includes/DifferenceEngine.php

   1 <?php
   2 # See diff.doc
   3
   4 class DifferenceEngine {
   5         /* private */ var $mOldid, $mNewid;
   6         /* private */ var $mOldtitle, $mNewtitle;
   7         /* private */ var $mOldtext, $mNewtext;
   8         /* private */ var $mOldUser, $mNewUser;
   9         /* private */ var $mOldComment, $mNewComment;
  10         /* private */ var $mOldPage, $mNewPage;
  11
  12         function DifferenceEngine( $old, $new )
  13         {
  14                 $this->mOldid = $old;
  15                 $this->mNewid = $new;
  16         }
  17
  18         function showDiffPage()
  19         {
  20                 global $wgUser, $wgTitle, $wgOut, $wgLang;
  21                 $fname = "DifferenceEngine::showDiffPage";
  22                 wfProfileIn( $fname );
  23
  24                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  25                   "{$this->mNewid})";
  26                 $mtext = wfMsg( "missingarticle", $t );
  27
  28                 $wgOut->setArticleFlag( false );
  29                 if ( ! $this->loadText() ) {
  30                         $wgOut->setPagetitle( wfMsg( "errorpagetitle" ) );
  31                         $wgOut->addHTML( $mtext );
  32                         wfProfileOut( $fname );
  33                         return;
  34                 }
  35                 $wgOut->suppressQuickbar();
  36
  37                 $oldTitle = $this->mOldPage->getPrefixedText();
  38                 $newTitle = $this->mNewPage->getPrefixedText();
  39                 if( $oldTitle == $newTitle ) {
  40                         $wgOut->setPageTitle( $newTitle );
  41                 } else {
  42                         $wgOut->setPageTitle( $oldTitle . ", " . $newTitle );
  43                 }
  44                 $wgOut->setSubtitle( wfMsg( "difference" ) );
  45                 $wgOut->setRobotpolicy( "noindex,follow" );
  46
  47                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  48                         $wgOut->loginToUse();
  49                         $wgOut->output();
  50                         wfProfileOut( $fname );
  51                         exit;
  52                 }
  53
  54                 $sk = $wgUser->getSkin();
  55                 $talk = $wgLang->getNsText( NS_TALK );
  56                 $contribs = wfMsg( "contribslink" );
  57
  58
  59                 $this->mOldComment = $sk->formatComment($this->mOldComment);
  60                 $this->mNewComment = $sk->formatComment($this->mNewComment);
  61
  62
  63                 $oldUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mOldUser ), $this->mOldUser );
  64                 $newUserLink = $sk->makeLinkObj( Title::makeTitle( NS_USER, $this->mNewUser ), $this->mNewUser );
  65                 $oldUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mOldUser ), $talk );
  66                 $newUTLink = $sk->makeLinkObj( Title::makeTitle( NS_USER_TALK, $this->mNewUser ), $talk );
  67                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  68                         "target=" . urlencode($this->mOldUser) );
  69                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, "Contributions" ), $contribs,
  70                         "target=" . urlencode($this->mNewUser) );
  71                 if ( !$this->mNewid && $wgUser->isSysop() ) {
  72                         $rollback = "&nbsp;&nbsp;&nbsp;<strong>[" . $sk->makeKnownLinkObj( $wgTitle, wfMsg( "rollbacklink" ),
  73                                 "action=rollback&from=" . urlencode($this->mNewUser) ) . "]</strong>";
  74                 } else {
  75                         $rollback = "";
  76                 }
  77
  78                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink ($oldUTLink | $oldContribs)<br />".
  79                   $this->mOldComment;
  80                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink ($newUTLink | $newContribs) $rollback<br />".
  81                   $this->mNewComment;
  82
  83                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
  84                   $oldHeader, $newHeader );
  85                 $wgOut->addHTML( "<hr /><h2>{$this->mNewtitle}</h2>\n" );
  86                 $wgOut->addWikiText( $this->mNewtext );
  87
  88                 wfProfileOut( $fname );
  89         }
  90
  91         function showDiff( $otext, $ntext, $otitle, $ntitle )
  92         {
  93                 global $wgOut, $wgUseExternalDiffEngine;
  94
  95                 $otext = str_replace( "\r\n", "\n", htmlspecialchars( $otext ) );
  96                 $ntext = str_replace( "\r\n", "\n", htmlspecialchars( $ntext ) );
  97
  98
  99                         $wgOut->addHTML( "<table border='0' width='98%'
 100 cellpadding='0' cellspacing='4px' class='diff'><tr>
 101 <td colspan='2' width='50%' align='center' class='diff-otitle'>
 102 {$otitle}</td>
 103 <td colspan='2' width='50%' align='center' class='diff-ntitle'>
 104 {$ntitle}</td>
 105 </tr>\n" );
 106                 if ( $wgUseExternalDiffEngine ) {
 107                         dl("php_wikidiff.so");
 108                         $wgOut->addHTML( wikidiff_do_diff( $otext, $ntext, 2) );
 109                 } else {
 110                         $ota = explode( "\n", $otext);
 111                         $nta = explode( "\n", $ntext);
 112                         $diffs = new Diff( $ota, $nta );
 113                         $formatter = new TableDiffFormatter();
 114                         $formatter->format( $diffs );
 115                 }
 116                 $wgOut->addHTML( "</table>\n" );
 117         }
 118
 119         # Load the text of the articles to compare.  If newid is 0, then compare
 120         # the old article in oldid to the current article; if oldid is 0, then
 121         # compare the current article to the immediately previous one (ignoring
 122         # the value of newid).
 123         #
 124         function loadText()
 125         {
 126                 global $wgTitle, $wgOut, $wgLang;
 127                 $fname = "DifferenceEngine::loadText";
 128
 129                 $dbr =& wfGetDB( DB_SLAVE );
 130                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 131                         $wgOut->setArticleFlag( true );
 132                         $this->mNewtitle = wfMsg( "currentrev" );
 133                         $id = $wgTitle->getArticleID();
 134
 135                         $s = $dbr->getArray( 'cur', array( 'cur_text', 'cur_user_text', 'cur_comment' ),
 136                                 array( 'cur_id' => $id ), $fname );
 137                         if ( $s === false ) {
 138                                 return false;
 139                         }
 140
 141                         $this->mNewPage = &$wgTitle;
 142                         $this->mNewtext = $s->cur_text;
 143                         $this->mNewUser = $s->cur_user_text;
 144                         $this->mNewComment = $s->cur_comment;
 145                 } else {
 146                         $s = $dbr->getArray( 'old', array( 'old_namespace','old_title','old_timestamp', 'old_text',
 147                                 'old_flags','old_user_text','old_comment' ), array( 'old_id' => $this->mNewid ), $fname );
 148
 149                         if ( $s === false ) {
 150                                 return false;
 151                         }
 152
 153                         $this->mNewtext = Article::getRevisionText( $s );
 154
 155                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 156                         $this->mNewPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 157                         $this->mNewtitle = wfMsg( "revisionasof", $t );
 158                         $this->mNewUser = $s->old_user_text;
 159                         $this->mNewComment = $s->old_comment;
 160                 }
 161                 if ( 0 == $this->mOldid ) {
 162                         $s = $dbr->getArray( 'old',
 163                                 array( 'old_namespace','old_title','old_timestamp','old_text', 'old_flags','old_user_text','old_comment' ),
 164                                 array( /* WHERE */
 165                                         'old_namespace' => $this->mNewPage->getNamespace(),
 166                                         'old_title' => $this->mNewPage->getDBkey()
 167                                 ), $fname, array( 'ORDER BY' => 'inverse_timestamp', 'USE INDEX' => 'name_title_timestamp' )
 168                         );
 169                 } else {
 170                         $s = $dbr->getArray( 'old',
 171                                 array( 'old_namespace','old_title','old_timestamp','old_text','old_flags','old_user_text','old_comment'),
 172                                 array( 'old_id' => $this->mOldid ),
 173                                 $fname
 174                         );
 175                 }
 176                 if ( $s === false ) {
 177                         return false;
 178                 }
 179                 $this->mOldPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 180                 $this->mOldtext = Article::getRevisionText( $s );
 181
 182                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 183                 $this->mOldtitle = wfMsg( "revisionasof", $t );
 184                 $this->mOldUser = $s->old_user_text;
 185                 $this->mOldComment = $s->old_comment;
 186
 187                 return true;
 188         }
 189 }
 190
 191 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 192 //
 193 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 194 // You may copy this code freely under the conditions of the GPL.
 195 //
 196
 197 define('USE_ASSERTS', function_exists('assert'));
 198
 199 class _DiffOp {
 200         var $type;
 201         var $orig;
 202         var $closing;
 203
 204         function reverse() {
 205                 trigger_error("pure virtual", E_USER_ERROR);
 206         }
 207
 208         function norig() {
 209                 return $this->orig ? sizeof($this->orig) : 0;
 210         }
 211
 212         function nclosing() {
 213                 return $this->closing ? sizeof($this->closing) : 0;
 214         }
 215 }
 216
 217 class _DiffOp_Copy extends _DiffOp {
 218         var $type = 'copy';
 219
 220         function _DiffOp_Copy ($orig, $closing = false) {
 221                 if (!is_array($closing))
 222                         $closing = $orig;
 223                 $this->orig = $orig;
 224                 $this->closing = $closing;
 225         }
 226
 227         function reverse() {
 228                 return new _DiffOp_Copy($this->closing, $this->orig);
 229         }
 230 }
 231
 232 class _DiffOp_Delete extends _DiffOp {
 233         var $type = 'delete';
 234
 235         function _DiffOp_Delete ($lines) {
 236                 $this->orig = $lines;
 237                 $this->closing = false;
 238         }
 239
 240         function reverse() {
 241                 return new _DiffOp_Add($this->orig);
 242         }
 243 }
 244
 245 class _DiffOp_Add extends _DiffOp {
 246         var $type = 'add';
 247
 248         function _DiffOp_Add ($lines) {
 249                 $this->closing = $lines;
 250                 $this->orig = false;
 251         }
 252
 253         function reverse() {
 254                 return new _DiffOp_Delete($this->closing);
 255         }
 256 }
 257
 258 class _DiffOp_Change extends _DiffOp {
 259         var $type = 'change';
 260
 261         function _DiffOp_Change ($orig, $closing) {
 262                 $this->orig = $orig;
 263                 $this->closing = $closing;
 264         }
 265
 266         function reverse() {
 267                 return new _DiffOp_Change($this->closing, $this->orig);
 268         }
 269 }
 270
 271
 272 /**
 273  * Class used internally by Diff to actually compute the diffs.
 274  *
 275  * The algorithm used here is mostly lifted from the perl module
 276  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 277  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 278  *
 279  * More ideas are taken from:
 280  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 281  *
 282  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 283  * diffutils-2.7, which can be found at:
 284  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 285  *
 286  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 287  * are my own.
 288  *
 289  * @author Geoffrey T. Dairiki
 290  * @access private
 291  */
 292 class _DiffEngine
 293 {
 294         function diff ($from_lines, $to_lines) {
 295                 $n_from = sizeof($from_lines);
 296                 $n_to = sizeof($to_lines);
 297
 298                 $this->xchanged = $this->ychanged = array();
 299                 $this->xv = $this->yv = array();
 300                 $this->xind = $this->yind = array();
 301                 unset($this->seq);
 302                 unset($this->in_seq);
 303                 unset($this->lcs);
 304
 305                 // Skip leading common lines.
 306                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 307                         if ($from_lines[$skip] != $to_lines[$skip])
 308                                 break;
 309                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 310                 }
 311                 // Skip trailing common lines.
 312                 $xi = $n_from; $yi = $n_to;
 313                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 314                         if ($from_lines[$xi] != $to_lines[$yi])
 315                                 break;
 316                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 317                 }
 318
 319                 // Ignore lines which do not exist in both files.
 320                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 321                         $xhash[$from_lines[$xi]] = 1;
 322                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 323                         $line = $to_lines[$yi];
 324                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 325                                 continue;
 326                         $yhash[$line] = 1;
 327                         $this->yv[] = $line;
 328                         $this->yind[] = $yi;
 329                 }
 330                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 331                         $line = $from_lines[$xi];
 332                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 333                                 continue;
 334                         $this->xv[] = $line;
 335                         $this->xind[] = $xi;
 336                 }
 337
 338                 // Find the LCS.
 339                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 340
 341                 // Merge edits when possible
 342                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 343                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 344
 345                 // Compute the edit operations.
 346                 $edits = array();
 347                 $xi = $yi = 0;
 348                 while ($xi < $n_from || $yi < $n_to) {
 349                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 350                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 351
 352                         // Skip matching "snake".
 353                         $copy = array();
 354                         while ( $xi < $n_from && $yi < $n_to
 355                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 356                                 $copy[] = $from_lines[$xi++];
 357                                 ++$yi;
 358                         }
 359                         if ($copy)
 360                                 $edits[] = new _DiffOp_Copy($copy);
 361
 362                         // Find deletes & adds.
 363                         $delete = array();
 364                         while ($xi < $n_from && $this->xchanged[$xi])
 365                                 $delete[] = $from_lines[$xi++];
 366
 367                         $add = array();
 368                         while ($yi < $n_to && $this->ychanged[$yi])
 369                                 $add[] = $to_lines[$yi++];
 370
 371                         if ($delete && $add)
 372                                 $edits[] = new _DiffOp_Change($delete, $add);
 373                         elseif ($delete)
 374                                 $edits[] = new _DiffOp_Delete($delete);
 375                         elseif ($add)
 376                                 $edits[] = new _DiffOp_Add($add);
 377                 }
 378                 return $edits;
 379         }
 380
 381
 382         /* Divide the Largest Common Subsequence (LCS) of the sequences
 383          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 384          * sized segments.
 385          *
 386          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 387          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 388          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 389          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 390          * that (X0, Y0) == (XOFF, YOFF) and
 391          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 392          *
 393          * This function assumes that the first lines of the specified portions
 394          * of the two files do not match, and likewise that the last lines do not
 395          * match.  The caller must trim matching lines from the beginning and end
 396          * of the portions it is going to specify.
 397          */
 398         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 399         $flip = false;
 400
 401         if ($xlim - $xoff > $ylim - $yoff) {
 402                 // Things seems faster (I'm not sure I understand why)
 403                         // when the shortest sequence in X.
 404                         $flip = true;
 405                 list ($xoff, $xlim, $yoff, $ylim)
 406                 = array( $yoff, $ylim, $xoff, $xlim);
 407                 }
 408
 409         if ($flip)
 410                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 411                 $ymatches[$this->xv[$i]][] = $i;
 412         else
 413                 for ($i = $ylim - 1; $i >= $yoff; $i--)
 414                 $ymatches[$this->yv[$i]][] = $i;
 415
 416         $this->lcs = 0;
 417         $this->seq[0]= $yoff - 1;
 418         $this->in_seq = array();
 419         $ymids[0] = array();
 420
 421         $numer = $xlim - $xoff + $nchunks - 1;
 422         $x = $xoff;
 423         for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 424                 if ($chunk > 0)
 425                 for ($i = 0; $i <= $this->lcs; $i++)
 426                         $ymids[$i][$chunk-1] = $this->seq[$i];
 427
 428                 $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 429                 for ( ; $x < $x1; $x++) {
 430                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 431                                 if (empty($ymatches[$line]))
 432                         continue;
 433                 $matches = $ymatches[$line];
 434                                 reset($matches);
 435                 while (list ($junk, $y) = each($matches))
 436                         if (empty($this->in_seq[$y])) {
 437                         $k = $this->_lcs_pos($y);
 438                         USE_ASSERTS && assert($k > 0);
 439                         $ymids[$k] = $ymids[$k-1];
 440                         break;
 441                                         }
 442                 while (list ($junk, $y) = each($matches)) {
 443                         if ($y > $this->seq[$k-1]) {
 444                         USE_ASSERTS && assert($y < $this->seq[$k]);
 445                         // Optimization: this is a common case:
 446                         //      next match is just replacing previous match.
 447                         $this->in_seq[$this->seq[$k]] = false;
 448                         $this->seq[$k] = $y;
 449                         $this->in_seq[$y] = 1;
 450                                         }
 451                         else if (empty($this->in_seq[$y])) {
 452                         $k = $this->_lcs_pos($y);
 453                         USE_ASSERTS && assert($k > 0);
 454                         $ymids[$k] = $ymids[$k-1];
 455                                         }
 456                                 }
 457                         }
 458                 }
 459
 460         $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 461         $ymid = $ymids[$this->lcs];
 462         for ($n = 0; $n < $nchunks - 1; $n++) {
 463                 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 464                 $y1 = $ymid[$n] + 1;
 465                 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 466                 }
 467         $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 468
 469         return array($this->lcs, $seps);
 470         }
 471
 472         function _lcs_pos ($ypos) {
 473         $end = $this->lcs;
 474         if ($end == 0 || $ypos > $this->seq[$end]) {
 475                 $this->seq[++$this->lcs] = $ypos;
 476                 $this->in_seq[$ypos] = 1;
 477                 return $this->lcs;
 478                 }
 479
 480         $beg = 1;
 481         while ($beg < $end) {
 482                 $mid = (int)(($beg + $end) / 2);
 483                 if ( $ypos > $this->seq[$mid] )
 484                 $beg = $mid + 1;
 485                 else
 486                 $end = $mid;
 487                 }
 488
 489         USE_ASSERTS && assert($ypos != $this->seq[$end]);
 490
 491         $this->in_seq[$this->seq[$end]] = false;
 492         $this->seq[$end] = $ypos;
 493         $this->in_seq[$ypos] = 1;
 494         return $end;
 495         }
 496
 497         /* Find LCS of two sequences.
 498          *
 499          * The results are recorded in the vectors $this->{x,y}changed[], by
 500          * storing a 1 in the element for each line that is an insertion
 501          * or deletion (ie. is not in the LCS).
 502          *
 503          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 504          *
 505          * Note that XLIM, YLIM are exclusive bounds.
 506          * All line numbers are origin-0 and discarded lines are not counted.
 507          */
 508         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 509         // Slide down the bottom initial diagonal.
 510         while ($xoff < $xlim && $yoff < $ylim
 511                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 512                 ++$xoff;
 513                 ++$yoff;
 514                 }
 515
 516         // Slide up the top initial diagonal.
 517         while ($xlim > $xoff && $ylim > $yoff
 518                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 519                 --$xlim;
 520                 --$ylim;
 521                 }
 522
 523         if ($xoff == $xlim || $yoff == $ylim)
 524                 $lcs = 0;
 525         else {
 526                 // This is ad hoc but seems to work well.
 527                 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 528                 //$nchunks = max(2,min(8,(int)$nchunks));
 529                 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 530                 list ($lcs, $seps)
 531                 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 532                 }
 533
 534         if ($lcs == 0) {
 535                 // X and Y sequences have no common subsequence:
 536                 // mark all changed.
 537                 while ($yoff < $ylim)
 538                 $this->ychanged[$this->yind[$yoff++]] = 1;
 539                 while ($xoff < $xlim)
 540                 $this->xchanged[$this->xind[$xoff++]] = 1;
 541                 }
 542         else {
 543                 // Use the partitions to split this problem into subproblems.
 544                 reset($seps);
 545                 $pt1 = $seps[0];
 546                 while ($pt2 = next($seps)) {
 547                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 548                 $pt1 = $pt2;
 549                         }
 550                 }
 551         }
 552
 553         /* Adjust inserts/deletes of identical lines to join changes
 554          * as much as possible.
 555          *
 556          * We do something when a run of changed lines include a
 557          * line at one end and has an excluded, identical line at the other.
 558          * We are free to choose which identical line is included.
 559          * `compareseq' usually chooses the one at the beginning,
 560          * but usually it is cleaner to consider the following identical line
 561          * to be the "change".
 562          *
 563          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 564          */
 565         function _shift_boundaries ($lines, &$changed, $other_changed) {
 566         $i = 0;
 567         $j = 0;
 568
 569         USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 570         $len = sizeof($lines);
 571         $other_len = sizeof($other_changed);
 572
 573         while (1) {
 574                 /*
 575                  * Scan forwards to find beginning of another run of changes.
 576                  * Also keep track of the corresponding point in the other file.
 577                  *
 578                  * Throughout this code, $i and $j are adjusted together so that
 579                  * the first $i elements of $changed and the first $j elements
 580                  * of $other_changed both contain the same number of zeros
 581                  * (unchanged lines).
 582                  * Furthermore, $j is always kept so that $j == $other_len or
 583                  * $other_changed[$j] == false.
 584                  */
 585                 while ($j < $other_len && $other_changed[$j])
 586                 $j++;
 587
 588                 while ($i < $len && ! $changed[$i]) {
 589                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 590                 $i++; $j++;
 591                 while ($j < $other_len && $other_changed[$j])
 592                         $j++;
 593                         }
 594
 595                 if ($i == $len)
 596                 break;
 597
 598                 $start = $i;
 599
 600                 // Find the end of this run of changes.
 601                 while (++$i < $len && $changed[$i])
 602                 continue;
 603
 604                 do {
 605                 /*
 606                  * Record the length of this run of changes, so that
 607                  * we can later determine whether the run has grown.
 608                  */
 609                 $runlength = $i - $start;
 610
 611                 /*
 612                  * Move the changed region back, so long as the
 613                  * previous unchanged line matches the last changed one.
 614                  * This merges with previous changed regions.
 615                  */
 616                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 617                         $changed[--$start] = 1;
 618                         $changed[--$i] = false;
 619                         while ($start > 0 && $changed[$start - 1])
 620                         $start--;
 621                         USE_ASSERTS && assert('$j > 0');
 622                         while ($other_changed[--$j])
 623                         continue;
 624                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 625                                 }
 626
 627                 /*
 628                  * Set CORRESPONDING to the end of the changed run, at the last
 629                  * point where it corresponds to a changed run in the other file.
 630                  * CORRESPONDING == LEN means no such point has been found.
 631                  */
 632                 $corresponding = $j < $other_len ? $i : $len;
 633
 634                 /*
 635                  * Move the changed region forward, so long as the
 636                  * first changed line matches the following unchanged one.
 637                  * This merges with following changed regions.
 638                  * Do this second, so that if there are no merges,
 639                  * the changed region is moved forward as far as possible.
 640                  */
 641                 while ($i < $len && $lines[$start] == $lines[$i]) {
 642                         $changed[$start++] = false;
 643                         $changed[$i++] = 1;
 644                         while ($i < $len && $changed[$i])
 645                         $i++;
 646
 647                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 648                         $j++;
 649                         if ($j < $other_len && $other_changed[$j]) {
 650                         $corresponding = $i;
 651                         while ($j < $other_len && $other_changed[$j])
 652                                 $j++;
 653                                         }
 654                                 }
 655                         } while ($runlength != $i - $start);
 656
 657                 /*
 658                  * If possible, move the fully-merged run of changes
 659                  * back to a corresponding run in the other file.
 660                  */
 661                 while ($corresponding < $i) {
 662                 $changed[--$start] = 1;
 663                 $changed[--$i] = 0;
 664                 USE_ASSERTS && assert('$j > 0');
 665                 while ($other_changed[--$j])
 666                         continue;
 667                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 668                         }
 669                 }
 670         }
 671 }
 672
 673 /**
 674  * Class representing a 'diff' between two sequences of strings.
 675  */
 676 class Diff
 677 {
 678         var $edits;
 679
 680         /**
 681          * Constructor.
 682          * Computes diff between sequences of strings.
 683          *
 684          * @param $from_lines array An array of strings.
 685          *                (Typically these are lines from a file.)
 686          * @param $to_lines array An array of strings.
 687          */
 688         function Diff($from_lines, $to_lines) {
 689                 $eng = new _DiffEngine;
 690                 $this->edits = $eng->diff($from_lines, $to_lines);
 691                 //$this->_check($from_lines, $to_lines);
 692         }
 693
 694         /**
 695          * Compute reversed Diff.
 696          *
 697          * SYNOPSIS:
 698          *
 699          *      $diff = new Diff($lines1, $lines2);
 700          *      $rev = $diff->reverse();
 701          * @return object A Diff object representing the inverse of the
 702          *                                original diff.
 703          */
 704         function reverse () {
 705         $rev = $this;
 706                 $rev->edits = array();
 707                 foreach ($this->edits as $edit) {
 708                         $rev->edits[] = $edit->reverse();
 709                 }
 710         return $rev;
 711         }
 712
 713         /**
 714          * Check for empty diff.
 715          *
 716          * @return bool True iff two sequences were identical.
 717          */
 718         function isEmpty () {
 719                 foreach ($this->edits as $edit) {
 720                         if ($edit->type != 'copy')
 721                                 return false;
 722                 }
 723                 return true;
 724         }
 725
 726         /**
 727          * Compute the length of the Longest Common Subsequence (LCS).
 728          *
 729          * This is mostly for diagnostic purposed.
 730          *
 731          * @return int The length of the LCS.
 732          */
 733         function lcs () {
 734         $lcs = 0;
 735                 foreach ($this->edits as $edit) {
 736                         if ($edit->type == 'copy')
 737                                 $lcs += sizeof($edit->orig);
 738                 }
 739         return $lcs;
 740         }
 741
 742         /**
 743          * Get the original set of lines.
 744          *
 745          * This reconstructs the $from_lines parameter passed to the
 746          * constructor.
 747          *
 748          * @return array The original sequence of strings.
 749          */
 750         function orig() {
 751                 $lines = array();
 752
 753                 foreach ($this->edits as $edit) {
 754                         if ($edit->orig)
 755                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 756                 }
 757                 return $lines;
 758         }
 759
 760         /**
 761          * Get the closing set of lines.
 762          *
 763          * This reconstructs the $to_lines parameter passed to the
 764          * constructor.
 765          *
 766          * @return array The sequence of strings.
 767          */
 768         function closing() {
 769                 $lines = array();
 770
 771                 foreach ($this->edits as $edit) {
 772                         if ($edit->closing)
 773                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 774                 }
 775                 return $lines;
 776         }
 777
 778         /**
 779          * Check a Diff for validity.
 780          *
 781          * This is here only for debugging purposes.
 782          */
 783         function _check ($from_lines, $to_lines) {
 784                 if (serialize($from_lines) != serialize($this->orig()))
 785                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 786                 if (serialize($to_lines) != serialize($this->closing()))
 787                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 788
 789                 $rev = $this->reverse();
 790                 if (serialize($to_lines) != serialize($rev->orig()))
 791                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 792                 if (serialize($from_lines) != serialize($rev->closing()))
 793                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 794
 795
 796                 $prevtype = 'none';
 797                 foreach ($this->edits as $edit) {
 798                         if ( $prevtype == $edit->type )
 799                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 800                         $prevtype = $edit->type;
 801                 }
 802
 803                 $lcs = $this->lcs();
 804                 trigger_error("Diff okay: LCS = $lcs", E_USER_NOTICE);
 805         }
 806 }
 807
 808 /**
 809  * FIXME: bad name.
 810  */
 811 class MappedDiff
 812 extends Diff
 813 {
 814         /**
 815          * Constructor.
 816          *
 817          * Computes diff between sequences of strings.
 818          *
 819          * This can be used to compute things like
 820          * case-insensitve diffs, or diffs which ignore
 821          * changes in white-space.
 822          *
 823          * @param $from_lines array An array of strings.
 824          *      (Typically these are lines from a file.)
 825          *
 826          * @param $to_lines array An array of strings.
 827          *
 828          * @param $mapped_from_lines array This array should
 829          *      have the same size number of elements as $from_lines.
 830          *      The elements in $mapped_from_lines and
 831          *      $mapped_to_lines are what is actually compared
 832          *      when computing the diff.
 833          *
 834          * @param $mapped_to_lines array This array should
 835          *      have the same number of elements as $to_lines.
 836          */
 837         function MappedDiff($from_lines, $to_lines,
 838                                                 $mapped_from_lines, $mapped_to_lines) {
 839
 840                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
 841                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
 842
 843                 $this->Diff($mapped_from_lines, $mapped_to_lines);
 844
 845                 $xi = $yi = 0;
 846                 for ($i = 0; $i < sizeof($this->edits); $i++) {
 847                         $orig = &$this->edits[$i]->orig;
 848                         if (is_array($orig)) {
 849                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
 850                                 $xi += sizeof($orig);
 851                         }
 852
 853                         $closing = &$this->edits[$i]->closing;
 854                         if (is_array($closing)) {
 855                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
 856                                 $yi += sizeof($closing);
 857                         }
 858                 }
 859         }
 860 }
 861
 862 /**
 863  * A class to format Diffs
 864  *
 865  * This class formats the diff in classic diff format.
 866  * It is intended that this class be customized via inheritance,
 867  * to obtain fancier outputs.
 868  */
 869 class DiffFormatter
 870 {
 871         /**
 872          * Number of leading context "lines" to preserve.
 873          *
 874          * This should be left at zero for this class, but subclasses
 875          * may want to set this to other values.
 876          */
 877         var $leading_context_lines = 0;
 878
 879         /**
 880          * Number of trailing context "lines" to preserve.
 881          *
 882          * This should be left at zero for this class, but subclasses
 883          * may want to set this to other values.
 884          */
 885         var $trailing_context_lines = 0;
 886
 887         /**
 888          * Format a diff.
 889          *
 890          * @param $diff object A Diff object.
 891          * @return string The formatted output.
 892          */
 893         function format($diff) {
 894
 895                 $xi = $yi = 1;
 896                 $block = false;
 897                 $context = array();
 898
 899                 $nlead = $this->leading_context_lines;
 900                 $ntrail = $this->trailing_context_lines;
 901
 902                 $this->_start_diff();
 903
 904                 foreach ($diff->edits as $edit) {
 905                         if ($edit->type == 'copy') {
 906                                 if (is_array($block)) {
 907                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
 908                                                 $block[] = $edit;
 909                                         }
 910                                         else{
 911                                                 if ($ntrail) {
 912                                                         $context = array_slice($edit->orig, 0, $ntrail);
 913                                                         $block[] = new _DiffOp_Copy($context);
 914                                                 }
 915                                                 $this->_block($x0, $ntrail + $xi - $x0,
 916                                                                           $y0, $ntrail + $yi - $y0,
 917                                                                           $block);
 918                                                 $block = false;
 919                                         }
 920                                 }
 921                                 $context = $edit->orig;
 922                         }
 923                         else {
 924                                 if (! is_array($block)) {
 925                                         $context = array_slice($context, sizeof($context) - $nlead);
 926                                         $x0 = $xi - sizeof($context);
 927                                         $y0 = $yi - sizeof($context);
 928                                         $block = array();
 929                                         if ($context)
 930                                                 $block[] = new _DiffOp_Copy($context);
 931                                 }
 932                                 $block[] = $edit;
 933                         }
 934
 935                         if ($edit->orig)
 936                                 $xi += sizeof($edit->orig);
 937                         if ($edit->closing)
 938                                 $yi += sizeof($edit->closing);
 939                 }
 940
 941                 if (is_array($block))
 942                         $this->_block($x0, $xi - $x0,
 943                                                   $y0, $yi - $y0,
 944                                                   $block);
 945
 946                 return $this->_end_diff();
 947         }
 948
 949         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
 950                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
 951                 foreach ($edits as $edit) {
 952                         if ($edit->type == 'copy')
 953                                 $this->_context($edit->orig);
 954                         elseif ($edit->type == 'add')
 955                                 $this->_added($edit->closing);
 956                         elseif ($edit->type == 'delete')
 957                                 $this->_deleted($edit->orig);
 958                         elseif ($edit->type == 'change')
 959                                 $this->_changed($edit->orig, $edit->closing);
 960                         else
 961                                 trigger_error("Unknown edit type", E_USER_ERROR);
 962                 }
 963                 $this->_end_block();
 964         }
 965
 966         function _start_diff() {
 967                 ob_start();
 968         }
 969
 970         function _end_diff() {
 971                 $val = ob_get_contents();
 972                 ob_end_clean();
 973                 return $val;
 974         }
 975
 976         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
 977                 if ($xlen > 1)
 978                         $xbeg .= "," . ($xbeg + $xlen - 1);
 979                 if ($ylen > 1)
 980                         $ybeg .= "," . ($ybeg + $ylen - 1);
 981
 982                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
 983         }
 984
 985         function _start_block($header) {
 986                 echo $header;
 987         }
 988
 989         function _end_block() {
 990         }
 991
 992         function _lines($lines, $prefix = ' ') {
 993                 foreach ($lines as $line)
 994                         echo "$prefix $line\n";
 995         }
 996
 997         function _context($lines) {
 998                 $this->_lines($lines);
 999         }
1000
1001         function _added($lines) {
1002                 $this->_lines($lines, ">");
1003         }
1004         function _deleted($lines) {
1005                 $this->_lines($lines, "<");
1006         }
1007
1008         function _changed($orig, $closing) {
1009                 $this->_deleted($orig);
1010                 echo "---\n";
1011                 $this->_added($closing);
1012         }
1013 }
1014
1015
1016 /**
1017  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1018  *
1019  */
1020
1021 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1022
1023 class _HWLDF_WordAccumulator {
1024         function _HWLDF_WordAccumulator () {
1025                 $this->_lines = array();
1026                 $this->_line = '';
1027                 $this->_group = '';
1028                 $this->_tag = '';
1029         }
1030
1031         function _flushGroup ($new_tag) {
1032                 if ($this->_group !== '') {
1033           if ($this->_tag == 'mark')
1034                         $this->_line .= '<span class="diffchange">'.$this->_group.'</span>';
1035           else
1036                 $this->_line .= $this->_group;
1037         }
1038                 $this->_group = '';
1039                 $this->_tag = $new_tag;
1040         }
1041
1042         function _flushLine ($new_tag) {
1043                 $this->_flushGroup($new_tag);
1044                 if ($this->_line != '')
1045                         $this->_lines[] = $this->_line;
1046                 $this->_line = '';
1047         }
1048
1049         function addWords ($words, $tag = '') {
1050                 if ($tag != $this->_tag)
1051                         $this->_flushGroup($tag);
1052
1053                 foreach ($words as $word) {
1054                         // new-line should only come as first char of word.
1055                         if ($word == '')
1056                                 continue;
1057                         if ($word[0] == "\n") {
1058                                 $this->_group .= NBSP;
1059                                 $this->_flushLine($tag);
1060                                 $word = substr($word, 1);
1061                         }
1062                         assert(!strstr($word, "\n"));
1063                         $this->_group .= $word;
1064                 }
1065         }
1066
1067         function getLines() {
1068                 $this->_flushLine('~done');
1069                 return $this->_lines;
1070         }
1071 }
1072
1073 class WordLevelDiff extends MappedDiff
1074 {
1075         function WordLevelDiff ($orig_lines, $closing_lines) {
1076                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1077                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1078
1079
1080                 $this->MappedDiff($orig_words, $closing_words,
1081                                                   $orig_stripped, $closing_stripped);
1082         }
1083
1084         function _split($lines) {
1085                 // FIXME: fix POSIX char class.
1086 #                if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1087                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1088                                                         implode("\n", $lines),
1089                                                         $m)) {
1090                         return array(array(''), array(''));
1091                 }
1092                 return array($m[0], $m[1]);
1093         }
1094
1095         function orig () {
1096                 $orig = new _HWLDF_WordAccumulator;
1097
1098                 foreach ($this->edits as $edit) {
1099                         if ($edit->type == 'copy')
1100                                 $orig->addWords($edit->orig);
1101                         elseif ($edit->orig)
1102                                 $orig->addWords($edit->orig, 'mark');
1103                 }
1104                 return $orig->getLines();
1105         }
1106
1107         function closing () {
1108                 $closing = new _HWLDF_WordAccumulator;
1109
1110                 foreach ($this->edits as $edit) {
1111                         if ($edit->type == 'copy')
1112                                 $closing->addWords($edit->closing);
1113                         elseif ($edit->closing)
1114                                 $closing->addWords($edit->closing, 'mark');
1115                 }
1116                 return $closing->getLines();
1117         }
1118 }
1119
1120 /**
1121  *      Wikipedia Table style diff formatter.
1122  *
1123  */
1124 class TableDiffFormatter extends DiffFormatter
1125 {
1126         function TableDiffFormatter() {
1127                 $this->leading_context_lines = 2;
1128                 $this->trailing_context_lines = 2;
1129         }
1130
1131         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1132                 $l1 = wfMsg( "lineno", $xbeg );
1133                 $l2 = wfMsg( "lineno", $ybeg );
1134
1135                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1136                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1137                 return $r;
1138         }
1139
1140         function _start_block( $header ) {
1141                 global $wgOut;
1142                 $wgOut->addHTML( $header );
1143         }
1144
1145         function _end_block() {
1146         }
1147
1148         function _lines( $lines, $prefix=' ', $color="white" ) {
1149         }
1150
1151         function addedLine( $line ) {
1152                 return '<td>+</td><td class="diff-addedline">' .
1153                   $line.'</td>';
1154         }
1155
1156         function deletedLine( $line ) {
1157                 return '<td>-</td><td class="diff-deletedline">' .
1158                   $line.'</td>';
1159         }
1160
1161         function emptyLine() {
1162                 return '<td colspan="2">&nbsp;</td>';
1163         }
1164
1165         function contextLine( $line ) {
1166                 return '<td> </td><td class="diff-context">'.$line.'</td>';
1167         }
1168
1169         function _added($lines) {
1170                 global $wgOut;
1171                 foreach ($lines as $line) {
1172                         $wgOut->addHTML( '<tr>' . $this->emptyLine() .
1173                           $this->addedLine( $line ) . "</tr>\n" );
1174                 }
1175         }
1176
1177         function _deleted($lines) {
1178                 global $wgOut;
1179                 foreach ($lines as $line) {
1180                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1181                           $this->emptyLine() . "</tr>\n" );
1182                 }
1183         }
1184
1185         function _context( $lines ) {
1186                 global $wgOut;
1187                 foreach ($lines as $line) {
1188                         $wgOut->addHTML( '<tr>' . $this->contextLine( $line ) .
1189                           $this->contextLine( $line ) . "</tr>\n" );
1190                 }
1191         }
1192
1193         function _changed( $orig, $closing ) {
1194                 global $wgOut;
1195                 $diff = new WordLevelDiff( $orig, $closing );
1196                 $del = $diff->orig();
1197                 $add = $diff->closing();
1198
1199                 while ( $line = array_shift( $del ) ) {
1200                         $aline = array_shift( $add );
1201                         $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1202                           $this->addedLine( $aline ) . "</tr>\n" );
1203                 }
1204                 $this->_added( $add ); # If any leftovers
1205         }
1206 }
1207
1208 ?>