 419
 420                 $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 421                 for ( ; $x < $x1; $x++) {
 422                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 423                                 if (empty($ymatches[$line]))
 424                         continue;
 425                 $matches = $ymatches[$line];
 426                                 reset($matches);
 427                 while (list ($junk, $y) = each($matches))
 428                         if (empty($this->in_seq[$y])) {
 429                         $k = $this->_lcs_pos($y);
 430                         USE_ASSERTS && assert($k > 0);
 431                         $ymids[$k] = $ymids[$k-1];
 432                         break;
 433                                         }
 434                 while (list ($junk, $y) = each($matches)) {
 435                         if ($y > $this->seq[$k-1]) {
 436                         USE_ASSERTS && assert($y < $this->seq[$k]);
 437                         // Optimization: this is a common case:
 438                         //      next match is just replacing previous match.
 439                         $this->in_seq[$this->seq[$k]] = false;
 440                         $this->seq[$k] = $y;
 441                         $this->in_seq[$y] = 1;
 442                                         }
 443                         else if (empty($this->in_seq[$y])) {
 444                         $k = $this->_lcs_pos($y);
 445                         USE_ASSERTS && assert($k > 0);
 446                         $ymids[$k] = $ymids[$k-1];
 447                                         }
 448                                 }
 449                         }
 450                 }
 451
 452         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 453         $ymid = $ymids[$this->lcs];
 454         for ($n = 0; $n < $nchunks - 1; $n++) {
 455                 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 456                 $y1 = $ymid[$n] + 1;
 457                 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 458                 }
 459         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 460
 461         return array($this->lcs, $seps);
 462         }
 463
 464         function _lcs_pos ($ypos) {
 465         $end = $this->lcs;
 466         if ($end == 0 || $ypos > $this->seq[$end]) {
 467                 $this->seq[++$this->lcs] = $ypos;
 468                 $this->in_seq[$ypos] = 1;
 469                 return $this->lcs;
 470                 }
 471
 472         $beg = 1;
 473         while ($beg < $end) {
 474                 $mid = (int)(($beg + $end) / 2);
 475                 if ( $ypos > $this->seq[$mid] )
 476                 $beg = $mid + 1;
 477                 else
 478                 $end = $mid;
 479                 }
 480
 481         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 482
 483         $this->in_seq[$this->seq[$end]] = false;
 484         $this->seq[$end] = $ypos;
 485         $this->in_seq[$ypos] = 1;
 486         return $end;
 487         }
 488
 489         /* Find LCS of two sequences.
 490          *
 491          * The results are recorded in the vectors $this->{x,y}changed[], by
 492          * storing a 1 in the element for each line that is an insertion
 493          * or deletion (ie. is not in the LCS).
 494          *
 495          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 496          *
 497          * Note that XLIM, YLIM are exclusive bounds.
 498          * All line numbers are origin-0 and discarded lines are not counted.
 499          */
 500         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 501         // Slide down the bottom initial diagonal.
 502         while ($xoff < $xlim && $yoff < $ylim
 503                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 504                 ++$xoff;
 505                 ++$yoff;
 506                 }
 507
 508         // Slide up the top initial diagonal.
 509         while ($xlim > $xoff && $ylim > $yoff
 510                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 511                 --$xlim;
 512                 --$ylim;
 513                 }
 514
 515         if ($xoff == $xlim || $yoff == $ylim)
 516                 $lcs = 0;
 517         else {
 518                 // This is ad hoc but seems to work well.
 519                 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 520                 //$nchunks = max(2,min(8,(int)$nchunks));
 521                 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 522                 list ($lcs, $seps)
 523                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 524                 }
 525
 526         if ($lcs == 0) {
 527                 // X and Y sequences have no common subsequence:
 528                 // mark all changed.
 529                 while ($yoff < $ylim)
 530                 $this->ychanged[$this->yind[$yoff++]] = 1;
 531                 while ($xoff < $xlim)
 532                 $this->xchanged[$this->xind[$xoff++]] = 1;
 533                 }
 534         else {
 535                 // Use the partitions to split this problem into subproblems.
 536                 reset($seps);
 537                 $pt1 = $seps[0];
 538                 while ($pt2 = next($seps)) {
 539                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 540                 $pt1 = $pt2;
 541                         }
 542                 }
 543         }
 544
 545         /* Adjust inserts/deletes of identical lines to join changes
 546          * as much as possible.
 547          *
 548          * We do something when a run of changed lines include a
 549          * line at one end and has an excluded, identical line at the other.
 550          * We are free to choose which identical line is included.
 551          * `compareseq' usually chooses the one at the beginning,
 552          * but usually it is cleaner to consider the following identical line
 553          * to be the "change".
 554          *
 555          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 556          */
 557         function _shift_boundaries ($lines, &$changed, $other_changed) {
 558         $i = 0;
 559         $j = 0;
 560
 561         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 562         $len = sizeof($lines);
 563         $other_len = sizeof($other_changed);
 564
 565         while (1) {
 566                 /*
 567                  * Scan forwards to find beginning of another run of changes.
 568                  * Also keep track of the corresponding point in the other file.
 569                  *
 570                  * Throughout this code, $i and $j are adjusted together so that
 571                  * the first $i elements of $changed and the first $j elements
 572                  * of $other_changed both contain the same number of zeros
 573                  * (unchanged lines).
 574                  * Furthermore, $j is always kept so that $j == $other_len or
 575                  * $other_changed[$j] == false.
 576                  */
 577                 while ($j < $other_len && $other_changed[$j])
 578                 $j++;
 579
 580                 while ($i < $len && ! $changed[$i]) {
 581                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 582                 $i++; $j++;
 583                 while ($j < $other_len && $other_changed[$j])
 584                         $j++;
 585                         }
 586
 587                 if ($i == $len)
 588                 break;
 589
 590                 $start = $i;
 591
 592                 // Find the end of this run of changes.
 593                 while (++$i < $len && $changed[$i])
 594                 continue;
 595
 596                 do {
 597                 /*
 598                  * Record the length of this run of changes, so that
 599                  * we can later determine whether the run has grown.
 600                  */
 601                 $runlength = $i - $start;
 602
 603                 /*
 604                  * Move the changed region back, so long as the
 605                  * previous unchanged line matches the last changed one.
 606                  * This merges with previous changed regions.
 607                  */
 608                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 609                         $changed[--$start] = 1;
 610                         $changed[--$i] = false;
 611                         while ($start > 0 && $changed[$start - 1])
 612                         $start--;
 613                         USE_ASSERTS && assert('$j > 0');
 614                         while ($other_changed[--$j])
 615                         continue;
 616                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 617                                 }
 618
 619                 /*
 620                  * Set CORRESPONDING to the end of the changed run, at the last
 621                  * point where it corresponds to a changed run in the other file.
 622                  * CORRESPONDING == LEN means no such point has been found.
 623                  */
 624                 $corresponding = $j < $other_len ? $i : $len;
 625
 626                 /*
 627                  * Move the changed region forward, so long as the
 628                  * first changed line matches the following unchanged one.
 629                  * This merges with following changed regions.
 630                  * Do this second, so that if there are no merges,
 631                  * the changed region is moved forward as far as possible.
 632                  */
 633                 while ($i < $len && $lines[$start] == $lines[$i]) {
 634                         $changed[$start++] = false;
 635                         $changed[$i++] = 1;
 636                         while ($i < $len && $changed[$i])
 637                         $i++;
 638
 639                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 640                         $j++;
 641                         if ($j < $other_len && $other_changed[$j]) {
 642                         $corresponding = $i;
 643                         while ($j < $other_len && $other_changed[$j])
 644                                 $j++;
 645                                         }
 646                                 }
 647                         } while ($runlength != $i - $start);
 648
 649                 /*
 650                  * If possible, move the fully-merged run of changes
 651                  * back to a corresponding run in the other file.
 652                  */
 653                 while ($corresponding < $i) {
 654                 $changed[--$start] = 1;
 655                 $changed[--$i] = 0;
 656                 USE_ASSERTS && assert('$j > 0');
 657                 while ($other_changed[--$j])
 658                         continue;
 659                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 660                         }
 661                 }
 662         }
 663 }
 664
 665 /**
 666  * Class representing a 'diff' between two sequences of strings.
 667  */
 668 class Diff
 669 {
 670         var $edits;
 671
 672         /**
 673          * Constructor.
 674          * Computes diff between sequences of strings.
 675          *
 676          * @param $from_lines array An array of strings.
 677          *                (Typically these are lines from a file.)
 678          * @param $to_lines array An array of strings.
 679          */
 680         function Diff($from_lines, $to_lines) {
 681                 $eng = new _DiffEngine;
 682                 $this->edits = $eng->diff($from_lines, $to_lines);
 683                 //$this->_check($from_lines, $to_lines);
 684         }
 685
 686         /**
 687          * Compute reversed Diff.
 688          *
 689          * SYNOPSIS:
 690          *
 691          *      $diff = new Diff($lines1, $lines2);
 692          *      $rev = $diff->reverse();
 693          * @return object A Diff object representing the inverse of the
 694          *                                original diff.
 695          */
 696         function reverse () {
 697         $rev = $this;
 698                 $rev->edits = array();
 699                 foreach ($this->edits as $edit) {
 700                         $rev->edits[] = $edit->reverse();
 701                 }
 702         return $rev;
 703         }
 704
 705         /**
 706          * Check for empty diff.
 707          *
 708          * @return bool True iff two sequences were identical.
 709          */
 710         function isEmpty () {
 711                 foreach ($this->edits as $edit) {
 712                         if ($edit->type != 'copy')
 713                                 return false;
 714                 }
 715                 return true;
 716         }
 717
 718         /**
 719          * Compute the length of the Longest Common Subsequence (LCS).
 720          *
 721          * This is mostly for diagnostic purposed.
 722          *
 723          * @return int The length of the LCS.
 724          */
 725         function lcs () {
 726         $lcs = 0;
 727                 foreach ($this->edits as $edit) {
 728                         if ($edit->type == 'copy')
 729                                 $lcs += sizeof($edit->orig);
 730                 }
 731         return $lcs;
 732         }
 733
 734         /**
 735          * Get the original set of lines.
 736          *
 737          * This reconstructs the $from_lines parameter passed to the
 738          * constructor.
 739          *
 740          * @return array The original sequence of strings.
 741          */
 742         function orig() {
 743                 $lines = array();
 744
 745                 foreach ($this->edits as $edit) {
 746                         if ($edit->orig)
 747                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 748                 }
 749                 return $lines;
 750         }
 751
 752         /**
 753          * Get the closing set of lines.
 754          *
 755          * This reconstructs the $to_lines parameter passed to the
 756          * constructor.
 757          *
 758          * @return array The sequence of strings.
 759          */
 760         function closing() {
 761                 $lines = array();
 762
 763                 foreach ($this->edits as $edit) {
 764                         if ($edit->closing)
 765                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 766                 }
 767                 return $lines;
 768         }
 769
 770         /**
 771          * Check a Diff for validity.
 772          *
 773          * This is here only for debugging purposes.
 774          */
 775         function _check ($from_lines, $to_lines) {
 776                 if (serialize($from_lines) != serialize($this->orig()))
 777                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 778                 if (serialize($to_lines) != serialize($this->closing()))
 779                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 780
 781                 $rev = $this->reverse();
 782                 if (serialize($to_lines) != serialize($rev->orig()))
 783                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 784                 if (serialize($from_lines) != serialize($rev->closing()))
 785                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 786
 787
 788                 $prevtype = 'none';
 789                 foreach ($this->edits as $edit) {
 790                         if ( $prevtype == $edit->type )
 791                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 792                         $prevtype = $edit->type;
 793                 }
 794
 795                 $lcs = $this->lcs();
 796                 trigger_error("Diff okay: LCS = $lcs", E_USER_NOTICE);
 797         }
 798 }
 799
 800 /**
 801  * FIXME: bad name.
 802  */
 803 class MappedDiff
 804 extends Diff
 805 {
 806         /**
 807          * Constructor.
 808          *
 809          * Computes diff between sequences of strings.
 810          *
 811          * This can be used to compute things like
 812          * case-insensitve diffs, or diffs which ignore
 813          * changes in white-space.
 814          *
 815          * @param $from_lines array An array of strings.
 816          *      (Typically these are lines from a file.)
 817          *
 818          * @param $to_lines array An array of strings.
 819          *
 820          * @param $mapped_from_lines array This array should
 821          *      have the same size number of elements as $from_lines.
 822          *      The elements in $mapped_from_lines and
 823          *      $mapped_to_lines are what is actually compared
 824          *      when computing the diff.
 825          *
 826          * @param $mapped_to_lines array This array should
 827          *      have the same number of elements as $to_lines.
 828          */
 829         function MappedDiff($from_lines, $to_lines,
 830                                                 $mapped_from_lines, $mapped_to_lines) {
 831
 832                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 833                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 834
 835                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 836
 837                 $xi = $yi = 0;
 838                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 839                         $orig = &$this->edits[$i]->orig;
 840                         if (is_array($orig)) {
 841                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 842                                 $xi += sizeof($orig);
 843                         }
 844
 845                         $closing = &$this->edits[$i]->closing;
 846                         if (is_array($closing)) {
 847                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 848                                 $yi += sizeof($closing);
 849                         }
 850                 }
 851         }
 852 }
 853
 854 /**
 855  * A class to format Diffs
 856  *
 857  * This class formats the diff in classic diff format.
 858  * It is intended that this class be customized via inheritance,
 859  * to obtain fancier outputs.
 860  */
 861 class DiffFormatter
 862 {
 863         /**
 864          * Number of leading context "lines" to preserve.
 865          *
 866          * This should be left at zero for this class, but subclasses
 867          * may want to set this to other values.
 868          */
 869         var $leading_context_lines = 0;
 870
 871         /**
 872          * Number of trailing context "lines" to preserve.
 873          *
 874          * This should be left at zero for this class, but subclasses
 875          * may want to set this to other values.
 876          */
 877         var $trailing_context_lines = 0;
 878
 879         /**
 880          * Format a diff.
 881          *
 882          * @param $diff object A Diff object.
 883          * @return string The formatted output.
 884          */
 885         function format($diff) {
 886
 887                 $xi = $yi = 1;
 888                 $block = false;
 889                 $context = array();
 890
 891                 $nlead = $this->leading_context_lines;
 892                 $ntrail = $this->trailing_context_lines;
 893
 894                 $this->_start_diff();
 895
 896                 foreach ($diff->edits as $edit) {
 897                         if ($edit->type == 'copy') {
 898                                 if (is_array($block)) {
 899                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 900                                                 $block[] = $edit;
 901                                         }
 902                                         else{
 903                                                 if ($ntrail) {
 904                                                         $context = array_slice($edit->orig, 0, $ntrail);
 905                                                         $block[] = new _DiffOp_Copy($context);
 906                                                 }
 907                                                 $this->_block($x0, $ntrail + $xi - $x0,
 908                                                                           $y0, $ntrail + $yi - $y0,
 909                                                                           $block);
 910                                                 $block = false;
 911                                         }
 912                                 }
 913                                 $context = $edit->orig;
 914                         }
 915                         else {
 916                                 if (! is_array($block)) {
 917                                         $context = array_slice($context, sizeof($context) - $nlead);
 918                                         $x0 = $xi - sizeof($context);
 919                                         $y0 = $yi - sizeof($context);
 920                                         $block = array();
 921                                         if ($context)
 922                                                 $block[] = new _DiffOp_Copy($context);
 923                                 }
 924                                 $block[] = $edit;
 925                         }
 926
 927                         if ($edit->orig)
 928                                 $xi += sizeof($edit->orig);
 929                         if ($edit->closing)
 930                                 $yi += sizeof($edit->closing);
 931                 }
 932
 933                 if (is_array($block))
 934                         $this->_block($x0, $xi - $x0,
 935                                                   $y0, $yi - $y0,
 936                                                   $block);
 937
 938                 return $this->_end_diff();
 939         }
 940
 941         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 942                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 943                 foreach ($edits as $edit) {
 944                         if ($edit->type == 'copy')
 945                                 $this->_context($edit->orig);
 946                         elseif ($edit->type == 'add')
 947                                 $this->_added($edit->closing);
 948                         elseif ($edit->type == 'delete')
 949                                 $this->_deleted($edit->orig);
 950                         elseif ($edit->type == 'change')
 951                                 $this->_changed($edit->orig, $edit->closing);
 952                         else
 953                                 trigger_error("Unknown edit type", E_USER_ERROR);
 954                 }
 955                 $this->_end_block();
 956         }
 957
 958         function _start_diff() {
 959                 ob_start();
 960         }
 961
 962         function _end_diff() {
 963                 $val = ob_get_contents();
 964                 ob_end_clean();
 965                 return $val;
 966         }
 967
 968         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 969                 if ($xlen > 1)
 970                         $xbeg .= "," . ($xbeg + $xlen - 1);
 971                 if ($ylen > 1)
 972                         $ybeg .= "," . ($ybeg + $ylen - 1);
 973
 974                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 975         }
 976
 977         function _start_block($header) {
 978                 echo $header;
 979         }
 980
 981         function _end_block() {
 982         }
 983
 984         function _lines($lines, $prefix = ' ') {
 985                 foreach ($lines as $line)
 986                         echo "$prefix $line\n";
 987         }
 988
 989         function _context($lines) {
 990                 $this->_lines($lines);
 991         }
 992
 993         function _added($lines) {
 994                 $this->_lines($lines, ">");
 995         }
 996         function _deleted($lines) {
 997                 $this->_lines($lines, "<");
 998         }
 999
1000         function _changed($orig, $closing) {
1001                 $this->_deleted($orig);
1002                 echo "---\n";
1003                 $this->_added($closing);
1004         }
1005 }
1006
1007
1008 /**
1009  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1010  *
1011  */
1012
1013 define('NBSP', "\xA0");                 // iso-8859-x non-breaking space.
1014
1015 class _HWLDF_WordAccumulator {
1016         function _HWLDF_WordAccumulator () {
1017                 $this->_lines = array();
1018                 $this->_line = '';
1019                 $this->_group = '';
1020                 $this->_tag = '';
1021         }
1022
1023         function _flushGroup ($new_tag) {
1024                 if ($this->_group !== '') {
1025           if ($this->_tag == 'mark')
1026                         $this->_line .= '<span class="diffchange">'.$this->_group.'</span>';
1027           else
1028                 $this->_line .= $this->_group;
1029         }
1030                 $this->_group = '';
1031                 $this->_tag = $new_tag;
1032         }
1033
1034         function _flushLine ($new_tag) {
1035                 $this->_flushGroup($new_tag);
1036                 if ($this->_line != '')
1037                         $this->_lines[] = $this->_line;
1038                 $this->_line = '';
1039         }
1040
1041         function addWords ($words, $tag = '') {
1042                 if ($tag != $this->_tag)
1043                         $this->_flushGroup($tag);
1044
1045                 foreach ($words as $word) {
1046                         // new-line should only come as first char of word.
1047                         if ($word == '')
1048                                 continue;
1049                         if ($word[0] == "\n") {
1050                                 $this->_group .= NBSP;
1051                                 $this->_flushLine($tag);
1052                                 $word = substr($word, 1);
1053                         }
1054                         assert(!strstr($word, "\n"));
1055                         $this->_group .= $word;
1056                 }
1057         }
1058
1059         function getLines() {
1060                 $this->_flushLine('~done');
1061                 return $this->_lines;
1062         }
1063 }
1064
1065 class WordLevelDiff extends MappedDiff
1066 {
1067         function WordLevelDiff ($orig_lines, $closing_lines) {
1068                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1069                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1070
1071
1072                 $this->MappedDiff($orig_words, $closing_words,
1073                                                   $orig_stripped, $closing_stripped);
1074         }
1075
1076         function _split($lines) {
1077                 // FIXME: fix POSIX char class.
1078 #                if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1079                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1080                                                         implode("\n", $lines),
1081                                                         $m)) {
1082                         return array(array(''), array(''));
1083                 }
1084                 return array($m[0], $m[1]);
1085         }
1086
1087         function orig () {
1088                 $orig = new _HWLDF_WordAccumulator;
1089
1090                 foreach ($this->edits as $edit) {
1091                         if ($edit->type == 'copy')
1092                                 $orig->addWords($edit->orig);
1093                         elseif ($edit->orig)
1094                                 $orig->addWords($edit->orig, 'mark');
1095                 }
1096                 return $orig->getLines();
1097         }
1098
1099         function closing () {
1100                 $closing = new _HWLDF_WordAccumulator;
1101
1102                 foreach ($this->edits as $edit) {
1103                         if ($edit->type == 'copy')
1104                                 $closing->addWords($edit->closing);
1105                         elseif ($edit->closing)
1106                                 $closing->addWords($edit->closing, 'mark');
1107                 }
1108                 return $closing->getLines();
1109         }
1110 }
1111
1112 /**
1113  *      Wikipedia Table style diff formatter.
1114  *
1115  */
1116 class TableDiffFormatter extends DiffFormatter
1117 {
1118         function TableDiffFormatter() {
1119                 $this->leading_context_lines = 2;
1120                 $this->trailing_context_lines = 2;
1121         }
1122
1123         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1124                 $l1 = wfMsg( "lineno", $xbeg );
1125                 $l2 = wfMsg( "lineno", $ybeg );
1126
1127                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1128                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1129                 return $r;
1130         }
1131
1132         function _start_block( $header ) {
1133                 global $wgOut;
1134                 $wgOut->addHTML( $header );
1135         }
1136
1137         function _end_block() {
1138         }
1139
1140         function _lines( $lines, $prefix=' ', $color="white" ) {
1141         }
1142
1143         function addedLine( $line ) {
1144                 return '<td>+</td><td class="diff-addedline">' .
1145                   $line.'</td>';
1146         }
1147
1148         function deletedLine( $line ) {
1149                 return '<td>-</td><td class="diff-deletedline">' .
1150                   $line.'</td>';
1151         }
1152
1153         function emptyLine() {
1154                 return '<td colspan="2">&nbsp;</td>';
1155         }
1156
1157         function contextLine( $line ) {
1158                 return '<td> </td><td class="diff-context">'.$line.'</td>';
1159         }
1160
1161         function _added($lines) {
1162                 global $wgOut;
1163                 foreach ($lines as $line) {
1164                         $wgOut->addHTML( '<tr>' . $this->emptyLine() .
1165                           $this->addedLine( $line ) . "</tr>\n" );
1166                 }
1167         }
1168
1169         function _deleted($lines) {
1170                 global $wgOut;
1171                 foreach ($lines as $line) {
1172                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1173                           $this->emptyLine() . "</tr>\n" );
1174                 }
1175         }
1176
1177         function _context( $lines ) {
1178                 global $wgOut;
1179                 foreach ($lines as $line) {
1180                         $wgOut->addHTML( '<tr>' . $this->contextLine( $line ) .
1181                           $this->contextLine( $line ) . "</tr>\n" );
1182                 }
1183         }
1184
1185         function _changed( $orig, $closing ) {
1186                 global $wgOut;
1187                 $diff = new WordLevelDiff( $orig, $closing );
1188                 $del = $diff->orig();
1189                 $add = $diff->closing();
1190
1191                 while ( $line = array_shift( $del ) ) {
1192                         $aline = array_shift( $add );
1193                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1194                           $this->addedLine( $aline ) . "</tr>\n" );
1195                 }
1196                 $this->_added( $add ); # If any leftovers
1197         }
1198 }
1199
1200 ?>