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