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;
 122                 $fname = "DifferenceEngine::loadText";
 123
 124                 $dbr =& wfGetDB( DB_SLAVE );
 125                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 126                         $wgOut->setArticleFlag( true );
 127                         $this->mNewtitle = wfMsg( "currentrev" );
 128                         $id = $wgTitle->getArticleID();
 129
 130                         $s = $dbr->getArray( 'cur', array( 'cur_text', 'cur_user_text', 'cur_comment' ),
 131                                 array( 'cur_id' => $id ), $fname );
 132                         if ( $s === false ) {
 133                                 return false;
 134                         }
 135
 136                         $this->mNewPage = &$wgTitle;
 137                         $this->mNewtext = $s->cur_text;
 138                         $this->mNewUser = $s->cur_user_text;
 139                         $this->mNewComment = $s->cur_comment;
 140                 } else {
 141                         $s = $dbr->getArray( 'old', array( 'old_namespace','old_title','old_timestamp', 'old_text',
 142                                 'old_flags','old_user_text','old_comment' ), array( 'old_id' => $this->mNewid ), $fname );
 143
 144                         if ( $s === false ) {
 145                                 return false;
 146                         }
 147
 148                         $this->mNewtext = Article::getRevisionText( $s );
 149
 150                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 151                         $this->mNewPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 152                         $this->mNewtitle = wfMsg( "revisionasof", $t );
 153                         $this->mNewUser = $s->old_user_text;
 154                         $this->mNewComment = $s->old_comment;
 155                 }
 156                 if ( 0 == $this->mOldid ) {
 157                         $s = $dbr->getArray( 'old',
 158                                 array( 'old_namespace','old_title','old_timestamp','old_text', 'old_flags','old_user_text','old_comment' ),
 159                                 array( /* WHERE */
 160                                         'old_namespace' => $this->mNewPage->getNamespace(),
 161                                         'old_title' => $this->mNewPage->getDBkey()
 162                                 ), $fname, array( 'ORDER BY' => 'inverse_timestamp', 'USE INDEX' => 'name_title_timestamp' )
 163                         );
 164                 } else {
 165                         $s = $dbr->getArray( 'old',
 166                                 array( 'old_namespace','old_title','old_timestamp','old_text','old_flags','old_user_text','old_comment'),
 167                                 array( 'old_id' => $this->mOldid ),
 168                                 $fname
 169                         );
 170                 }
 171                 if ( $s === false ) {
 172                         return false;
 173                 }
 174                 $this->mOldPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 175                 $this->mOldtext = Article::getRevisionText( $s );
 176
 177                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 178                 $this->mOldtitle = wfMsg( "revisionasof", $t );
 179                 $this->mOldUser = $s->old_user_text;
 180                 $this->mOldComment = $s->old_comment;
 181
 182                 return true;
 183         }
 184 }
 185
 186 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 187 //
 188 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 189 // You may copy this code freely under the conditions of the GPL.
 190 //
 191
 192 define('USE_ASSERTS', function_exists('assert'));
 193
 194 class _DiffOp {
 195         var $type;
 196         var $orig;
 197         var $closing;
 198
 199         function reverse() {
 200                 trigger_error("pure virtual", E_USER_ERROR);
 201         }
 202
 203         function norig() {
 204                 return $this->orig ? sizeof($this->orig) : 0;
 205         }
 206
 207         function nclosing() {
 208                 return $this->closing ? sizeof($this->closing) : 0;
 209         }
 210 }
 211
 212 class _DiffOp_Copy extends _DiffOp {
 213         var $type = 'copy';
 214
 215         function _DiffOp_Copy ($orig, $closing = false) {
 216                 if (!is_array($closing))
 217                         $closing = $orig;
 218                 $this->orig = $orig;
 219                 $this->closing = $closing;
 220         }
 221
 222         function reverse() {
 223                 return new _DiffOp_Copy($this->closing, $this->orig);
 224         }
 225 }
 226
 227 class _DiffOp_Delete extends _DiffOp {
 228         var $type = 'delete';
 229
 230         function _DiffOp_Delete ($lines) {
 231                 $this->orig = $lines;
 232                 $this->closing = false;
 233         }
 234
 235         function reverse() {
 236                 return new _DiffOp_Add($this->orig);
 237         }
 238 }
 239
 240 class _DiffOp_Add extends _DiffOp {
 241         var $type = 'add';
 242
 243         function _DiffOp_Add ($lines) {
 244                 $this->closing = $lines;
 245                 $this->orig = false;
 246         }
 247
 248         function reverse() {
 249                 return new _DiffOp_Delete($this->closing);
 250         }
 251 }
 252
 253 class _DiffOp_Change extends _DiffOp {
 254         var $type = 'change';
 255
 256         function _DiffOp_Change ($orig, $closing) {
 257                 $this->orig = $orig;
 258                 $this->closing = $closing;
 259         }
 260
 261         function reverse() {
 262                 return new _DiffOp_Change($this->closing, $this->orig);
 263         }
 264 }
 265
 266
 267 /**
 268  * Class used internally by Diff to actually compute the diffs.
 269  *
 270  * The algorithm used here is mostly lifted from the perl module
 271  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 272  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 273  *
 274  * More ideas are taken from:
 275  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 276  *
 277  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 278  * diffutils-2.7, which can be found at:
 279  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 280  *
 281  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 282  * are my own.
 283  *
 284  * @author Geoffrey T. Dairiki
 285  * @access private
 286  */
 287 class _DiffEngine
 288 {
 289         function diff ($from_lines, $to_lines) {
 290                 $n_from = sizeof($from_lines);
 291                 $n_to = sizeof($to_lines);
 292
 293                 $this->xchanged = $this->ychanged = array();
 294                 $this->xv = $this->yv = array();
 295                 $this->xind = $this->yind = array();
 296                 unset($this->seq);
 297                 unset($this->in_seq);
 298                 unset($this->lcs);
 299
 300                 // Skip leading common lines.
 301                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 302                         if ($from_lines[$skip] != $to_lines[$skip])
 303                                 break;
 304                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 305                 }
 306                 // Skip trailing common lines.
 307                 $xi = $n_from; $yi = $n_to;
 308                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 309                         if ($from_lines[$xi] != $to_lines[$yi])
 310                                 break;
 311                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 312                 }
 313
 314                 // Ignore lines which do not exist in both files.
 315                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 316                         $xhash[$from_lines[$xi]] = 1;
 317                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 318                         $line = $to_lines[$yi];
 319                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 320                                 continue;
 321                         $yhash[$line] = 1;
 322                         $this->yv[] = $line;
 323                         $this->yind[] = $yi;
 324                 }
 325                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 326                         $line = $from_lines[$xi];
 327                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 328                                 continue;
 329                         $this->xv[] = $line;
 330                         $this->xind[] = $xi;
 331                 }
 332
 333                 // Find the LCS.
 334                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 335
 336                 // Merge edits when possible
 337                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 338                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 339
 340                 // Compute the edit operations.
 341                 $edits = array();
 342                 $xi = $yi = 0;
 343                 while ($xi < $n_from || $yi < $n_to) {
 344                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 345                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 346
 347                         // Skip matching "snake".
 348                         $copy = array();
 349                         while ( $xi < $n_from && $yi < $n_to
 350                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 351                                 $copy[] = $from_lines[$xi++];
 352                                 ++$yi;
 353                         }
 354                         if ($copy)
 355                                 $edits[] = new _DiffOp_Copy($copy);
 356
 357                         // Find deletes & adds.
 358                         $delete = array();
 359                         while ($xi < $n_from && $this->xchanged[$xi])
 360                                 $delete[] = $from_lines[$xi++];
 361
 362                         $add = array();
 363                         while ($yi < $n_to && $this->ychanged[$yi])
 364                                 $add[] = $to_lines[$yi++];
 365
 366                         if ($delete && $add)
 367                                 $edits[] = new _DiffOp_Change($delete, $add);
 368                         elseif ($delete)
 369                                 $edits[] = new _DiffOp_Delete($delete);
 370                         elseif ($add)
 371                                 $edits[] = new _DiffOp_Add($add);
 372                 }
 373                 return $edits;
 374         }
 375
 376
 377         /* Divide the Largest Common Subsequence (LCS) of the sequences
 378          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 379          * sized segments.
 380          *
 381          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 382          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 383          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 384          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 385          * that (X0, Y0) == (XOFF, YOFF) and
 386          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 387          *
 388          * This function assumes that the first lines of the specified portions
 389          * of the two files do not match, and likewise that the last lines do not
 390          * match.  The caller must trim matching lines from the beginning and end
 391          * of the portions it is going to specify.
 392          */
 393         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 394         $flip = false;
 395
 396         if ($xlim - $xoff > $ylim - $yoff) {
 397                 // Things seems faster (I'm not sure I understand why)
 398                         // when the shortest sequence in X.
 399                         $flip = true;
 400                 list ($xoff, $xlim, $yoff, $ylim)
 401                 = array( $yoff, $ylim, $xoff, $xlim);
 402                 }
 403
 404         if ($flip)
 405                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 406                 $ymatches[$this->xv[$i]][] = $i;
 407         else
 408                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 409                 $ymatches[$this->yv[$i]][] = $i;
 410
 411         $this->lcs = 0;
 412         $this->seq[0]= $yoff - 1;
 413         $this->in_seq = array();
 414         $ymids[0] = array();
 415
 416         $numer = $xlim - $xoff + $nchunks - 1;
 417         $x = $xoff;
 418         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 419                 if ($chunk > 0)
 420                 for ($i = 0; $i <= $this->lcs; $i++)
 421                         $ymids[$i][$chunk-1] = $this->seq[$i];
 422
 423                 $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 424                 for ( ; $x < $x1; $x++) {
 425                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 426                                 if (empty($ymatches[$line]))
 427                         continue;
 428                 $matches = $ymatches[$line];
 429                                 reset($matches);
 430                 while (list ($junk, $y) = each($matches))
 431                         if (empty($this->in_seq[$y])) {
 432                         $k = $this->_lcs_pos($y);
 433                         USE_ASSERTS && assert($k > 0);
 434                         $ymids[$k] = $ymids[$k-1];
 435                         break;
 436                                         }
 437                 while (list ($junk, $y) = each($matches)) {
 438                         if ($y > $this->seq[$k-1]) {
 439                         USE_ASSERTS && assert($y < $this->seq[$k]);
 440                         // Optimization: this is a common case:
 441                         //      next match is just replacing previous match.
 442                         $this->in_seq[$this->seq[$k]] = false;
 443                         $this->seq[$k] = $y;
 444                         $this->in_seq[$y] = 1;
 445                                         }
 446                         else if (empty($this->in_seq[$y])) {
 447                         $k = $this->_lcs_pos($y);
 448                         USE_ASSERTS && assert($k > 0);
 449                         $ymids[$k] = $ymids[$k-1];
 450                                         }
 451                                 }
 452                         }
 453                 }
 454
 455         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 456         $ymid = $ymids[$this->lcs];
 457         for ($n = 0; $n < $nchunks - 1; $n++) {
 458                 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 459                 $y1 = $ymid[$n] + 1;
 460                 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 461                 }
 462         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 463
 464         return array($this->lcs, $seps);
 465         }
 466
 467         function _lcs_pos ($ypos) {
 468         $end = $this->lcs;
 469         if ($end == 0 || $ypos > $this->seq[$end]) {
 470                 $this->seq[++$this->lcs] = $ypos;
 471                 $this->in_seq[$ypos] = 1;
 472                 return $this->lcs;
 473                 }
 474
 475         $beg = 1;
 476         while ($beg < $end) {
 477                 $mid = (int)(($beg + $end) / 2);
 478                 if ( $ypos > $this->seq[$mid] )
 479                 $beg = $mid + 1;
 480                 else
 481                 $end = $mid;
 482                 }
 483
 484         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 485
 486         $this->in_seq[$this->seq[$end]] = false;
 487         $this->seq[$end] = $ypos;
 488         $this->in_seq[$ypos] = 1;
 489         return $end;
 490         }
 491
 492         /* Find LCS of two sequences.
 493          *
 494          * The results are recorded in the vectors $this->{x,y}changed[], by
 495          * storing a 1 in the element for each line that is an insertion
 496          * or deletion (ie. is not in the LCS).
 497          *
 498          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 499          *
 500          * Note that XLIM, YLIM are exclusive bounds.
 501          * All line numbers are origin-0 and discarded lines are not counted.
 502          */
 503         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 504         // Slide down the bottom initial diagonal.
 505         while ($xoff < $xlim && $yoff < $ylim
 506                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 507                 ++$xoff;
 508                 ++$yoff;
 509                 }
 510
 511         // Slide up the top initial diagonal.
 512         while ($xlim > $xoff && $ylim > $yoff
 513                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 514                 --$xlim;
 515                 --$ylim;
 516                 }
 517
 518         if ($xoff == $xlim || $yoff == $ylim)
 519                 $lcs = 0;
 520         else {
 521                 // This is ad hoc but seems to work well.
 522                 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 523                 //$nchunks = max(2,min(8,(int)$nchunks));
 524                 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 525                 list ($lcs, $seps)
 526                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 527                 }
 528
 529         if ($lcs == 0) {
 530                 // X and Y sequences have no common subsequence:
 531                 // mark all changed.
 532                 while ($yoff < $ylim)
 533                 $this->ychanged[$this->yind[$yoff++]] = 1;
 534                 while ($xoff < $xlim)
 535                 $this->xchanged[$this->xind[$xoff++]] = 1;
 536                 }
 537         else {
 538                 // Use the partitions to split this problem into subproblems.
 539                 reset($seps);
 540                 $pt1 = $seps[0];
 541                 while ($pt2 = next($seps)) {
 542                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 543                 $pt1 = $pt2;
 544                         }
 545                 }
 546         }
 547
 548         /* Adjust inserts/deletes of identical lines to join changes
 549          * as much as possible.
 550          *
 551          * We do something when a run of changed lines include a
 552          * line at one end and has an excluded, identical line at the other.
 553          * We are free to choose which identical line is included.
 554          * `compareseq' usually chooses the one at the beginning,
 555          * but usually it is cleaner to consider the following identical line
 556          * to be the "change".
 557          *
 558          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 559          */
 560         function _shift_boundaries ($lines, &$changed, $other_changed) {
 561         $i = 0;
 562         $j = 0;
 563
 564         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 565         $len = sizeof($lines);
 566         $other_len = sizeof($other_changed);
 567
 568         while (1) {
 569                 /*
 570                  * Scan forwards to find beginning of another run of changes.
 571                  * Also keep track of the corresponding point in the other file.
 572                  *
 573                  * Throughout this code, $i and $j are adjusted together so that
 574                  * the first $i elements of $changed and the first $j elements
 575                  * of $other_changed both contain the same number of zeros
 576                  * (unchanged lines).
 577                  * Furthermore, $j is always kept so that $j == $other_len or
 578                  * $other_changed[$j] == false.
 579                  */
 580                 while ($j < $other_len && $other_changed[$j])
 581                 $j++;
 582
 583                 while ($i < $len && ! $changed[$i]) {
 584                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 585                 $i++; $j++;
 586                 while ($j < $other_len && $other_changed[$j])
 587                         $j++;
 588                         }
 589
 590                 if ($i == $len)
 591                 break;
 592
 593                 $start = $i;
 594
 595                 // Find the end of this run of changes.
 596                 while (++$i < $len && $changed[$i])
 597                 continue;
 598
 599                 do {
 600                 /*
 601                  * Record the length of this run of changes, so that
 602                  * we can later determine whether the run has grown.
 603                  */
 604                 $runlength = $i - $start;
 605
 606                 /*
 607                  * Move the changed region back, so long as the
 608                  * previous unchanged line matches the last changed one.
 609                  * This merges with previous changed regions.
 610                  */
 611                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 612                         $changed[--$start] = 1;
 613                         $changed[--$i] = false;
 614                         while ($start > 0 && $changed[$start - 1])
 615                         $start--;
 616                         USE_ASSERTS && assert('$j > 0');
 617                         while ($other_changed[--$j])
 618                         continue;
 619                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 620                                 }
 621
 622                 /*
 623                  * Set CORRESPONDING to the end of the changed run, at the last
 624                  * point where it corresponds to a changed run in the other file.
 625                  * CORRESPONDING == LEN means no such point has been found.
 626                  */
 627                 $corresponding = $j < $other_len ? $i : $len;
 628
 629                 /*
 630                  * Move the changed region forward, so long as the
 631                  * first changed line matches the following unchanged one.
 632                  * This merges with following changed regions.
 633                  * Do this second, so that if there are no merges,
 634                  * the changed region is moved forward as far as possible.
 635                  */
 636                 while ($i < $len && $lines[$start] == $lines[$i]) {
 637                         $changed[$start++] = false;
 638                         $changed[$i++] = 1;
 639                         while ($i < $len && $changed[$i])
 640                         $i++;
 641
 642                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 643                         $j++;
 644                         if ($j < $other_len && $other_changed[$j]) {
 645                         $corresponding = $i;
 646                         while ($j < $other_len && $other_changed[$j])
 647                                 $j++;
 648                                         }
 649                                 }
 650                         } while ($runlength != $i - $start);
 651
 652                 /*
 653                  * If possible, move the fully-merged run of changes
 654                  * back to a corresponding run in the other file.
 655                  */
 656                 while ($corresponding < $i) {
 657                 $changed[--$start] = 1;
 658                 $changed[--$i] = 0;
 659                 USE_ASSERTS && assert('$j > 0');
 660                 while ($other_changed[--$j])
 661                         continue;
 662                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 663                         }
 664                 }
 665         }
 666 }
 667
 668 /**
 669  * Class representing a 'diff' between two sequences of strings.
 670  */
 671 class Diff
 672 {
 673         var $edits;
 674
 675         /**
 676          * Constructor.
 677          * Computes diff between sequences of strings.
 678          *
 679          * @param $from_lines array An array of strings.
 680          *                (Typically these are lines from a file.)
 681          * @param $to_lines array An array of strings.
 682          */
 683         function Diff($from_lines, $to_lines) {
 684                 $eng = new _DiffEngine;
 685                 $this->edits = $eng->diff($from_lines, $to_lines);
 686                 //$this->_check($from_lines, $to_lines);
 687         }
 688
 689         /**
 690          * Compute reversed Diff.
 691          *
 692          * SYNOPSIS:
 693          *
 694          *      $diff = new Diff($lines1, $lines2);
 695          *      $rev = $diff->reverse();
 696          * @return object A Diff object representing the inverse of the
 697          *                                original diff.
 698          */
 699         function reverse () {
 700         $rev = $this;
 701                 $rev->edits = array();
 702                 foreach ($this->edits as $edit) {
 703                         $rev->edits[] = $edit->reverse();
 704                 }
 705         return $rev;
 706         }
 707
 708         /**
 709          * Check for empty diff.
 710          *
 711          * @return bool True iff two sequences were identical.
 712          */
 713         function isEmpty () {
 714                 foreach ($this->edits as $edit) {
 715                         if ($edit->type != 'copy')
 716                                 return false;
 717                 }
 718                 return true;
 719         }
 720
 721         /**
 722          * Compute the length of the Longest Common Subsequence (LCS).
 723          *
 724          * This is mostly for diagnostic purposed.
 725          *
 726          * @return int The length of the LCS.
 727          */
 728         function lcs () {
 729         $lcs = 0;
 730                 foreach ($this->edits as $edit) {
 731                         if ($edit->type == 'copy')
 732                                 $lcs += sizeof($edit->orig);
 733                 }
 734         return $lcs;
 735         }
 736
 737         /**
 738          * Get the original set of lines.
 739          *
 740          * This reconstructs the $from_lines parameter passed to the
 741          * constructor.
 742          *
 743          * @return array The original sequence of strings.
 744          */
 745         function orig() {
 746                 $lines = array();
 747
 748                 foreach ($this->edits as $edit) {
 749                         if ($edit->orig)
 750                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 751                 }
 752                 return $lines;
 753         }
 754
 755         /**
 756          * Get the closing set of lines.
 757          *
 758          * This reconstructs the $to_lines parameter passed to the
 759          * constructor.
 760          *
 761          * @return array The sequence of strings.
 762          */
 763         function closing() {
 764                 $lines = array();
 765
 766                 foreach ($this->edits as $edit) {
 767                         if ($edit->closing)
 768                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 769                 }
 770                 return $lines;
 771         }
 772
 773         /**
 774          * Check a Diff for validity.
 775          *
 776          * This is here only for debugging purposes.
 777          */
 778         function _check ($from_lines, $to_lines) {
 779                 if (serialize($from_lines) != serialize($this->orig()))
 780                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 781                 if (serialize($to_lines) != serialize($this->closing()))
 782                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 783
 784                 $rev = $this->reverse();
 785                 if (serialize($to_lines) != serialize($rev->orig()))
 786                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 787                 if (serialize($from_lines) != serialize($rev->closing()))
 788                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 789
 790
 791                 $prevtype = 'none';
 792                 foreach ($this->edits as $edit) {
 793                         if ( $prevtype == $edit->type )
 794                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 795                         $prevtype = $edit->type;
 796                 }
 797
 798                 $lcs = $this->lcs();
 799                 trigger_error("Diff okay: LCS = $lcs", E_USER_NOTICE);
 800         }
 801 }
 802
 803 /**
 804  * FIXME: bad name.
 805  */
 806 class MappedDiff
 807 extends Diff
 808 {
 809         /**
 810          * Constructor.
 811          *
 812          * Computes diff between sequences of strings.
 813          *
 814          * This can be used to compute things like
 815          * case-insensitve diffs, or diffs which ignore
 816          * changes in white-space.
 817          *
 818          * @param $from_lines array An array of strings.
 819          *      (Typically these are lines from a file.)
 820          *
 821          * @param $to_lines array An array of strings.
 822          *
 823          * @param $mapped_from_lines array This array should
 824          *      have the same size number of elements as $from_lines.
 825          *      The elements in $mapped_from_lines and
 826          *      $mapped_to_lines are what is actually compared
 827          *      when computing the diff.
 828          *
 829          * @param $mapped_to_lines array This array should
 830          *      have the same number of elements as $to_lines.
 831          */
 832         function MappedDiff($from_lines, $to_lines,
 833                                                 $mapped_from_lines, $mapped_to_lines) {
 834
 835                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 836                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 837
 838                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 839
 840                 $xi = $yi = 0;
 841                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 842                         $orig = &$this->edits[$i]->orig;
 843                         if (is_array($orig)) {
 844                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 845                                 $xi += sizeof($orig);
 846                         }
 847
 848                         $closing = &$this->edits[$i]->closing;
 849                         if (is_array($closing)) {
 850                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 851                                 $yi += sizeof($closing);
 852                         }
 853                 }
 854         }
 855 }
 856
 857 /**
 858  * A class to format Diffs
 859  *
 860  * This class formats the diff in classic diff format.
 861  * It is intended that this class be customized via inheritance,
 862  * to obtain fancier outputs.
 863  */
 864 class DiffFormatter
 865 {
 866         /**
 867          * Number of leading context "lines" to preserve.
 868          *
 869          * This should be left at zero for this class, but subclasses
 870          * may want to set this to other values.
 871          */
 872         var $leading_context_lines = 0;
 873
 874         /**
 875          * Number of trailing context "lines" to preserve.
 876          *
 877          * This should be left at zero for this class, but subclasses
 878          * may want to set this to other values.
 879          */
 880         var $trailing_context_lines = 0;
 881
 882         /**
 883          * Format a diff.
 884          *
 885          * @param $diff object A Diff object.
 886          * @return string The formatted output.
 887          */
 888         function format($diff) {
 889
 890                 $xi = $yi = 1;
 891                 $block = false;
 892                 $context = array();
 893
 894                 $nlead = $this->leading_context_lines;
 895                 $ntrail = $this->trailing_context_lines;
 896
 897                 $this->_start_diff();
 898
 899                 foreach ($diff->edits as $edit) {
 900                         if ($edit->type == 'copy') {
 901                                 if (is_array($block)) {
 902                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 903                                                 $block[] = $edit;
 904                                         }
 905                                         else{
 906                                                 if ($ntrail) {
 907                                                         $context = array_slice($edit->orig, 0, $ntrail);
 908                                                         $block[] = new _DiffOp_Copy($context);
 909                                                 }
 910                                                 $this->_block($x0, $ntrail + $xi - $x0,
 911                                                                           $y0, $ntrail + $yi - $y0,
 912                                                                           $block);
 913                                                 $block = false;
 914                                         }
 915                                 }
 916                                 $context = $edit->orig;
 917                         }
 918                         else {
 919                                 if (! is_array($block)) {
 920                                         $context = array_slice($context, sizeof($context) - $nlead);
 921                                         $x0 = $xi - sizeof($context);
 922                                         $y0 = $yi - sizeof($context);
 923                                         $block = array();
 924                                         if ($context)
 925                                                 $block[] = new _DiffOp_Copy($context);
 926                                 }
 927                                 $block[] = $edit;
 928                         }
 929
 930                         if ($edit->orig)
 931                                 $xi += sizeof($edit->orig);
 932                         if ($edit->closing)
 933                                 $yi += sizeof($edit->closing);
 934                 }
 935
 936                 if (is_array($block))
 937                         $this->_block($x0, $xi - $x0,
 938                                                   $y0, $yi - $y0,
 939                                                   $block);
 940
 941                 return $this->_end_diff();
 942         }
 943
 944         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 945                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 946                 foreach ($edits as $edit) {
 947                         if ($edit->type == 'copy')
 948                                 $this->_context($edit->orig);
 949                         elseif ($edit->type == 'add')
 950                                 $this->_added($edit->closing);
 951                         elseif ($edit->type == 'delete')
 952                                 $this->_deleted($edit->orig);
 953                         elseif ($edit->type == 'change')
 954                                 $this->_changed($edit->orig, $edit->closing);
 955                         else
 956                                 trigger_error("Unknown edit type", E_USER_ERROR);
 957                 }
 958                 $this->_end_block();
 959         }
 960
 961         function _start_diff() {
 962                 ob_start();
 963         }
 964
 965         function _end_diff() {
 966                 $val = ob_get_contents();
 967                 ob_end_clean();
 968                 return $val;
 969         }
 970
 971         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 972                 if ($xlen > 1)
 973                         $xbeg .= "," . ($xbeg + $xlen - 1);
 974                 if ($ylen > 1)
 975                         $ybeg .= "," . ($ybeg + $ylen - 1);
 976
 977                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 978         }
 979
 980         function _start_block($header) {
 981                 echo $header;
 982         }
 983
 984         function _end_block() {
 985         }
 986
 987         function _lines($lines, $prefix = ' ') {
 988                 foreach ($lines as $line)
 989                         echo "$prefix $line\n";
 990         }
 991
 992         function _context($lines) {
 993                 $this->_lines($lines);
 994         }
 995
 996         function _added($lines) {
 997                 $this->_lines($lines, ">");
 998         }
 999         function _deleted($lines) {
1000                 $this->_lines($lines, "<");
1001         }
1002
1003         function _changed($orig, $closing) {
1004                 $this->_deleted($orig);
1005                 echo "---\n";
1006                 $this->_added($closing);
1007         }
1008 }
1009
1010
1011 /**
1012  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1013  *
1014  */
1015
1016 define('NBSP', "\xA0");                 // iso-8859-x non-breaking space.
1017
1018 class _HWLDF_WordAccumulator {
1019         function _HWLDF_WordAccumulator () {
1020                 $this->_lines = array();
1021                 $this->_line = '';
1022                 $this->_group = '';
1023                 $this->_tag = '';
1024         }
1025
1026         function _flushGroup ($new_tag) {
1027                 if ($this->_group !== '') {
1028           if ($this->_tag == 'mark')
1029                         $this->_line .= '<span class="diffchange">'.$this->_group.'</span>';
1030           else
1031                 $this->_line .= $this->_group;
1032         }
1033                 $this->_group = '';
1034                 $this->_tag = $new_tag;
1035         }
1036
1037         function _flushLine ($new_tag) {
1038                 $this->_flushGroup($new_tag);
1039                 if ($this->_line != '')
1040                         $this->_lines[] = $this->_line;
1041                 $this->_line = '';
1042         }
1043
1044         function addWords ($words, $tag = '') {
1045                 if ($tag != $this->_tag)
1046                         $this->_flushGroup($tag);
1047
1048                 foreach ($words as $word) {
1049                         // new-line should only come as first char of word.
1050                         if ($word == '')
1051                                 continue;
1052                         if ($word[0] == "\n") {
1053                                 $this->_group .= NBSP;
1054                                 $this->_flushLine($tag);
1055                                 $word = substr($word, 1);
1056                         }
1057                         assert(!strstr($word, "\n"));
1058                         $this->_group .= $word;
1059                 }
1060         }
1061
1062         function getLines() {
1063                 $this->_flushLine('~done');
1064                 return $this->_lines;
1065         }
1066 }
1067
1068 class WordLevelDiff extends MappedDiff
1069 {
1070         function WordLevelDiff ($orig_lines, $closing_lines) {
1071                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1072                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1073
1074
1075                 $this->MappedDiff($orig_words, $closing_words,
1076                                                   $orig_stripped, $closing_stripped);
1077         }
1078
1079         function _split($lines) {
1080                 // FIXME: fix POSIX char class.
1081 #                if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1082                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1083                                                         implode("\n", $lines),
1084                                                         $m)) {
1085                         return array(array(''), array(''));
1086                 }
1087                 return array($m[0], $m[1]);
1088         }
1089
1090         function orig () {
1091                 $orig = new _HWLDF_WordAccumulator;
1092
1093                 foreach ($this->edits as $edit) {
1094                         if ($edit->type == 'copy')
1095                                 $orig->addWords($edit->orig);
1096                         elseif ($edit->orig)
1097                                 $orig->addWords($edit->orig, 'mark');
1098                 }
1099                 return $orig->getLines();
1100         }
1101
1102         function closing () {
1103                 $closing = new _HWLDF_WordAccumulator;
1104
1105                 foreach ($this->edits as $edit) {
1106                         if ($edit->type == 'copy')
1107                                 $closing->addWords($edit->closing);
1108                         elseif ($edit->closing)
1109                                 $closing->addWords($edit->closing, 'mark');
1110                 }
1111                 return $closing->getLines();
1112         }
1113 }
1114
1115 /**
1116  *      Wikipedia Table style diff formatter.
1117  *
1118  */
1119 class TableDiffFormatter extends DiffFormatter
1120 {
1121         function TableDiffFormatter() {
1122                 $this->leading_context_lines = 2;
1123                 $this->trailing_context_lines = 2;
1124         }
1125
1126         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1127                 $l1 = wfMsg( "lineno", $xbeg );
1128                 $l2 = wfMsg( "lineno", $ybeg );
1129
1130                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1131                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1132                 return $r;
1133         }
1134
1135         function _start_block( $header ) {
1136                 global $wgOut;
1137                 $wgOut->addHTML( $header );
1138         }
1139
1140         function _end_block() {
1141         }
1142
1143         function _lines( $lines, $prefix=' ', $color="white" ) {
1144         }
1145
1146         function addedLine( $line ) {
1147                 return '<td>+</td><td class="diff-addedline">' .
1148                   $line.'</td>';
1149         }
1150
1151         function deletedLine( $line ) {
1152                 return '<td>-</td><td class="diff-deletedline">' .
1153                   $line.'</td>';
1154         }
1155
1156         function emptyLine() {
1157                 return '<td colspan="2">&nbsp;</td>';
1158         }
1159
1160         function contextLine( $line ) {
1161                 return '<td> </td><td class="diff-context">'.$line.'</td>';
1162         }
1163
1164         function _added($lines) {
1165                 global $wgOut;
1166                 foreach ($lines as $line) {
1167                         $wgOut->addHTML( '<tr>' . $this->emptyLine() .
1168                           $this->addedLine( $line ) . "</tr>\n" );
1169                 }
1170         }
1171
1172         function _deleted($lines) {
1173                 global $wgOut;
1174                 foreach ($lines as $line) {
1175                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1176                           $this->emptyLine() . "</tr>\n" );
1177                 }
1178         }
1179
1180         function _context( $lines ) {
1181                 global $wgOut;
1182                 foreach ($lines as $line) {
1183                         $wgOut->addHTML( '<tr>' . $this->contextLine( $line ) .
1184                           $this->contextLine( $line ) . "</tr>\n" );
1185                 }
1186         }
1187
1188         function _changed( $orig, $closing ) {
1189                 global $wgOut;
1190                 $diff = new WordLevelDiff( $orig, $closing );
1191                 $del = $diff->orig();
1192                 $add = $diff->closing();
1193
1194                 while ( $line = array_shift( $del ) ) {
1195                         $aline = array_shift( $add );
1196                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1197                           $this->addedLine( $aline ) . "</tr>\n" );
1198                 }
1199                 $this->_added( $add ); # If any leftovers
1200         }
1201 }
1202
1203 ?>