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