includes/DifferenceEngine.php

   1 <?php
   2 # See diff.doc
   3
   4 class DifferenceEngine {
   5         /* private */ var $mOldid, $mNewid;
   6         /* private */ var $mOldtitle, $mNewtitle, $mPagetitle;
   7         /* private */ var $mOldtext, $mNewtext;
   8         /* private */ var $mOldUser, $mNewUser;
   9         /* private */ var $mOldComment, $mNewComment;
  10         /* private */ var $mOldPage, $mNewPage;
  11         /* private */ var $mRcidMarkPatrolled;
  12
  13         function DifferenceEngine( $old, $new, $rcid = 0 )
  14         {
  15                 $this->mOldid = $old;
  16                 $this->mNewid = $new;
  17                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  18         }
  19
  20         function showDiffPage()
  21         {
  22                 global $wgUser, $wgTitle, $wgOut, $wgLang, $wgOnlySysopsCanPatrol, $wgUseRCPatrol;
  23                 $fname = 'DifferenceEngine::showDiffPage';
  24                 wfProfileIn( $fname );
  25
  26                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  27                   "{$this->mNewid})";
  28                 $mtext = wfMsg( 'missingarticle', $t );
  29
  30                 $wgOut->setArticleFlag( false );
  31                 if ( ! $this->loadText() ) {
  32                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
  33                         $wgOut->addHTML( $mtext );
  34                         wfProfileOut( $fname );
  35                         return;
  36                 }
  37                 $wgOut->suppressQuickbar();
  38
  39                 $oldTitle = $this->mOldPage->getPrefixedText();
  40                 $newTitle = $this->mNewPage->getPrefixedText();
  41                 if( $oldTitle == $newTitle ) {
  42                         $wgOut->setPageTitle( $newTitle );
  43                 } else {
  44                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
  45                 }
  46                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
  47                 $wgOut->setRobotpolicy( 'noindex,follow' );
  48
  49                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  50                         $wgOut->loginToUse();
  51                         $wgOut->output();
  52                         wfProfileOut( $fname );
  53                         exit;
  54                 }
  55
  56                 $sk = $wgUser->getSkin();
  57                 $talk = $wgLang->getNsText( NS_TALK );
  58                 $contribs = wfMsg( 'contribslink' );
  59
  60                 $this->mOldComment = $sk->formatComment($this->mOldComment);
  61                 $this->mNewComment = $sk->formatComment($this->mNewComment);
  62
  63                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
  64                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
  65                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
  66                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( 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                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->getID() != 0 &&
  78                      ( $wgUser->isSysop() || !$wgOnlySysopsCanPatrol ) )
  79                 {
  80                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
  81                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
  82                 } else {
  83                         $patrol = '';
  84                 }
  85
  86                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
  87                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment;
  88                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
  89                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment . $patrol;
  90
  91                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
  92                   $oldHeader, $newHeader );
  93                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
  94                 $wgOut->addWikiText( $this->mNewtext );
  95
  96                 wfProfileOut( $fname );
  97         }
  98
  99         function showDiff( $otext, $ntext, $otitle, $ntitle )
 100         {
 101                 global $wgOut, $wgUseExternalDiffEngine;
 102
 103                 $otext = str_replace( "\r\n", "\n", htmlspecialchars( $otext ) );
 104                 $ntext = str_replace( "\r\n", "\n", htmlspecialchars( $ntext ) );
 105
 106
 107                         $wgOut->addHTML( "<table border='0' width='98%'
 108 cellpadding='0' cellspacing='4px' class='diff'><tr>
 109 <td colspan='2' width='50%' align='center' class='diff-otitle'>
 110 {$otitle}</td>
 111 <td colspan='2' width='50%' align='center' class='diff-ntitle'>
 112 {$ntitle}</td>
 113 </tr>\n" );
 114                 if ( $wgUseExternalDiffEngine ) {
 115                         dl('php_wikidiff.so');
 116                         $wgOut->addHTML( wikidiff_do_diff( $otext, $ntext, 2) );
 117                 } else {
 118                         $ota = explode( "\n", $otext);
 119                         $nta = explode( "\n", $ntext);
 120                         $diffs = new Diff( $ota, $nta );
 121                         $formatter = new TableDiffFormatter();
 122                         $formatter->format( $diffs );
 123                 }
 124                 $wgOut->addHTML( "</table>\n" );
 125         }
 126
 127         # Load the text of the articles to compare.  If newid is 0, then compare
 128         # the old article in oldid to the current article; if oldid is 0, then
 129         # compare the current article to the immediately previous one (ignoring
 130         # the value of newid).
 131         #
 132         function loadText()
 133         {
 134                 global $wgTitle, $wgOut, $wgLang;
 135                 $fname = 'DifferenceEngine::loadText';
 136
 137                 $dbr =& wfGetDB( DB_SLAVE );
 138                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 139                         $wgOut->setArticleFlag( true );
 140                         $newLink = $wgTitle->getLocalUrl();
 141                         $this->mPagetitle = wfMsg( 'currentrev' );
 142                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 143                         $id = $wgTitle->getArticleID();
 144
 145                         $s = $dbr->getArray( 'cur', array( 'cur_text', 'cur_user_text', 'cur_comment' ),
 146                                 array( 'cur_id' => $id ), $fname );
 147                         if ( $s === false ) {
 148                                 return false;
 149                         }
 150
 151                         $this->mNewPage = &$wgTitle;
 152                         $this->mNewtext = $s->cur_text;
 153                         $this->mNewUser = $s->cur_user_text;
 154                         $this->mNewComment = $s->cur_comment;
 155                 } else {
 156                         $s = $dbr->getArray( 'old', array( 'old_namespace','old_title','old_timestamp', 'old_text',
 157                                 'old_flags','old_user_text','old_comment' ), array( 'old_id' => $this->mNewid ), $fname );
 158
 159                         if ( $s === false ) {
 160                                 return false;
 161                         }
 162
 163                         $this->mNewtext = Article::getRevisionText( $s );
 164
 165                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 166                         $this->mNewPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 167                         $newLink = $wgTitle->getLocalUrl ('oldid=' . $this->mNewid);
 168                         $this->mPagetitle = wfMsg( 'revisionasof', $t );
 169                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 170                         $this->mNewUser = $s->old_user_text;
 171                         $this->mNewComment = $s->old_comment;
 172                 }
 173                 if ( 0 == $this->mOldid ) {
 174                         $s = $dbr->getArray( 'old',
 175                                 array( 'old_namespace','old_title','old_timestamp','old_text', 'old_flags','old_user_text','old_comment' ),
 176                                 array( /* WHERE */
 177                                         'old_namespace' => $this->mNewPage->getNamespace(),
 178                                         'old_title' => $this->mNewPage->getDBkey()
 179                                 ), $fname, array( 'ORDER BY' => 'inverse_timestamp', 'USE INDEX' => 'name_title_timestamp' )
 180                         );
 181                 } else {
 182                         $s = $dbr->getArray( 'old',
 183                                 array( 'old_namespace','old_title','old_timestamp','old_text','old_flags','old_user_text','old_comment'),
 184                                 array( 'old_id' => $this->mOldid ),
 185                                 $fname
 186                         );
 187                 }
 188                 if ( $s === false ) {
 189                         return false;
 190                 }
 191                 $this->mOldPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 192                 $this->mOldtext = Article::getRevisionText( $s );
 193
 194                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 195                 $oldLink = $this->mOldPage->getLocalUrl ('oldid=' . $this->mOldid);
 196                 $this->mOldtitle = "<a href='$oldLink'>" . wfMsg( 'revisionasof', $t ) . '</a>';
 197                 $this->mOldUser = $s->old_user_text;
 198                 $this->mOldComment = $s->old_comment;
 199
 200                 return true;
 201         }
 202 }
 203
 204 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 205 //
 206 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 207 // You may copy this code freely under the conditions of the GPL.
 208 //
 209
 210 define('USE_ASSERTS', function_exists('assert'));
 211
 212 class _DiffOp {
 213         var $type;
 214         var $orig;
 215         var $closing;
 216
 217         function reverse() {
 218                 trigger_error('pure virtual', E_USER_ERROR);
 219         }
 220
 221         function norig() {
 222                 return $this->orig ? sizeof($this->orig) : 0;
 223         }
 224
 225         function nclosing() {
 226                 return $this->closing ? sizeof($this->closing) : 0;
 227         }
 228 }
 229
 230 class _DiffOp_Copy extends _DiffOp {
 231         var $type = 'copy';
 232
 233         function _DiffOp_Copy ($orig, $closing = false) {
 234                 if (!is_array($closing))
 235                         $closing = $orig;
 236                 $this->orig = $orig;
 237                 $this->closing = $closing;
 238         }
 239
 240         function reverse() {
 241                 return new _DiffOp_Copy($this->closing, $this->orig);
 242         }
 243 }
 244
 245 class _DiffOp_Delete extends _DiffOp {
 246         var $type = 'delete';
 247
 248         function _DiffOp_Delete ($lines) {
 249                 $this->orig = $lines;
 250                 $this->closing = false;
 251         }
 252
 253         function reverse() {
 254                 return new _DiffOp_Add($this->orig);
 255         }
 256 }
 257
 258 class _DiffOp_Add extends _DiffOp {
 259         var $type = 'add';
 260
 261         function _DiffOp_Add ($lines) {
 262                 $this->closing = $lines;
 263                 $this->orig = false;
 264         }
 265
 266         function reverse() {
 267                 return new _DiffOp_Delete($this->closing);
 268         }
 269 }
 270
 271 class _DiffOp_Change extends _DiffOp {
 272         var $type = 'change';
 273
 274         function _DiffOp_Change ($orig, $closing) {
 275                 $this->orig = $orig;
 276                 $this->closing = $closing;
 277         }
 278
 279         function reverse() {
 280                 return new _DiffOp_Change($this->closing, $this->orig);
 281         }
 282 }
 283
 284
 285 /**
 286  * Class used internally by Diff to actually compute the diffs.
 287  *
 288  * The algorithm used here is mostly lifted from the perl module
 289  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 290  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 291  *
 292  * More ideas are taken from:
 293  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 294  *
 295  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 296  * diffutils-2.7, which can be found at:
 297  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 298  *
 299  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 300  * are my own.
 301  *
 302  * @author Geoffrey T. Dairiki
 303  * @access private
 304  */
 305 class _DiffEngine
 306 {
 307         function diff ($from_lines, $to_lines) {
 308                 $n_from = sizeof($from_lines);
 309                 $n_to = sizeof($to_lines);
 310
 311                 $this->xchanged = $this->ychanged = array();
 312                 $this->xv = $this->yv = array();
 313                 $this->xind = $this->yind = array();
 314                 unset($this->seq);
 315                 unset($this->in_seq);
 316                 unset($this->lcs);
 317
 318                 // Skip leading common lines.
 319                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 320                         if ($from_lines[$skip] != $to_lines[$skip])
 321                                 break;
 322                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 323                 }
 324                 // Skip trailing common lines.
 325                 $xi = $n_from; $yi = $n_to;
 326                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 327                         if ($from_lines[$xi] != $to_lines[$yi])
 328                                 break;
 329                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 330                 }
 331
 332                 // Ignore lines which do not exist in both files.
 333                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 334                         $xhash[$from_lines[$xi]] = 1;
 335                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 336                         $line = $to_lines[$yi];
 337                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 338                                 continue;
 339                         $yhash[$line] = 1;
 340                         $this->yv[] = $line;
 341                         $this->yind[] = $yi;
 342                 }
 343                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 344                         $line = $from_lines[$xi];
 345                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 346                                 continue;
 347                         $this->xv[] = $line;
 348                         $this->xind[] = $xi;
 349                 }
 350
 351                 // Find the LCS.
 352                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 353
 354                 // Merge edits when possible
 355                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 356                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 357
 358                 // Compute the edit operations.
 359                 $edits = array();
 360                 $xi = $yi = 0;
 361                 while ($xi < $n_from || $yi < $n_to) {
 362                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 363                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 364
 365                         // Skip matching "snake".
 366                         $copy = array();
 367                         while ( $xi < $n_from && $yi < $n_to
 368                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 369                                 $copy[] = $from_lines[$xi++];
 370                                 ++$yi;
 371                         }
 372                         if ($copy)
 373                                 $edits[] = new _DiffOp_Copy($copy);
 374
 375                         // Find deletes & adds.
 376                         $delete = array();
 377                         while ($xi < $n_from && $this->xchanged[$xi])
 378                                 $delete[] = $from_lines[$xi++];
 379
 380                         $add = array();
 381                         while ($yi < $n_to && $this->ychanged[$yi])
 382                                 $add[] = $to_lines[$yi++];
 383
 384                         if ($delete && $add)
 385                                 $edits[] = new _DiffOp_Change($delete, $add);
 386                         elseif ($delete)
 387                                 $edits[] = new _DiffOp_Delete($delete);
 388                         elseif ($add)
 389                                 $edits[] = new _DiffOp_Add($add);
 390                 }
 391                 return $edits;
 392         }
 393
 394
 395         /* Divide the Largest Common Subsequence (LCS) of the sequences
 396          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 397          * sized segments.
 398          *
 399          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 400          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 401          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 402          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 403          * that (X0, Y0) == (XOFF, YOFF) and
 404          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 405          *
 406          * This function assumes that the first lines of the specified portions
 407          * of the two files do not match, and likewise that the last lines do not
 408          * match.  The caller must trim matching lines from the beginning and end
 409          * of the portions it is going to specify.
 410          */
 411         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 412         $flip = false;
 413
 414         if ($xlim - $xoff > $ylim - $yoff) {
 415                 // Things seems faster (I'm not sure I understand why)
 416                         // when the shortest sequence in X.
 417                         $flip = true;
 418                 list ($xoff, $xlim, $yoff, $ylim)
 419                 = array( $yoff, $ylim, $xoff, $xlim);
 420                 }
 421
 422         if ($flip)
 423                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 424                 $ymatches[$this->xv[$i]][] = $i;
 425         else
 426                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 427                 $ymatches[$this->yv[$i]][] = $i;
 428
 429         $this->lcs = 0;
 430         $this->seq[0]= $yoff - 1;
 431         $this->in_seq = array();
 432         $ymids[0] = array();
 433
 434         $numer = $xlim - $xoff + $nchunks - 1;
 435         $x = $xoff;
 436         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 437                 if ($chunk > 0)
 438                 for ($i = 0; $i <= $this->lcs; $i++)
 439                         $ymids[$i][$chunk-1] = $this->seq[$i];
 440
 441                 $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 442                 for ( ; $x < $x1; $x++) {
 443                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 444                                 if (empty($ymatches[$line]))
 445                         continue;
 446                 $matches = $ymatches[$line];
 447                                 reset($matches);
 448                 while (list ($junk, $y) = each($matches))
 449                         if (empty($this->in_seq[$y])) {
 450                         $k = $this->_lcs_pos($y);
 451                         USE_ASSERTS && assert($k > 0);
 452                         $ymids[$k] = $ymids[$k-1];
 453                         break;
 454                                         }
 455                 while (list ($junk, $y) = each($matches)) {
 456                         if ($y > $this->seq[$k-1]) {
 457                         USE_ASSERTS && assert($y < $this->seq[$k]);
 458                         // Optimization: this is a common case:
 459                         //      next match is just replacing previous match.
 460                         $this->in_seq[$this->seq[$k]] = false;
 461                         $this->seq[$k] = $y;
 462                         $this->in_seq[$y] = 1;
 463                                         }
 464                         else if (empty($this->in_seq[$y])) {
 465                         $k = $this->_lcs_pos($y);
 466                         USE_ASSERTS && assert($k > 0);
 467                         $ymids[$k] = $ymids[$k-1];
 468                                         }
 469                                 }
 470                         }
 471                 }
 472
 473         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 474         $ymid = $ymids[$this->lcs];
 475         for ($n = 0; $n < $nchunks - 1; $n++) {
 476                 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 477                 $y1 = $ymid[$n] + 1;
 478                 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 479                 }
 480         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 481
 482         return array($this->lcs, $seps);
 483         }
 484
 485         function _lcs_pos ($ypos) {
 486         $end = $this->lcs;
 487         if ($end == 0 || $ypos > $this->seq[$end]) {
 488                 $this->seq[++$this->lcs] = $ypos;
 489                 $this->in_seq[$ypos] = 1;
 490                 return $this->lcs;
 491                 }
 492
 493         $beg = 1;
 494         while ($beg < $end) {
 495                 $mid = (int)(($beg + $end) / 2);
 496                 if ( $ypos > $this->seq[$mid] )
 497                 $beg = $mid + 1;
 498                 else
 499                 $end = $mid;
 500                 }
 501
 502         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 503
 504         $this->in_seq[$this->seq[$end]] = false;
 505         $this->seq[$end] = $ypos;
 506         $this->in_seq[$ypos] = 1;
 507         return $end;
 508         }
 509
 510         /* Find LCS of two sequences.
 511          *
 512          * The results are recorded in the vectors $this->{x,y}changed[], by
 513          * storing a 1 in the element for each line that is an insertion
 514          * or deletion (ie. is not in the LCS).
 515          *
 516          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 517          *
 518          * Note that XLIM, YLIM are exclusive bounds.
 519          * All line numbers are origin-0 and discarded lines are not counted.
 520          */
 521         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 522         // Slide down the bottom initial diagonal.
 523         while ($xoff < $xlim && $yoff < $ylim
 524                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 525                 ++$xoff;
 526                 ++$yoff;
 527                 }
 528
 529         // Slide up the top initial diagonal.
 530         while ($xlim > $xoff && $ylim > $yoff
 531                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 532                 --$xlim;
 533                 --$ylim;
 534                 }
 535
 536         if ($xoff == $xlim || $yoff == $ylim)
 537                 $lcs = 0;
 538         else {
 539                 // This is ad hoc but seems to work well.
 540                 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 541                 //$nchunks = max(2,min(8,(int)$nchunks));
 542                 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 543                 list ($lcs, $seps)
 544                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 545                 }
 546
 547         if ($lcs == 0) {
 548                 // X and Y sequences have no common subsequence:
 549                 // mark all changed.
 550                 while ($yoff < $ylim)
 551                 $this->ychanged[$this->yind[$yoff++]] = 1;
 552                 while ($xoff < $xlim)
 553                 $this->xchanged[$this->xind[$xoff++]] = 1;
 554                 }
 555         else {
 556                 // Use the partitions to split this problem into subproblems.
 557                 reset($seps);
 558                 $pt1 = $seps[0];
 559                 while ($pt2 = next($seps)) {
 560                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 561                 $pt1 = $pt2;
 562                         }
 563                 }
 564         }
 565
 566         /* Adjust inserts/deletes of identical lines to join changes
 567          * as much as possible.
 568          *
 569          * We do something when a run of changed lines include a
 570          * line at one end and has an excluded, identical line at the other.
 571          * We are free to choose which identical line is included.
 572          * `compareseq' usually chooses the one at the beginning,
 573          * but usually it is cleaner to consider the following identical line
 574          * to be the "change".
 575          *
 576          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 577          */
 578         function _shift_boundaries ($lines, &$changed, $other_changed) {
 579         $i = 0;
 580         $j = 0;
 581
 582         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 583         $len = sizeof($lines);
 584         $other_len = sizeof($other_changed);
 585
 586         while (1) {
 587                 /*
 588                  * Scan forwards to find beginning of another run of changes.
 589                  * Also keep track of the corresponding point in the other file.
 590                  *
 591                  * Throughout this code, $i and $j are adjusted together so that
 592                  * the first $i elements of $changed and the first $j elements
 593                  * of $other_changed both contain the same number of zeros
 594                  * (unchanged lines).
 595                  * Furthermore, $j is always kept so that $j == $other_len or
 596                  * $other_changed[$j] == false.
 597                  */
 598                 while ($j < $other_len && $other_changed[$j])
 599                 $j++;
 600
 601                 while ($i < $len && ! $changed[$i]) {
 602                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 603                 $i++; $j++;
 604                 while ($j < $other_len && $other_changed[$j])
 605                         $j++;
 606                         }
 607
 608                 if ($i == $len)
 609                 break;
 610
 611                 $start = $i;
 612
 613                 // Find the end of this run of changes.
 614                 while (++$i < $len && $changed[$i])
 615                 continue;
 616
 617                 do {
 618                 /*
 619                  * Record the length of this run of changes, so that
 620                  * we can later determine whether the run has grown.
 621                  */
 622                 $runlength = $i - $start;
 623
 624                 /*
 625                  * Move the changed region back, so long as the
 626                  * previous unchanged line matches the last changed one.
 627                  * This merges with previous changed regions.
 628                  */
 629                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 630                         $changed[--$start] = 1;
 631                         $changed[--$i] = false;
 632                         while ($start > 0 && $changed[$start - 1])
 633                         $start--;
 634                         USE_ASSERTS && assert('$j > 0');
 635                         while ($other_changed[--$j])
 636                         continue;
 637                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 638                                 }
 639
 640                 /*
 641                  * Set CORRESPONDING to the end of the changed run, at the last
 642                  * point where it corresponds to a changed run in the other file.
 643                  * CORRESPONDING == LEN means no such point has been found.
 644                  */
 645                 $corresponding = $j < $other_len ? $i : $len;
 646
 647                 /*
 648                  * Move the changed region forward, so long as the
 649                  * first changed line matches the following unchanged one.
 650                  * This merges with following changed regions.
 651                  * Do this second, so that if there are no merges,
 652                  * the changed region is moved forward as far as possible.
 653                  */
 654                 while ($i < $len && $lines[$start] == $lines[$i]) {
 655                         $changed[$start++] = false;
 656                         $changed[$i++] = 1;
 657                         while ($i < $len && $changed[$i])
 658                         $i++;
 659
 660                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 661                         $j++;
 662                         if ($j < $other_len && $other_changed[$j]) {
 663                         $corresponding = $i;
 664                         while ($j < $other_len && $other_changed[$j])
 665                                 $j++;
 666                                         }
 667                                 }
 668                         } while ($runlength != $i - $start);
 669
 670                 /*
 671                  * If possible, move the fully-merged run of changes
 672                  * back to a corresponding run in the other file.
 673                  */
 674                 while ($corresponding < $i) {
 675                 $changed[--$start] = 1;
 676                 $changed[--$i] = 0;
 677                 USE_ASSERTS && assert('$j > 0');
 678                 while ($other_changed[--$j])
 679                         continue;
 680                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 681                         }
 682                 }
 683         }
 684 }
 685
 686 /**
 687  * Class representing a 'diff' between two sequences of strings.
 688  */
 689 class Diff
 690 {
 691         var $edits;
 692
 693         /**
 694          * Constructor.
 695          * Computes diff between sequences of strings.
 696          *
 697          * @param $from_lines array An array of strings.
 698          *                (Typically these are lines from a file.)
 699          * @param $to_lines array An array of strings.
 700          */
 701         function Diff($from_lines, $to_lines) {
 702                 $eng = new _DiffEngine;
 703                 $this->edits = $eng->diff($from_lines, $to_lines);
 704                 //$this->_check($from_lines, $to_lines);
 705         }
 706
 707         /**
 708          * Compute reversed Diff.
 709          *
 710          * SYNOPSIS:
 711          *
 712          *      $diff = new Diff($lines1, $lines2);
 713          *      $rev = $diff->reverse();
 714          * @return object A Diff object representing the inverse of the
 715          *                                original diff.
 716          */
 717         function reverse () {
 718         $rev = $this;
 719                 $rev->edits = array();
 720                 foreach ($this->edits as $edit) {
 721                         $rev->edits[] = $edit->reverse();
 722                 }
 723         return $rev;
 724         }
 725
 726         /**
 727          * Check for empty diff.
 728          *
 729          * @return bool True iff two sequences were identical.
 730          */
 731         function isEmpty () {
 732                 foreach ($this->edits as $edit) {
 733                         if ($edit->type != 'copy')
 734                                 return false;
 735                 }
 736                 return true;
 737         }
 738
 739         /**
 740          * Compute the length of the Longest Common Subsequence (LCS).
 741          *
 742          * This is mostly for diagnostic purposed.
 743          *
 744          * @return int The length of the LCS.
 745          */
 746         function lcs () {
 747         $lcs = 0;
 748                 foreach ($this->edits as $edit) {
 749                         if ($edit->type == 'copy')
 750                                 $lcs += sizeof($edit->orig);
 751                 }
 752         return $lcs;
 753         }
 754
 755         /**
 756          * Get the original set of lines.
 757          *
 758          * This reconstructs the $from_lines parameter passed to the
 759          * constructor.
 760          *
 761          * @return array The original sequence of strings.
 762          */
 763         function orig() {
 764                 $lines = array();
 765
 766                 foreach ($this->edits as $edit) {
 767                         if ($edit->orig)
 768                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 769                 }
 770                 return $lines;
 771         }
 772
 773         /**
 774          * Get the closing set of lines.
 775          *
 776          * This reconstructs the $to_lines parameter passed to the
 777          * constructor.
 778          *
 779          * @return array The sequence of strings.
 780          */
 781         function closing() {
 782                 $lines = array();
 783
 784                 foreach ($this->edits as $edit) {
 785                         if ($edit->closing)
 786                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 787                 }
 788                 return $lines;
 789         }
 790
 791         /**
 792          * Check a Diff for validity.
 793          *
 794          * This is here only for debugging purposes.
 795          */
 796         function _check ($from_lines, $to_lines) {
 797                 if (serialize($from_lines) != serialize($this->orig()))
 798                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 799                 if (serialize($to_lines) != serialize($this->closing()))
 800                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 801
 802                 $rev = $this->reverse();
 803                 if (serialize($to_lines) != serialize($rev->orig()))
 804                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 805                 if (serialize($from_lines) != serialize($rev->closing()))
 806                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 807
 808
 809                 $prevtype = 'none';
 810                 foreach ($this->edits as $edit) {
 811                         if ( $prevtype == $edit->type )
 812                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 813                         $prevtype = $edit->type;
 814                 }
 815
 816                 $lcs = $this->lcs();
 817                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 818         }
 819 }
 820
 821 /**
 822  * FIXME: bad name.
 823  */
 824 class MappedDiff
 825 extends Diff
 826 {
 827         /**
 828          * Constructor.
 829          *
 830          * Computes diff between sequences of strings.
 831          *
 832          * This can be used to compute things like
 833          * case-insensitve diffs, or diffs which ignore
 834          * changes in white-space.
 835          *
 836          * @param $from_lines array An array of strings.
 837          *      (Typically these are lines from a file.)
 838          *
 839          * @param $to_lines array An array of strings.
 840          *
 841          * @param $mapped_from_lines array This array should
 842          *      have the same size number of elements as $from_lines.
 843          *      The elements in $mapped_from_lines and
 844          *      $mapped_to_lines are what is actually compared
 845          *      when computing the diff.
 846          *
 847          * @param $mapped_to_lines array This array should
 848          *      have the same number of elements as $to_lines.
 849          */
 850         function MappedDiff($from_lines, $to_lines,
 851                                                 $mapped_from_lines, $mapped_to_lines) {
 852
 853                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 854                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 855
 856                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 857
 858                 $xi = $yi = 0;
 859                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 860                         $orig = &$this->edits[$i]->orig;
 861                         if (is_array($orig)) {
 862                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 863                                 $xi += sizeof($orig);
 864                         }
 865
 866                         $closing = &$this->edits[$i]->closing;
 867                         if (is_array($closing)) {
 868                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 869                                 $yi += sizeof($closing);
 870                         }
 871                 }
 872         }
 873 }
 874
 875 /**
 876  * A class to format Diffs
 877  *
 878  * This class formats the diff in classic diff format.
 879  * It is intended that this class be customized via inheritance,
 880  * to obtain fancier outputs.
 881  */
 882 class DiffFormatter
 883 {
 884         /**
 885          * Number of leading context "lines" to preserve.
 886          *
 887          * This should be left at zero for this class, but subclasses
 888          * may want to set this to other values.
 889          */
 890         var $leading_context_lines = 0;
 891
 892         /**
 893          * Number of trailing context "lines" to preserve.
 894          *
 895          * This should be left at zero for this class, but subclasses
 896          * may want to set this to other values.
 897          */
 898         var $trailing_context_lines = 0;
 899
 900         /**
 901          * Format a diff.
 902          *
 903          * @param $diff object A Diff object.
 904          * @return string The formatted output.
 905          */
 906         function format($diff) {
 907
 908                 $xi = $yi = 1;
 909                 $block = false;
 910                 $context = array();
 911
 912                 $nlead = $this->leading_context_lines;
 913                 $ntrail = $this->trailing_context_lines;
 914
 915                 $this->_start_diff();
 916
 917                 foreach ($diff->edits as $edit) {
 918                         if ($edit->type == 'copy') {
 919                                 if (is_array($block)) {
 920                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 921                                                 $block[] = $edit;
 922                                         }
 923                                         else{
 924                                                 if ($ntrail) {
 925                                                         $context = array_slice($edit->orig, 0, $ntrail);
 926                                                         $block[] = new _DiffOp_Copy($context);
 927                                                 }
 928                                                 $this->_block($x0, $ntrail + $xi - $x0,
 929                                                                           $y0, $ntrail + $yi - $y0,
 930                                                                           $block);
 931                                                 $block = false;
 932                                         }
 933                                 }
 934                                 $context = $edit->orig;
 935                         }
 936                         else {
 937                                 if (! is_array($block)) {
 938                                         $context = array_slice($context, sizeof($context) - $nlead);
 939                                         $x0 = $xi - sizeof($context);
 940                                         $y0 = $yi - sizeof($context);
 941                                         $block = array();
 942                                         if ($context)
 943                                                 $block[] = new _DiffOp_Copy($context);
 944                                 }
 945                                 $block[] = $edit;
 946                         }
 947
 948                         if ($edit->orig)
 949                                 $xi += sizeof($edit->orig);
 950                         if ($edit->closing)
 951                                 $yi += sizeof($edit->closing);
 952                 }
 953
 954                 if (is_array($block))
 955                         $this->_block($x0, $xi - $x0,
 956                                                   $y0, $yi - $y0,
 957                                                   $block);
 958
 959                 return $this->_end_diff();
 960         }
 961
 962         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 963                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 964                 foreach ($edits as $edit) {
 965                         if ($edit->type == 'copy')
 966                                 $this->_context($edit->orig);
 967                         elseif ($edit->type == 'add')
 968                                 $this->_added($edit->closing);
 969                         elseif ($edit->type == 'delete')
 970                                 $this->_deleted($edit->orig);
 971                         elseif ($edit->type == 'change')
 972                                 $this->_changed($edit->orig, $edit->closing);
 973                         else
 974                                 trigger_error('Unknown edit type', E_USER_ERROR);
 975                 }
 976                 $this->_end_block();
 977         }
 978
 979         function _start_diff() {
 980                 ob_start();
 981         }
 982
 983         function _end_diff() {
 984                 $val = ob_get_contents();
 985                 ob_end_clean();
 986                 return $val;
 987         }
 988
 989         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 990                 if ($xlen > 1)
 991                         $xbeg .= "," . ($xbeg + $xlen - 1);
 992                 if ($ylen > 1)
 993                         $ybeg .= "," . ($ybeg + $ylen - 1);
 994
 995                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 996         }
 997
 998         function _start_block($header) {
 999                 echo $header;
1000         }
1001
1002         function _end_block() {
1003         }
1004
1005         function _lines($lines, $prefix = ' ') {
1006                 foreach ($lines as $line)
1007                         echo "$prefix $line\n";
1008         }
1009
1010         function _context($lines) {
1011                 $this->_lines($lines);
1012         }
1013
1014         function _added($lines) {
1015                 $this->_lines($lines, '>');
1016         }
1017         function _deleted($lines) {
1018                 $this->_lines($lines, '<');
1019         }
1020
1021         function _changed($orig, $closing) {
1022                 $this->_deleted($orig);
1023                 echo "---\n";
1024                 $this->_added($closing);
1025         }
1026 }
1027
1028
1029 /**
1030  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1031  *
1032  */
1033
1034 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1035
1036 class _HWLDF_WordAccumulator {
1037         function _HWLDF_WordAccumulator () {
1038                 $this->_lines = array();
1039                 $this->_line = '';
1040                 $this->_group = '';
1041                 $this->_tag = '';
1042         }
1043
1044         function _flushGroup ($new_tag) {
1045                 if ($this->_group !== '') {
1046           if ($this->_tag == 'mark')
1047                         $this->_line .= '<span class="diffchange">'.$this->_group.'</span>';
1048           else
1049                 $this->_line .= $this->_group;
1050         }
1051                 $this->_group = '';
1052                 $this->_tag = $new_tag;
1053         }
1054
1055         function _flushLine ($new_tag) {
1056                 $this->_flushGroup($new_tag);
1057                 if ($this->_line != '')
1058                         $this->_lines[] = $this->_line;
1059                 $this->_line = '';
1060         }
1061
1062         function addWords ($words, $tag = '') {
1063                 if ($tag != $this->_tag)
1064                         $this->_flushGroup($tag);
1065
1066                 foreach ($words as $word) {
1067                         // new-line should only come as first char of word.
1068                         if ($word == '')
1069                                 continue;
1070                         if ($word[0] == "\n") {
1071                                 $this->_group .= NBSP;
1072                                 $this->_flushLine($tag);
1073                                 $word = substr($word, 1);
1074                         }
1075                         assert(!strstr($word, "\n"));
1076                         $this->_group .= $word;
1077                 }
1078         }
1079
1080         function getLines() {
1081                 $this->_flushLine('~done');
1082                 return $this->_lines;
1083         }
1084 }
1085
1086 class WordLevelDiff extends MappedDiff
1087 {
1088         function WordLevelDiff ($orig_lines, $closing_lines) {
1089                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1090                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1091
1092
1093                 $this->MappedDiff($orig_words, $closing_words,
1094                                                   $orig_stripped, $closing_stripped);
1095         }
1096
1097         function _split($lines) {
1098                 // FIXME: fix POSIX char class.
1099 #                if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1100                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1101                                                         implode("\n", $lines),
1102                                                         $m)) {
1103                         return array(array(''), array(''));
1104                 }
1105                 return array($m[0], $m[1]);
1106         }
1107
1108         function orig () {
1109                 $orig = new _HWLDF_WordAccumulator;
1110
1111                 foreach ($this->edits as $edit) {
1112                         if ($edit->type == 'copy')
1113                                 $orig->addWords($edit->orig);
1114                         elseif ($edit->orig)
1115                                 $orig->addWords($edit->orig, 'mark');
1116                 }
1117                 return $orig->getLines();
1118         }
1119
1120         function closing () {
1121                 $closing = new _HWLDF_WordAccumulator;
1122
1123                 foreach ($this->edits as $edit) {
1124                         if ($edit->type == 'copy')
1125                                 $closing->addWords($edit->closing);
1126                         elseif ($edit->closing)
1127                                 $closing->addWords($edit->closing, 'mark');
1128                 }
1129                 return $closing->getLines();
1130         }
1131 }
1132
1133 /**
1134  *      Wikipedia Table style diff formatter.
1135  *
1136  */
1137 class TableDiffFormatter extends DiffFormatter
1138 {
1139         function TableDiffFormatter() {
1140                 $this->leading_context_lines = 2;
1141                 $this->trailing_context_lines = 2;
1142         }
1143
1144         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1145                 $l1 = wfMsg( 'lineno', $xbeg );
1146                 $l2 = wfMsg( 'lineno', $ybeg );
1147
1148                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1149                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1150                 return $r;
1151         }
1152
1153         function _start_block( $header ) {
1154                 global $wgOut;
1155                 $wgOut->addHTML( $header );
1156         }
1157
1158         function _end_block() {
1159         }
1160
1161         function _lines( $lines, $prefix=' ', $color='white' ) {
1162         }
1163
1164         function addedLine( $line ) {
1165                 return '<td>+</td><td class="diff-addedline">' .
1166                   $line.'</td>';
1167         }
1168
1169         function deletedLine( $line ) {
1170                 return '<td>-</td><td class="diff-deletedline">' .
1171                   $line.'</td>';
1172         }
1173
1174         function emptyLine() {
1175                 return '<td colspan="2">&nbsp;</td>';
1176         }
1177
1178         function contextLine( $line ) {
1179                 return '<td> </td><td class="diff-context">'.$line.'</td>';
1180         }
1181
1182         function _added($lines) {
1183                 global $wgOut;
1184                 foreach ($lines as $line) {
1185                         $wgOut->addHTML( '<tr>' . $this->emptyLine() .
1186                           $this->addedLine( $line ) . "</tr>\n" );
1187                 }
1188         }
1189
1190         function _deleted($lines) {
1191                 global $wgOut;
1192                 foreach ($lines as $line) {
1193                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1194                           $this->emptyLine() . "</tr>\n" );
1195                 }
1196         }
1197
1198         function _context( $lines ) {
1199                 global $wgOut;
1200                 foreach ($lines as $line) {
1201                         $wgOut->addHTML( '<tr>' . $this->contextLine( $line ) .
1202                           $this->contextLine( $line ) . "</tr>\n" );
1203                 }
1204         }
1205
1206         function _changed( $orig, $closing ) {
1207                 global $wgOut;
1208                 $diff = new WordLevelDiff( $orig, $closing );
1209                 $del = $diff->orig();
1210                 $add = $diff->closing();
1211
1212                 while ( $line = array_shift( $del ) ) {
1213                         $aline = array_shift( $add );
1214                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1215                           $this->addedLine( $aline ) . "</tr>\n" );
1216                 }
1217                 $this->_added( $add ); # If any leftovers
1218         }
1219 }
1220
1221 ?>