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