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                 global $wgTitle;
  16                 if ( 'prev' == $new ) {
  17                         $this->mNewid = intval($old);
  18                         $dbr =& wfGetDB( DB_SLAVE );
  19                         $this->mOldid = $dbr->selectField( 'old', 'old_id',
  20                                 "old_title='" . $wgTitle->getDBkey() . "'" .
  21                                 ' AND old_namespace=' . $wgTitle->getNamespace() .
  22                                 " AND old_id<{$this->mNewid} order by old_id desc" );
  23
  24                 } else {
  25                         $this->mOldid = intval($old);
  26                         $this->mNewid = intval($new);
  27                 }
  28                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  29         }
  30
  31         function showDiffPage()
  32         {
  33                 global $wgUser, $wgTitle, $wgOut, $wgLang, $wgOnlySysopsCanPatrol, $wgUseRCPatrol;
  34                 $fname = 'DifferenceEngine::showDiffPage';
  35                 wfProfileIn( $fname );
  36
  37                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  38                   "{$this->mNewid})";
  39                 $mtext = wfMsg( 'missingarticle', $t );
  40
  41                 $wgOut->setArticleFlag( false );
  42                 if ( ! $this->loadText() ) {
  43                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
  44                         $wgOut->addHTML( $mtext );
  45                         wfProfileOut( $fname );
  46                         return;
  47                 }
  48                 $wgOut->suppressQuickbar();
  49
  50                 $oldTitle = $this->mOldPage->getPrefixedText();
  51                 $newTitle = $this->mNewPage->getPrefixedText();
  52                 if( $oldTitle == $newTitle ) {
  53                         $wgOut->setPageTitle( $newTitle );
  54                 } else {
  55                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
  56                 }
  57                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
  58                 $wgOut->setRobotpolicy( 'noindex,follow' );
  59
  60                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  61                         $wgOut->loginToUse();
  62                         $wgOut->output();
  63                         wfProfileOut( $fname );
  64                         exit;
  65                 }
  66
  67                 $sk = $wgUser->getSkin();
  68                 $talk = $wgLang->getNsText( NS_TALK );
  69                 $contribs = wfMsg( 'contribslink' );
  70
  71                 $this->mOldComment = $sk->formatComment($this->mOldComment);
  72                 $this->mNewComment = $sk->formatComment($this->mNewComment);
  73
  74                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
  75                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
  76                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
  77                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
  78                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
  79                         'target=' . urlencode($this->mOldUser) );
  80                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
  81                         'target=' . urlencode($this->mNewUser) );
  82                 if ( !$this->mNewid && $wgUser->isSysop() ) {
  83                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
  84                                 'action=rollback&from=' . urlencode($this->mNewUser) ) . ']</strong>';
  85                 } else {
  86                         $rollback = '';
  87                 }
  88                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->getID() != 0 &&
  89                      ( $wgUser->isSysop() || !$wgOnlySysopsCanPatrol ) )
  90                 {
  91                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
  92                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
  93                 } else {
  94                         $patrol = '';
  95                 }
  96
  97                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
  98                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment;
  99                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 100                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment . $patrol;
 101
 102                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 103                   $oldHeader, $newHeader );
 104                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 105                 $wgOut->addWikiText( $this->mNewtext );
 106
 107                 wfProfileOut( $fname );
 108         }
 109
 110         function showDiff( $otext, $ntext, $otitle, $ntitle )
 111         {
 112                 global $wgOut, $wgUseExternalDiffEngine;
 113
 114                 $otext = str_replace( "\r\n", "\n", $otext );
 115                 $ntext = str_replace( "\r\n", "\n", $ntext );
 116
 117
 118                         $wgOut->addHTML( "<table border='0' width='98%'
 119 cellpadding='0' cellspacing='4px' class='diff'><tr>
 120 <td colspan='2' width='50%' align='center' class='diff-otitle'>
 121 {$otitle}</td>
 122 <td colspan='2' width='50%' align='center' class='diff-ntitle'>
 123 {$ntitle}</td>
 124 </tr>\n" );
 125                 if ( $wgUseExternalDiffEngine ) {
 126                         dl('php_wikidiff.so');
 127                         $wgOut->addHTML( wikidiff_do_diff( $otext, $ntext, 2) );
 128                 } else {
 129                         $ota = explode( "\n", $otext);
 130                         $nta = explode( "\n", $ntext);
 131                         $diffs = new Diff( $ota, $nta );
 132                         $formatter = new TableDiffFormatter();
 133                         $formatter->format( $diffs );
 134                 }
 135                 $wgOut->addHTML( "</table>\n" );
 136         }
 137
 138         # Load the text of the articles to compare.  If newid is 0, then compare
 139         # the old article in oldid to the current article; if oldid is 0, then
 140         # compare the current article to the immediately previous one (ignoring
 141         # the value of newid).
 142         #
 143         function loadText()
 144         {
 145                 global $wgTitle, $wgOut, $wgLang;
 146                 $fname = 'DifferenceEngine::loadText';
 147
 148                 $dbr =& wfGetDB( DB_SLAVE );
 149                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 150                         $wgOut->setArticleFlag( true );
 151                         $newLink = $wgTitle->getLocalUrl();
 152                         $this->mPagetitle = wfMsg( 'currentrev' );
 153                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 154                         $id = $wgTitle->getArticleID();
 155
 156                         $s = $dbr->getArray( 'cur', array( 'cur_text', 'cur_user_text', 'cur_comment' ),
 157                                 array( 'cur_id' => $id ), $fname );
 158                         if ( $s === false ) {
 159                                 return false;
 160                         }
 161
 162                         $this->mNewPage = &$wgTitle;
 163                         $this->mNewtext = $s->cur_text;
 164                         $this->mNewUser = $s->cur_user_text;
 165                         $this->mNewComment = $s->cur_comment;
 166                 } else {
 167                         $s = $dbr->getArray( 'old', array( 'old_namespace','old_title','old_timestamp', 'old_text',
 168                                 'old_flags','old_user_text','old_comment' ), array( 'old_id' => $this->mNewid ), $fname );
 169
 170                         if ( $s === false ) {
 171                                 return false;
 172                         }
 173
 174                         $this->mNewtext = Article::getRevisionText( $s );
 175
 176                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 177                         $this->mNewPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 178                         $newLink = $wgTitle->getLocalUrl ('oldid=' . $this->mNewid);
 179                         $this->mPagetitle = wfMsg( 'revisionasof', $t );
 180                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 181                         $this->mNewUser = $s->old_user_text;
 182                         $this->mNewComment = $s->old_comment;
 183                 }
 184                 if ( 0 == $this->mOldid ) {
 185                         $s = $dbr->getArray( 'old',
 186                                 array( 'old_namespace','old_title','old_timestamp','old_text', 'old_flags','old_user_text','old_comment' ),
 187                                 array( /* WHERE */
 188                                         'old_namespace' => $this->mNewPage->getNamespace(),
 189                                         'old_title' => $this->mNewPage->getDBkey()
 190                                 ), $fname, array( 'ORDER BY' => 'inverse_timestamp', 'USE INDEX' => 'name_title_timestamp' )
 191                         );
 192                 } else {
 193                         $s = $dbr->getArray( 'old',
 194                                 array( 'old_namespace','old_title','old_timestamp','old_text','old_flags','old_user_text','old_comment'),
 195                                 array( 'old_id' => $this->mOldid ),
 196                                 $fname
 197                         );
 198                 }
 199                 if ( $s === false ) {
 200                         return false;
 201                 }
 202                 $this->mOldPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 203                 $this->mOldtext = Article::getRevisionText( $s );
 204
 205                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 206                 $oldLink = $this->mOldPage->getLocalUrl ('oldid=' . $this->mOldid);
 207                 $this->mOldtitle = "<a href='$oldLink'>" . wfMsg( 'revisionasof', $t ) . '</a>';
 208                 $this->mOldUser = $s->old_user_text;
 209                 $this->mOldComment = $s->old_comment;
 210
 211                 return true;
 212         }
 213 }
 214
 215 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 216 //
 217 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 218 // You may copy this code freely under the conditions of the GPL.
 219 //
 220
 221 define('USE_ASSERTS', function_exists('assert'));
 222
 223 class _DiffOp {
 224         var $type;
 225         var $orig;
 226         var $closing;
 227
 228         function reverse() {
 229                 trigger_error('pure virtual', E_USER_ERROR);
 230         }
 231
 232         function norig() {
 233                 return $this->orig ? sizeof($this->orig) : 0;
 234         }
 235
 236         function nclosing() {
 237                 return $this->closing ? sizeof($this->closing) : 0;
 238         }
 239 }
 240
 241 class _DiffOp_Copy extends _DiffOp {
 242         var $type = 'copy';
 243
 244         function _DiffOp_Copy ($orig, $closing = false) {
 245                 if (!is_array($closing))
 246                         $closing = $orig;
 247                 $this->orig = $orig;
 248                 $this->closing = $closing;
 249         }
 250
 251         function reverse() {
 252                 return new _DiffOp_Copy($this->closing, $this->orig);
 253         }
 254 }
 255
 256 class _DiffOp_Delete extends _DiffOp {
 257         var $type = 'delete';
 258
 259         function _DiffOp_Delete ($lines) {
 260                 $this->orig = $lines;
 261                 $this->closing = false;
 262         }
 263
 264         function reverse() {
 265                 return new _DiffOp_Add($this->orig);
 266         }
 267 }
 268
 269 class _DiffOp_Add extends _DiffOp {
 270         var $type = 'add';
 271
 272         function _DiffOp_Add ($lines) {
 273                 $this->closing = $lines;
 274                 $this->orig = false;
 275         }
 276
 277         function reverse() {
 278                 return new _DiffOp_Delete($this->closing);
 279         }
 280 }
 281
 282 class _DiffOp_Change extends _DiffOp {
 283         var $type = 'change';
 284
 285         function _DiffOp_Change ($orig, $closing) {
 286                 $this->orig = $orig;
 287                 $this->closing = $closing;
 288         }
 289
 290         function reverse() {
 291                 return new _DiffOp_Change($this->closing, $this->orig);
 292         }
 293 }
 294
 295
 296 /**
 297  * Class used internally by Diff to actually compute the diffs.
 298  *
 299  * The algorithm used here is mostly lifted from the perl module
 300  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 301  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 302  *
 303  * More ideas are taken from:
 304  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 305  *
 306  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 307  * diffutils-2.7, which can be found at:
 308  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 309  *
 310  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 311  * are my own.
 312  *
 313  * @author Geoffrey T. Dairiki
 314  * @access private
 315  */
 316 class _DiffEngine
 317 {
 318         function diff ($from_lines, $to_lines) {
 319                 $n_from = sizeof($from_lines);
 320                 $n_to = sizeof($to_lines);
 321
 322                 $this->xchanged = $this->ychanged = array();
 323                 $this->xv = $this->yv = array();
 324                 $this->xind = $this->yind = array();
 325                 unset($this->seq);
 326                 unset($this->in_seq);
 327                 unset($this->lcs);
 328
 329                 // Skip leading common lines.
 330                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 331                         if ($from_lines[$skip] != $to_lines[$skip])
 332                                 break;
 333                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 334                 }
 335                 // Skip trailing common lines.
 336                 $xi = $n_from; $yi = $n_to;
 337                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 338                         if ($from_lines[$xi] != $to_lines[$yi])
 339                                 break;
 340                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 341                 }
 342
 343                 // Ignore lines which do not exist in both files.
 344                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 345                         $xhash[$from_lines[$xi]] = 1;
 346                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 347                         $line = $to_lines[$yi];
 348                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 349                                 continue;
 350                         $yhash[$line] = 1;
 351                         $this->yv[] = $line;
 352                         $this->yind[] = $yi;
 353                 }
 354                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 355                         $line = $from_lines[$xi];
 356                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 357                                 continue;
 358                         $this->xv[] = $line;
 359                         $this->xind[] = $xi;
 360                 }
 361
 362                 // Find the LCS.
 363                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 364
 365                 // Merge edits when possible
 366                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 367                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 368
 369                 // Compute the edit operations.
 370                 $edits = array();
 371                 $xi = $yi = 0;
 372                 while ($xi < $n_from || $yi < $n_to) {
 373                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 374                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 375
 376                         // Skip matching "snake".
 377                         $copy = array();
 378                         while ( $xi < $n_from && $yi < $n_to
 379                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 380                                 $copy[] = $from_lines[$xi++];
 381                                 ++$yi;
 382                         }
 383                         if ($copy)
 384                                 $edits[] = new _DiffOp_Copy($copy);
 385
 386                         // Find deletes & adds.
 387                         $delete = array();
 388                         while ($xi < $n_from && $this->xchanged[$xi])
 389                                 $delete[] = $from_lines[$xi++];
 390
 391                         $add = array();
 392                         while ($yi < $n_to && $this->ychanged[$yi])
 393                                 $add[] = $to_lines[$yi++];
 394
 395                         if ($delete && $add)
 396                                 $edits[] = new _DiffOp_Change($delete, $add);
 397                         elseif ($delete)
 398                                 $edits[] = new _DiffOp_Delete($delete);
 399                         elseif ($add)
 400                                 $edits[] = new _DiffOp_Add($add);
 401                 }
 402                 return $edits;
 403         }
 404
 405
 406         /* Divide the Largest Common Subsequence (LCS) of the sequences
 407          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 408          * sized segments.
 409          *
 410          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 411          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 412          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 413          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 414          * that (X0, Y0) == (XOFF, YOFF) and
 415          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 416          *
 417          * This function assumes that the first lines of the specified portions
 418          * of the two files do not match, and likewise that the last lines do not
 419          * match.  The caller must trim matching lines from the beginning and end
 420          * of the portions it is going to specify.
 421          */
 422         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 423         $flip = false;
 424
 425         if ($xlim - $xoff > $ylim - $yoff) {
 426                 // Things seems faster (I'm not sure I understand why)
 427                         // when the shortest sequence in X.
 428                         $flip = true;
 429                 list ($xoff, $xlim, $yoff, $ylim)
 430                 = array( $yoff, $ylim, $xoff, $xlim);
 431                 }
 432
 433         if ($flip)
 434                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 435                 $ymatches[$this->xv[$i]][] = $i;
 436         else
 437                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 438                 $ymatches[$this->yv[$i]][] = $i;
 439
 440         $this->lcs = 0;
 441         $this->seq[0]= $yoff - 1;
 442         $this->in_seq = array();
 443         $ymids[0] = array();
 444
 445         $numer = $xlim - $xoff + $nchunks - 1;
 446         $x = $xoff;
 447         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 448                 if ($chunk > 0)
 449                 for ($i = 0; $i <= $this->lcs; $i++)
 450                         $ymids[$i][$chunk-1] = $this->seq[$i];
 451
 452                 $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 453                 for ( ; $x < $x1; $x++) {
 454                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 455                                 if (empty($ymatches[$line]))
 456                         continue;
 457                 $matches = $ymatches[$line];
 458                                 reset($matches);
 459                 while (list ($junk, $y) = each($matches))
 460                         if (empty($this->in_seq[$y])) {
 461                         $k = $this->_lcs_pos($y);
 462                         USE_ASSERTS && assert($k > 0);
 463                         $ymids[$k] = $ymids[$k-1];
 464                         break;
 465                                         }
 466                 while (list ($junk, $y) = each($matches)) {
 467                         if ($y > $this->seq[$k-1]) {
 468                         USE_ASSERTS && assert($y < $this->seq[$k]);
 469                         // Optimization: this is a common case:
 470                         //      next match is just replacing previous match.
 471                         $this->in_seq[$this->seq[$k]] = false;
 472                         $this->seq[$k] = $y;
 473                         $this->in_seq[$y] = 1;
 474                                         }
 475                         else if (empty($this->in_seq[$y])) {
 476                         $k = $this->_lcs_pos($y);
 477                         USE_ASSERTS && assert($k > 0);
 478                         $ymids[$k] = $ymids[$k-1];
 479                                         }
 480                                 }
 481                         }
 482                 }
 483
 484         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 485         $ymid = $ymids[$this->lcs];
 486         for ($n = 0; $n < $nchunks - 1; $n++) {
 487                 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 488                 $y1 = $ymid[$n] + 1;
 489                 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 490                 }
 491         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 492
 493         return array($this->lcs, $seps);
 494         }
 495
 496         function _lcs_pos ($ypos) {
 497         $end = $this->lcs;
 498         if ($end == 0 || $ypos > $this->seq[$end]) {
 499                 $this->seq[++$this->lcs] = $ypos;
 500                 $this->in_seq[$ypos] = 1;
 501                 return $this->lcs;
 502                 }
 503
 504         $beg = 1;
 505         while ($beg < $end) {
 506                 $mid = (int)(($beg + $end) / 2);
 507                 if ( $ypos > $this->seq[$mid] )
 508                 $beg = $mid + 1;
 509                 else
 510                 $end = $mid;
 511                 }
 512
 513         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 514
 515         $this->in_seq[$this->seq[$end]] = false;
 516         $this->seq[$end] = $ypos;
 517         $this->in_seq[$ypos] = 1;
 518         return $end;
 519         }
 520
 521         /* Find LCS of two sequences.
 522          *
 523          * The results are recorded in the vectors $this->{x,y}changed[], by
 524          * storing a 1 in the element for each line that is an insertion
 525          * or deletion (ie. is not in the LCS).
 526          *
 527          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 528          *
 529          * Note that XLIM, YLIM are exclusive bounds.
 530          * All line numbers are origin-0 and discarded lines are not counted.
 531          */
 532         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 533         // Slide down the bottom initial diagonal.
 534         while ($xoff < $xlim && $yoff < $ylim
 535                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 536                 ++$xoff;
 537                 ++$yoff;
 538                 }
 539
 540         // Slide up the top initial diagonal.
 541         while ($xlim > $xoff && $ylim > $yoff
 542                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 543                 --$xlim;
 544                 --$ylim;
 545                 }
 546
 547         if ($xoff == $xlim || $yoff == $ylim)
 548                 $lcs = 0;
 549         else {
 550                 // This is ad hoc but seems to work well.
 551                 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 552                 //$nchunks = max(2,min(8,(int)$nchunks));
 553                 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 554                 list ($lcs, $seps)
 555                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 556                 }
 557
 558         if ($lcs == 0) {
 559                 // X and Y sequences have no common subsequence:
 560                 // mark all changed.
 561                 while ($yoff < $ylim)
 562                 $this->ychanged[$this->yind[$yoff++]] = 1;
 563                 while ($xoff < $xlim)
 564                 $this->xchanged[$this->xind[$xoff++]] = 1;
 565                 }
 566         else {
 567                 // Use the partitions to split this problem into subproblems.
 568                 reset($seps);
 569                 $pt1 = $seps[0];
 570                 while ($pt2 = next($seps)) {
 571                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 572                 $pt1 = $pt2;
 573                         }
 574                 }
 575         }
 576
 577         /* Adjust inserts/deletes of identical lines to join changes
 578          * as much as possible.
 579          *
 580          * We do something when a run of changed lines include a
 581          * line at one end and has an excluded, identical line at the other.
 582          * We are free to choose which identical line is included.
 583          * `compareseq' usually chooses the one at the beginning,
 584          * but usually it is cleaner to consider the following identical line
 585          * to be the "change".
 586          *
 587          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 588          */
 589         function _shift_boundaries ($lines, &$changed, $other_changed) {
 590         $i = 0;
 591         $j = 0;
 592
 593         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 594         $len = sizeof($lines);
 595         $other_len = sizeof($other_changed);
 596
 597         while (1) {
 598                 /*
 599                  * Scan forwards to find beginning of another run of changes.
 600                  * Also keep track of the corresponding point in the other file.
 601                  *
 602                  * Throughout this code, $i and $j are adjusted together so that
 603                  * the first $i elements of $changed and the first $j elements
 604                  * of $other_changed both contain the same number of zeros
 605                  * (unchanged lines).
 606                  * Furthermore, $j is always kept so that $j == $other_len or
 607                  * $other_changed[$j] == false.
 608                  */
 609                 while ($j < $other_len && $other_changed[$j])
 610                 $j++;
 611
 612                 while ($i < $len && ! $changed[$i]) {
 613                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 614                 $i++; $j++;
 615                 while ($j < $other_len && $other_changed[$j])
 616                         $j++;
 617                         }
 618
 619                 if ($i == $len)
 620                 break;
 621
 622                 $start = $i;
 623
 624                 // Find the end of this run of changes.
 625                 while (++$i < $len && $changed[$i])
 626                 continue;
 627
 628                 do {
 629                 /*
 630                  * Record the length of this run of changes, so that
 631                  * we can later determine whether the run has grown.
 632                  */
 633                 $runlength = $i - $start;
 634
 635                 /*
 636                  * Move the changed region back, so long as the
 637                  * previous unchanged line matches the last changed one.
 638                  * This merges with previous changed regions.
 639                  */
 640                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 641                         $changed[--$start] = 1;
 642                         $changed[--$i] = false;
 643                         while ($start > 0 && $changed[$start - 1])
 644                         $start--;
 645                         USE_ASSERTS && assert('$j > 0');
 646                         while ($other_changed[--$j])
 647                         continue;
 648                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 649                                 }
 650
 651                 /*
 652                  * Set CORRESPONDING to the end of the changed run, at the last
 653                  * point where it corresponds to a changed run in the other file.
 654                  * CORRESPONDING == LEN means no such point has been found.
 655                  */
 656                 $corresponding = $j < $other_len ? $i : $len;
 657
 658                 /*
 659                  * Move the changed region forward, so long as the
 660                  * first changed line matches the following unchanged one.
 661                  * This merges with following changed regions.
 662                  * Do this second, so that if there are no merges,
 663                  * the changed region is moved forward as far as possible.
 664                  */
 665                 while ($i < $len && $lines[$start] == $lines[$i]) {
 666                         $changed[$start++] = false;
 667                         $changed[$i++] = 1;
 668                         while ($i < $len && $changed[$i])
 669                         $i++;
 670
 671                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 672                         $j++;
 673                         if ($j < $other_len && $other_changed[$j]) {
 674                         $corresponding = $i;
 675                         while ($j < $other_len && $other_changed[$j])
 676                                 $j++;
 677                                         }
 678                                 }
 679                         } while ($runlength != $i - $start);
 680
 681                 /*
 682                  * If possible, move the fully-merged run of changes
 683                  * back to a corresponding run in the other file.
 684                  */
 685                 while ($corresponding < $i) {
 686                 $changed[--$start] = 1;
 687                 $changed[--$i] = 0;
 688                 USE_ASSERTS && assert('$j > 0');
 689                 while ($other_changed[--$j])
 690                         continue;
 691                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 692                         }
 693                 }
 694         }
 695 }
 696
 697 /**
 698  * Class representing a 'diff' between two sequences of strings.
 699  */
 700 class Diff
 701 {
 702         var $edits;
 703
 704         /**
 705          * Constructor.
 706          * Computes diff between sequences of strings.
 707          *
 708          * @param $from_lines array An array of strings.
 709          *                (Typically these are lines from a file.)
 710          * @param $to_lines array An array of strings.
 711          */
 712         function Diff($from_lines, $to_lines) {
 713                 $eng = new _DiffEngine;
 714                 $this->edits = $eng->diff($from_lines, $to_lines);
 715                 //$this->_check($from_lines, $to_lines);
 716         }
 717
 718         /**
 719          * Compute reversed Diff.
 720          *
 721          * SYNOPSIS:
 722          *
 723          *      $diff = new Diff($lines1, $lines2);
 724          *      $rev = $diff->reverse();
 725          * @return object A Diff object representing the inverse of the
 726          *                                original diff.
 727          */
 728         function reverse () {
 729         $rev = $this;
 730                 $rev->edits = array();
 731                 foreach ($this->edits as $edit) {
 732                         $rev->edits[] = $edit->reverse();
 733                 }
 734         return $rev;
 735         }
 736
 737         /**
 738          * Check for empty diff.
 739          *
 740          * @return bool True iff two sequences were identical.
 741          */
 742         function isEmpty () {
 743                 foreach ($this->edits as $edit) {
 744                         if ($edit->type != 'copy')
 745                                 return false;
 746                 }
 747                 return true;
 748         }
 749
 750         /**
 751          * Compute the length of the Longest Common Subsequence (LCS).
 752          *
 753          * This is mostly for diagnostic purposed.
 754          *
 755          * @return int The length of the LCS.
 756          */
 757         function lcs () {
 758         $lcs = 0;
 759                 foreach ($this->edits as $edit) {
 760                         if ($edit->type == 'copy')
 761                                 $lcs += sizeof($edit->orig);
 762                 }
 763         return $lcs;
 764         }
 765
 766         /**
 767          * Get the original set of lines.
 768          *
 769          * This reconstructs the $from_lines parameter passed to the
 770          * constructor.
 771          *
 772          * @return array The original sequence of strings.
 773          */
 774         function orig() {
 775                 $lines = array();
 776
 777                 foreach ($this->edits as $edit) {
 778                         if ($edit->orig)
 779                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 780                 }
 781                 return $lines;
 782         }
 783
 784         /**
 785          * Get the closing set of lines.
 786          *
 787          * This reconstructs the $to_lines parameter passed to the
 788          * constructor.
 789          *
 790          * @return array The sequence of strings.
 791          */
 792         function closing() {
 793                 $lines = array();
 794
 795                 foreach ($this->edits as $edit) {
 796                         if ($edit->closing)
 797                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 798                 }
 799                 return $lines;
 800         }
 801
 802         /**
 803          * Check a Diff for validity.
 804          *
 805          * This is here only for debugging purposes.
 806          */
 807         function _check ($from_lines, $to_lines) {
 808                 if (serialize($from_lines) != serialize($this->orig()))
 809                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 810                 if (serialize($to_lines) != serialize($this->closing()))
 811                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 812
 813                 $rev = $this->reverse();
 814                 if (serialize($to_lines) != serialize($rev->orig()))
 815                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 816                 if (serialize($from_lines) != serialize($rev->closing()))
 817                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 818
 819
 820                 $prevtype = 'none';
 821                 foreach ($this->edits as $edit) {
 822                         if ( $prevtype == $edit->type )
 823                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 824                         $prevtype = $edit->type;
 825                 }
 826
 827                 $lcs = $this->lcs();
 828                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 829         }
 830 }
 831
 832 /**
 833  * FIXME: bad name.
 834  */
 835 class MappedDiff
 836 extends Diff
 837 {
 838         /**
 839          * Constructor.
 840          *
 841          * Computes diff between sequences of strings.
 842          *
 843          * This can be used to compute things like
 844          * case-insensitve diffs, or diffs which ignore
 845          * changes in white-space.
 846          *
 847          * @param $from_lines array An array of strings.
 848          *      (Typically these are lines from a file.)
 849          *
 850          * @param $to_lines array An array of strings.
 851          *
 852          * @param $mapped_from_lines array This array should
 853          *      have the same size number of elements as $from_lines.
 854          *      The elements in $mapped_from_lines and
 855          *      $mapped_to_lines are what is actually compared
 856          *      when computing the diff.
 857          *
 858          * @param $mapped_to_lines array This array should
 859          *      have the same number of elements as $to_lines.
 860          */
 861         function MappedDiff($from_lines, $to_lines,
 862                                                 $mapped_from_lines, $mapped_to_lines) {
 863
 864                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 865                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 866
 867                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 868
 869                 $xi = $yi = 0;
 870                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 871                         $orig = &$this->edits[$i]->orig;
 872                         if (is_array($orig)) {
 873                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 874                                 $xi += sizeof($orig);
 875                         }
 876
 877                         $closing = &$this->edits[$i]->closing;
 878                         if (is_array($closing)) {
 879                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 880                                 $yi += sizeof($closing);
 881                         }
 882                 }
 883         }
 884 }
 885
 886 /**
 887  * A class to format Diffs
 888  *
 889  * This class formats the diff in classic diff format.
 890  * It is intended that this class be customized via inheritance,
 891  * to obtain fancier outputs.
 892  */
 893 class DiffFormatter
 894 {
 895         /**
 896          * Number of leading context "lines" to preserve.
 897          *
 898          * This should be left at zero for this class, but subclasses
 899          * may want to set this to other values.
 900          */
 901         var $leading_context_lines = 0;
 902
 903         /**
 904          * Number of trailing context "lines" to preserve.
 905          *
 906          * This should be left at zero for this class, but subclasses
 907          * may want to set this to other values.
 908          */
 909         var $trailing_context_lines = 0;
 910
 911         /**
 912          * Format a diff.
 913          *
 914          * @param $diff object A Diff object.
 915          * @return string The formatted output.
 916          */
 917         function format($diff) {
 918
 919                 $xi = $yi = 1;
 920                 $block = false;
 921                 $context = array();
 922
 923                 $nlead = $this->leading_context_lines;
 924                 $ntrail = $this->trailing_context_lines;
 925
 926                 $this->_start_diff();
 927
 928                 foreach ($diff->edits as $edit) {
 929                         if ($edit->type == 'copy') {
 930                                 if (is_array($block)) {
 931                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 932                                                 $block[] = $edit;
 933                                         }
 934                                         else{
 935                                                 if ($ntrail) {
 936                                                         $context = array_slice($edit->orig, 0, $ntrail);
 937                                                         $block[] = new _DiffOp_Copy($context);
 938                                                 }
 939                                                 $this->_block($x0, $ntrail + $xi - $x0,
 940                                                                           $y0, $ntrail + $yi - $y0,
 941                                                                           $block);
 942                                                 $block = false;
 943                                         }
 944                                 }
 945                                 $context = $edit->orig;
 946                         }
 947                         else {
 948                                 if (! is_array($block)) {
 949                                         $context = array_slice($context, sizeof($context) - $nlead);
 950                                         $x0 = $xi - sizeof($context);
 951                                         $y0 = $yi - sizeof($context);
 952                                         $block = array();
 953                                         if ($context)
 954                                                 $block[] = new _DiffOp_Copy($context);
 955                                 }
 956                                 $block[] = $edit;
 957                         }
 958
 959                         if ($edit->orig)
 960                                 $xi += sizeof($edit->orig);
 961                         if ($edit->closing)
 962                                 $yi += sizeof($edit->closing);
 963                 }
 964
 965                 if (is_array($block))
 966                         $this->_block($x0, $xi - $x0,
 967                                                   $y0, $yi - $y0,
 968                                                   $block);
 969
 970                 return $this->_end_diff();
 971         }
 972
 973         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 974                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 975                 foreach ($edits as $edit) {
 976                         if ($edit->type == 'copy')
 977                                 $this->_context($edit->orig);
 978                         elseif ($edit->type == 'add')
 979                                 $this->_added($edit->closing);
 980                         elseif ($edit->type == 'delete')
 981                                 $this->_deleted($edit->orig);
 982                         elseif ($edit->type == 'change')
 983                                 $this->_changed($edit->orig, $edit->closing);
 984                         else
 985                                 trigger_error('Unknown edit type', E_USER_ERROR);
 986                 }
 987                 $this->_end_block();
 988         }
 989
 990         function _start_diff() {
 991                 ob_start();
 992         }
 993
 994         function _end_diff() {
 995                 $val = ob_get_contents();
 996                 ob_end_clean();
 997                 return $val;
 998         }
 999
1000         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1001                 if ($xlen > 1)
1002                         $xbeg .= "," . ($xbeg + $xlen - 1);
1003                 if ($ylen > 1)
1004                         $ybeg .= "," . ($ybeg + $ylen - 1);
1005
1006                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1007         }
1008
1009         function _start_block($header) {
1010                 echo $header;
1011         }
1012
1013         function _end_block() {
1014         }
1015
1016         function _lines($lines, $prefix = ' ') {
1017                 foreach ($lines as $line)
1018                         echo "$prefix $line\n";
1019         }
1020
1021         function _context($lines) {
1022                 $this->_lines($lines);
1023         }
1024
1025         function _added($lines) {
1026                 $this->_lines($lines, '>');
1027         }
1028         function _deleted($lines) {
1029                 $this->_lines($lines, '<');
1030         }
1031
1032         function _changed($orig, $closing) {
1033                 $this->_deleted($orig);
1034                 echo "---\n";
1035                 $this->_added($closing);
1036         }
1037 }
1038
1039
1040 /**
1041  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1042  *
1043  */
1044
1045 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1046
1047 class _HWLDF_WordAccumulator {
1048         function _HWLDF_WordAccumulator () {
1049                 $this->_lines = array();
1050                 $this->_line = '';
1051                 $this->_group = '';
1052                 $this->_tag = '';
1053         }
1054
1055         function _flushGroup ($new_tag) {
1056                 if ($this->_group !== '') {
1057                         if ($this->_tag == 'mark')
1058                                 $this->_line .= '<span class="diffchange">'.htmlspecialchars ( $this->_group ).'</span>';
1059                         else
1060                                 $this->_line .= htmlspecialchars ( $this->_group );
1061                 }
1062                 $this->_group = '';
1063                 $this->_tag = $new_tag;
1064         }
1065
1066         function _flushLine ($new_tag) {
1067                 $this->_flushGroup($new_tag);
1068                 if ($this->_line != '')
1069                         array_push ( $this->_lines, $this->_line );
1070                 else
1071                         array_push ( $this->_lines, NBSP );
1072                 $this->_line = '';
1073         }
1074
1075         function addWords ($words, $tag = '') {
1076                 if ($tag != $this->_tag)
1077                         $this->_flushGroup($tag);
1078
1079                 foreach ($words as $word) {
1080                         // new-line should only come as first char of word.
1081                         if ($word == '')
1082                                 continue;
1083                         if ($word[0] == "\n") {
1084                                 $this->_flushLine($tag);
1085                                 $word = substr($word, 1);
1086                         }
1087                         assert(!strstr($word, "\n"));
1088                         $this->_group .= $word;
1089                 }
1090         }
1091
1092         function getLines() {
1093                 $this->_flushLine('~done');
1094                 return $this->_lines;
1095         }
1096 }
1097
1098 class WordLevelDiff extends MappedDiff
1099 {
1100         function WordLevelDiff ($orig_lines, $closing_lines) {
1101                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1102                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1103
1104
1105                 $this->MappedDiff($orig_words, $closing_words,
1106                                                   $orig_stripped, $closing_stripped);
1107         }
1108
1109         function _split($lines) {
1110                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1111                                                         implode("\n", $lines),
1112                                                         $m)) {
1113                         return array(array(''), array(''));
1114                 }
1115                 return array($m[0], $m[1]);
1116         }
1117
1118         function orig () {
1119                 $orig = new _HWLDF_WordAccumulator;
1120
1121                 foreach ($this->edits as $edit) {
1122                         if ($edit->type == 'copy')
1123                                 $orig->addWords($edit->orig);
1124                         elseif ($edit->orig)
1125                                 $orig->addWords($edit->orig, 'mark');
1126                 }
1127                 return $orig->getLines();
1128         }
1129
1130         function closing () {
1131                 $closing = new _HWLDF_WordAccumulator;
1132
1133                 foreach ($this->edits as $edit) {
1134                         if ($edit->type == 'copy')
1135                                 $closing->addWords($edit->closing);
1136                         elseif ($edit->closing)
1137                                 $closing->addWords($edit->closing, 'mark');
1138                 }
1139                 return $closing->getLines();
1140         }
1141 }
1142
1143 /**
1144  *      Wikipedia Table style diff formatter.
1145  *
1146  */
1147 class TableDiffFormatter extends DiffFormatter
1148 {
1149         function TableDiffFormatter() {
1150                 $this->leading_context_lines = 2;
1151                 $this->trailing_context_lines = 2;
1152         }
1153
1154         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1155                 $l1 = wfMsg( 'lineno', $xbeg );
1156                 $l2 = wfMsg( 'lineno', $ybeg );
1157
1158                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1159                         '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1160                 return $r;
1161         }
1162
1163         function _start_block( $header ) {
1164                 global $wgOut;
1165                 $wgOut->addHTML( $header );
1166         }
1167
1168         function _end_block() {
1169         }
1170
1171         function _lines( $lines, $prefix=' ', $color='white' ) {
1172         }
1173
1174         function addedLine( $line ) {
1175                 return '<td>+</td><td class="diff-addedline">' .
1176                         $line . '</td>';
1177         }
1178
1179         function deletedLine( $line ) {
1180                 return '<td>-</td><td class="diff-deletedline">' .
1181                         $line . '</td>';
1182         }
1183
1184         function emptyLine() {
1185                 return '<td colspan="2">&nbsp;</td>';
1186         }
1187
1188         function contextLine( $line ) {
1189                 return '<td> </td><td class="diff-context">' .
1190                         htmlspecialchars ( $line ) . '</td>';
1191         }
1192
1193         function _added($lines) {
1194                 global $wgOut;
1195                 foreach ($lines as $line) {
1196                         $wgOut->addHTML( '<tr>' . $this->emptyLine() .
1197                           $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n" );
1198                 }
1199         }
1200
1201         function _deleted($lines) {
1202                 global $wgOut;
1203                 foreach ($lines as $line) {
1204                         $wgOut->addHTML( '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1205                           $this->emptyLine() . "</tr>\n" );
1206                 }
1207         }
1208
1209         function _context( $lines ) {
1210                 global $wgOut;
1211                 foreach ($lines as $line) {
1212                         $wgOut->addHTML( '<tr>' . $this->contextLine( $line ) .
1213                           $this->contextLine( $line ) . "</tr>\n" );
1214                 }
1215         }
1216
1217         function _changed( $orig, $closing ) {
1218                 global $wgOut;
1219                 $diff = new WordLevelDiff( $orig, $closing );
1220                 $del = $diff->orig();
1221                 $add = $diff->closing();
1222
1223                 while ( $line = array_shift( $del ) ) {
1224                         $aline = array_shift( $add );
1225                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1226                           $this->addedLine( $aline ) . "</tr>\n" );
1227                 }
1228                 $this->_added( $add ); # If any leftovers
1229         }
1230 }
1231
1232 ?>