includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @package MediaWiki
   5  * @subpackage DifferenceEngine
   6  */
   7
   8 /**
   9  * @todo document
  10  * @access public
  11  */
  12 class DifferenceEngine {
  13         /* private */ var $mOldid, $mNewid;
  14         /* private */ var $mOldtitle, $mNewtitle, $mPagetitle;
  15         /* private */ var $mOldtext, $mNewtext;
  16         /* private */ var $mOldUser, $mNewUser;
  17         /* private */ var $mOldComment, $mNewComment;
  18         /* private */ var $mOldPage, $mNewPage;
  19         /* private */ var $mRcidMarkPatrolled;
  20
  21         function DifferenceEngine( $old, $new, $rcid = 0 )
  22         {
  23                 global $wgTitle;
  24                 if ( 'prev' == $new ) {
  25                         # Show diff between revision $old and the previous one.
  26                         # Get previous one from DB.
  27                         #
  28                         $this->mNewid = intval($old);
  29
  30                         $this->mOldid = $wgTitle->getPreviousRevisionID( $this->mNewid );
  31
  32                 } elseif ( 'next' == $new ) {
  33
  34                         # Show diff between revision $old and the previous one.
  35                         # Get previous one from DB.
  36                         #
  37                         $this->mOldid = intval($old);
  38                         $this->mNewid = $wgTitle->getNextRevisionID( $this->mOldid );
  39                         if ( false === $this->mNewid ) {
  40                                 # if no result, NewId points to the newest old revision. The only newer
  41                                 # revision is cur, which is "0".
  42                                 $this->mNewid = 0;
  43                         }
  44
  45                 } else {
  46
  47                         $this->mOldid = intval($old);
  48                         $this->mNewid = intval($new);
  49                 }
  50                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  51         }
  52
  53         function showDiffPage()
  54         {
  55                 global $wgUser, $wgTitle, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol, $wgUseRCPatrol;
  56                 $fname = 'DifferenceEngine::showDiffPage';
  57                 wfProfileIn( $fname );
  58
  59                 # mOldid is false if the difference engine is called with a "vague" query for
  60                 # a diff between a version V and its previous version V' AND the version V
  61                 # is the first version of that article. In that case, V' does not exist.
  62                 if ( $this->mOldid === false ) {
  63                         $this->showFirstRevision();
  64                         wfProfileOut( $fname );
  65                         return;
  66                 }
  67
  68                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  69                   "{$this->mNewid})";
  70                 $mtext = wfMsg( 'missingarticle', $t );
  71
  72                 $wgOut->setArticleFlag( false );
  73                 if ( ! $this->loadText() ) {
  74                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
  75                         $wgOut->addHTML( $mtext );
  76                         wfProfileOut( $fname );
  77                         return;
  78                 }
  79                 $wgOut->suppressQuickbar();
  80
  81                 $oldTitle = $this->mOldPage->getPrefixedText();
  82                 $newTitle = $this->mNewPage->getPrefixedText();
  83                 if( $oldTitle == $newTitle ) {
  84                         $wgOut->setPageTitle( $newTitle );
  85                 } else {
  86                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
  87                 }
  88                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
  89                 $wgOut->setRobotpolicy( 'noindex,follow' );
  90
  91                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  92                         $wgOut->loginToUse();
  93                         $wgOut->output();
  94                         wfProfileOut( $fname );
  95                         exit;
  96                 }
  97
  98                 $sk = $wgUser->getSkin();
  99                 $talk = $wgContLang->getNsText( NS_TALK );
 100                 $contribs = wfMsg( 'contribslink' );
 101
 102                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 103                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 104
 105                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 106                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 107                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 108                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 109                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 110                         'target=' . urlencode($this->mOldUser) );
 111                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 112                         'target=' . urlencode($this->mNewUser) );
 113                 if ( !$this->mNewid && $wgUser->isAllowed('rollback') ) {
 114                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
 115                                 'action=rollback&from=' . urlencode($this->mNewUser) ) . ']</strong>';
 116                 } else {
 117                         $rollback = '';
 118                 }
 119                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->getID() != 0 &&
 120                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 121                 {
 122                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 123                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 124                 } else {
 125                         $patrol = '';
 126                 }
 127
 128                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'previousdiff' ), 'diff=prev&oldid='.$this->mOldid );
 129                 if ( $this->mNewid == 0 ) {
 130                         $nextlink = '';
 131                 } else {
 132                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 133                 }
 134
 135                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 136                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 137                         '<br />' . $prevlink;
 138                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 139                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 140                         '<br />' . $nextlink . $patrol;
 141
 142                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 143                   $oldHeader, $newHeader );
 144                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 145                 $wgOut->addWikiText( $this->mNewtext );
 146
 147                 wfProfileOut( $fname );
 148         }
 149
 150         # Show the first revision of an article. Uses normal diff headers in contrast to normal
 151         # "old revision" display style.
 152         #
 153         function showFirstRevision()
 154         {
 155                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 156
 157                 $fname = 'DifferenceEngine::showFirstRevision';
 158                 wfProfileIn( $fname );
 159
 160
 161                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 162
 163                 # Get article text from the DB
 164                 #
 165                 if ( ! $this->loadText() ) {
 166                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 167                           "{$this->mNewid})";
 168                         $mtext = wfMsg( 'missingarticle', $t );
 169                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 170                         $wgOut->addHTML( $mtext );
 171                         wfProfileOut( $fname );
 172                         return;
 173                 }
 174
 175                 # Check if user is allowed to look at this page. If not, bail out.
 176                 #
 177                 if ( !( $this->mOldPage->userCanRead() ) ) {
 178                         $wgOut->loginToUse();
 179                         $wgOut->output();
 180                         wfProfileOut( $fname );
 181                         exit;
 182                 }
 183
 184                 # Prepare the header box
 185                 #
 186                 $sk = $wgUser->getSkin();
 187
 188                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 189                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 190                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 191                         'target=' . urlencode($this->mOldUser) );
 192                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 193                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 194                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 195                         '<br />' . $nextlink. "</div>\n";
 196
 197                 $wgOut->addHTML( $header );
 198
 199                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 200                 $wgOut->setRobotpolicy( 'noindex,follow' );
 201
 202
 203                 # Show current revision
 204                 #
 205                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 206                 $wgOut->addWikiText( $this->mNewtext );
 207
 208                 wfProfileOut( $fname );
 209         }
 210
 211         function showDiff( $otext, $ntext, $otitle, $ntitle )
 212         {
 213                 global $wgOut;
 214                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 215         }
 216
 217         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 218                 global $wgUseExternalDiffEngine;
 219                 $out = "
 220                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 221                         <tr>
 222                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 223                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 224                         </tr>
 225                 ";
 226
 227                 if ( $wgUseExternalDiffEngine ) {
 228                         # For historical reasons, external diff engine expects
 229                         # input text to be HTML-escaped already
 230                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 231                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 232                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 233                                 dl('php_wikidiff.so');
 234                         }
 235                         $out .= wikidiff_do_diff( $otext, $ntext, 2 );
 236                 } else {
 237                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 238                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 239                         $diffs =& new Diff( $ota, $nta );
 240                         $formatter =& new TableDiffFormatter();
 241                         $out .= $formatter->format( $diffs );
 242                 }
 243                 $out .= "</table>\n";
 244                 return $out;
 245         }
 246
 247         # Load the text of the articles to compare.  If newid is 0, then compare
 248         # the old article in oldid to the current article; if oldid is 0, then
 249         # compare the current article to the immediately previous one (ignoring
 250         # the value of newid).
 251         #
 252         function loadText()
 253         {
 254                 global $wgTitle, $wgOut, $wgLang;
 255                 $fname = 'DifferenceEngine::loadText';
 256
 257                 $dbr =& wfGetDB( DB_SLAVE );
 258                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 259                         $wgOut->setArticleFlag( true );
 260                         $newLink = $wgTitle->escapeLocalUrl();
 261                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 262                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 263                         $id = $wgTitle->getArticleID();
 264
 265                         $s = $dbr->selectRow( 'cur', array( 'cur_text', 'cur_user_text', 'cur_comment' ),
 266                                 array( 'cur_id' => $id ), $fname );
 267                         if ( $s === false ) {
 268                                 wfDebug( "Unable to load cur_id $id\n" );
 269                                 return false;
 270                         }
 271
 272                         $this->mNewPage = &$wgTitle;
 273                         $this->mNewtext = $s->cur_text;
 274                         $this->mNewUser = $s->cur_user_text;
 275                         $this->mNewComment = $s->cur_comment;
 276                 } else {
 277                         $s = $dbr->selectRow( 'old', array( 'old_namespace','old_title','old_timestamp', 'old_text',
 278                                 'old_flags','old_user_text','old_comment' ), array( 'old_id' => $this->mNewid ), $fname );
 279
 280                         if ( $s === false ) {
 281                                 wfDebug( "Unable to load old_id {$this->mNewid}\n" );
 282                                 return false;
 283                         }
 284
 285                         $this->mNewtext = Article::getRevisionText( $s );
 286
 287                         $t = $wgLang->timeanddate( $s->old_timestamp, true );
 288                         $this->mNewPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 289                         $newLink = $wgTitle->escapeLocalUrl ('oldid=' . $this->mNewid);
 290                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 291                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 292                         $this->mNewUser = $s->old_user_text;
 293                         $this->mNewComment = $s->old_comment;
 294                 }
 295                 if ( 0 == $this->mOldid ) {
 296                         $s = $dbr->selectRow( 'old',
 297                                 array( 'old_namespace','old_title','old_timestamp','old_text', 'old_flags','old_user_text','old_comment' ),
 298                                 array( /* WHERE */
 299                                         'old_namespace' => $this->mNewPage->getNamespace(),
 300                                         'old_title' => $this->mNewPage->getDBkey()
 301                                 ), $fname, array( 'ORDER BY' => 'inverse_timestamp', 'USE INDEX' => 'name_title_timestamp' )
 302                         );
 303                         if ( $s === false ) {
 304                                 wfDebug( 'Unable to load ' . $this->mNewPage->getPrefixedDBkey() . " from old\n" );
 305                                 return false;
 306                         }
 307                 } else {
 308                         $s = $dbr->selectRow( 'old',
 309                                 array( 'old_namespace','old_title','old_timestamp','old_text','old_flags','old_user_text','old_comment'),
 310                                 array( 'old_id' => $this->mOldid ),
 311                                 $fname
 312                         );
 313                         if ( $s === false ) {
 314                                 wfDebug( "Unable to load old_id {$this->mOldid}\n" );
 315                                 return false;
 316                         }
 317                 }
 318                 $this->mOldPage = Title::MakeTitle( $s->old_namespace, $s->old_title );
 319                 $this->mOldtext = Article::getRevisionText( $s );
 320
 321                 $t = $wgLang->timeanddate( $s->old_timestamp, true );
 322                 $oldLink = $this->mOldPage->escapeLocalUrl ('oldid=' . $this->mOldid);
 323                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 324                 $this->mOldUser = $s->old_user_text;
 325                 $this->mOldComment = $s->old_comment;
 326
 327                 return true;
 328         }
 329 }
 330
 331 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 332 //
 333 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 334 // You may copy this code freely under the conditions of the GPL.
 335 //
 336
 337 define('USE_ASSERTS', function_exists('assert'));
 338
 339 /**
 340  * @todo document
 341  * @access private
 342  */
 343 class _DiffOp {
 344         var $type;
 345         var $orig;
 346         var $closing;
 347
 348         function reverse() {
 349                 trigger_error('pure virtual', E_USER_ERROR);
 350         }
 351
 352         function norig() {
 353                 return $this->orig ? sizeof($this->orig) : 0;
 354         }
 355
 356         function nclosing() {
 357                 return $this->closing ? sizeof($this->closing) : 0;
 358         }
 359 }
 360
 361 /**
 362  * @todo document
 363  * @access private
 364  */
 365 class _DiffOp_Copy extends _DiffOp {
 366         var $type = 'copy';
 367
 368         function _DiffOp_Copy ($orig, $closing = false) {
 369                 if (!is_array($closing))
 370                         $closing = $orig;
 371                 $this->orig = $orig;
 372                 $this->closing = $closing;
 373         }
 374
 375         function reverse() {
 376                 return new _DiffOp_Copy($this->closing, $this->orig);
 377         }
 378 }
 379
 380 /**
 381  * @todo document
 382  * @access private
 383  */
 384 class _DiffOp_Delete extends _DiffOp {
 385         var $type = 'delete';
 386
 387         function _DiffOp_Delete ($lines) {
 388                 $this->orig = $lines;
 389                 $this->closing = false;
 390         }
 391
 392         function reverse() {
 393                 return new _DiffOp_Add($this->orig);
 394         }
 395 }
 396
 397 /**
 398  * @todo document
 399  * @access private
 400  */
 401 class _DiffOp_Add extends _DiffOp {
 402         var $type = 'add';
 403
 404         function _DiffOp_Add ($lines) {
 405                 $this->closing = $lines;
 406                 $this->orig = false;
 407         }
 408
 409         function reverse() {
 410                 return new _DiffOp_Delete($this->closing);
 411         }
 412 }
 413
 414 /**
 415  * @todo document
 416  * @access private
 417  */
 418 class _DiffOp_Change extends _DiffOp {
 419         var $type = 'change';
 420
 421         function _DiffOp_Change ($orig, $closing) {
 422                 $this->orig = $orig;
 423                 $this->closing = $closing;
 424         }
 425
 426         function reverse() {
 427                 return new _DiffOp_Change($this->closing, $this->orig);
 428         }
 429 }
 430
 431
 432 /**
 433  * Class used internally by Diff to actually compute the diffs.
 434  *
 435  * The algorithm used here is mostly lifted from the perl module
 436  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 437  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 438  *
 439  * More ideas are taken from:
 440  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 441  *
 442  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 443  * diffutils-2.7, which can be found at:
 444  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 445  *
 446  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 447  * are my own.
 448  *
 449  * @author Geoffrey T. Dairiki
 450  * @access private
 451  */
 452 class _DiffEngine
 453 {
 454         function diff ($from_lines, $to_lines) {
 455                 $fname = '_DiffEngine::diff';
 456                 wfProfileIn( $fname );
 457
 458                 $n_from = sizeof($from_lines);
 459                 $n_to = sizeof($to_lines);
 460
 461                 $this->xchanged = $this->ychanged = array();
 462                 $this->xv = $this->yv = array();
 463                 $this->xind = $this->yind = array();
 464                 unset($this->seq);
 465                 unset($this->in_seq);
 466                 unset($this->lcs);
 467
 468                 // Skip leading common lines.
 469                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 470                         if ($from_lines[$skip] != $to_lines[$skip])
 471                                 break;
 472                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 473                 }
 474                 // Skip trailing common lines.
 475                 $xi = $n_from; $yi = $n_to;
 476                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 477                         if ($from_lines[$xi] != $to_lines[$yi])
 478                                 break;
 479                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 480                 }
 481
 482                 // Ignore lines which do not exist in both files.
 483                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 484                         $xhash[$from_lines[$xi]] = 1;
 485                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 486                         $line = $to_lines[$yi];
 487                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 488                                 continue;
 489                         $yhash[$line] = 1;
 490                         $this->yv[] = $line;
 491                         $this->yind[] = $yi;
 492                 }
 493                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 494                         $line = $from_lines[$xi];
 495                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 496                                 continue;
 497                         $this->xv[] = $line;
 498                         $this->xind[] = $xi;
 499                 }
 500
 501                 // Find the LCS.
 502                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 503
 504                 // Merge edits when possible
 505                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 506                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 507
 508                 // Compute the edit operations.
 509                 $edits = array();
 510                 $xi = $yi = 0;
 511                 while ($xi < $n_from || $yi < $n_to) {
 512                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 513                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 514
 515                         // Skip matching "snake".
 516                         $copy = array();
 517                         while ( $xi < $n_from && $yi < $n_to
 518                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 519                                 $copy[] = $from_lines[$xi++];
 520                                 ++$yi;
 521                         }
 522                         if ($copy)
 523                                 $edits[] = new _DiffOp_Copy($copy);
 524
 525                         // Find deletes & adds.
 526                         $delete = array();
 527                         while ($xi < $n_from && $this->xchanged[$xi])
 528                                 $delete[] = $from_lines[$xi++];
 529
 530                         $add = array();
 531                         while ($yi < $n_to && $this->ychanged[$yi])
 532                                 $add[] = $to_lines[$yi++];
 533
 534                         if ($delete && $add)
 535                                 $edits[] = new _DiffOp_Change($delete, $add);
 536                         elseif ($delete)
 537                                 $edits[] = new _DiffOp_Delete($delete);
 538                         elseif ($add)
 539                                 $edits[] = new _DiffOp_Add($add);
 540                 }
 541                 wfProfileOut( $fname );
 542                 return $edits;
 543         }
 544
 545
 546         /* Divide the Largest Common Subsequence (LCS) of the sequences
 547          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 548          * sized segments.
 549          *
 550          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 551          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 552          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 553          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 554          * that (X0, Y0) == (XOFF, YOFF) and
 555          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 556          *
 557          * This function assumes that the first lines of the specified portions
 558          * of the two files do not match, and likewise that the last lines do not
 559          * match.  The caller must trim matching lines from the beginning and end
 560          * of the portions it is going to specify.
 561          */
 562         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 563                 $fname = '_DiffEngine::_diag';
 564                 wfProfileIn( $fname );
 565                 $flip = false;
 566
 567                 if ($xlim - $xoff > $ylim - $yoff) {
 568                         // Things seems faster (I'm not sure I understand why)
 569                                 // when the shortest sequence in X.
 570                                 $flip = true;
 571                         list ($xoff, $xlim, $yoff, $ylim)
 572                         = array( $yoff, $ylim, $xoff, $xlim);
 573                 }
 574
 575                 if ($flip)
 576                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 577                                 $ymatches[$this->xv[$i]][] = $i;
 578                 else
 579                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 580                                 $ymatches[$this->yv[$i]][] = $i;
 581
 582                 $this->lcs = 0;
 583                 $this->seq[0]= $yoff - 1;
 584                 $this->in_seq = array();
 585                 $ymids[0] = array();
 586
 587                 $numer = $xlim - $xoff + $nchunks - 1;
 588                 $x = $xoff;
 589                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 590                         if ($chunk > 0)
 591                                 for ($i = 0; $i <= $this->lcs; $i++)
 592                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 593
 594                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 595                         for ( ; $x < $x1; $x++) {
 596                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 597                                         if (empty($ymatches[$line]))
 598                                                 continue;
 599                                 $matches = $ymatches[$line];
 600                                 reset($matches);
 601                                 while (list ($junk, $y) = each($matches))
 602                                         if (empty($this->in_seq[$y])) {
 603                                                 $k = $this->_lcs_pos($y);
 604                                                 USE_ASSERTS && assert($k > 0);
 605                                                 $ymids[$k] = $ymids[$k-1];
 606                                                 break;
 607                                         }
 608                                 while (list ($junk, $y) = each($matches)) {
 609                                         if ($y > $this->seq[$k-1]) {
 610                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 611                                                 // Optimization: this is a common case:
 612                                                 //      next match is just replacing previous match.
 613                                                 $this->in_seq[$this->seq[$k]] = false;
 614                                                 $this->seq[$k] = $y;
 615                                                 $this->in_seq[$y] = 1;
 616                                         } else if (empty($this->in_seq[$y])) {
 617                                                 $k = $this->_lcs_pos($y);
 618                                                 USE_ASSERTS && assert($k > 0);
 619                                                 $ymids[$k] = $ymids[$k-1];
 620                                         }
 621                                 }
 622                         }
 623                 }
 624
 625                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 626                 $ymid = $ymids[$this->lcs];
 627                 for ($n = 0; $n < $nchunks - 1; $n++) {
 628                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 629                         $y1 = $ymid[$n] + 1;
 630                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 631                 }
 632                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 633
 634                 wfProfileOut( $fname );
 635                 return array($this->lcs, $seps);
 636         }
 637
 638         function _lcs_pos ($ypos) {
 639                 $fname = '_DiffEngine::_lcs_pos';
 640                 wfProfileIn( $fname );
 641
 642                 $end = $this->lcs;
 643                 if ($end == 0 || $ypos > $this->seq[$end]) {
 644                         $this->seq[++$this->lcs] = $ypos;
 645                         $this->in_seq[$ypos] = 1;
 646                         wfProfileOut( $fname );
 647                         return $this->lcs;
 648                 }
 649
 650                 $beg = 1;
 651                 while ($beg < $end) {
 652                         $mid = (int)(($beg + $end) / 2);
 653                         if ( $ypos > $this->seq[$mid] )
 654                                 $beg = $mid + 1;
 655                         else
 656                                 $end = $mid;
 657                 }
 658
 659                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 660
 661                 $this->in_seq[$this->seq[$end]] = false;
 662                 $this->seq[$end] = $ypos;
 663                 $this->in_seq[$ypos] = 1;
 664                 wfProfileOut( $fname );
 665                 return $end;
 666         }
 667
 668         /* Find LCS of two sequences.
 669          *
 670          * The results are recorded in the vectors $this->{x,y}changed[], by
 671          * storing a 1 in the element for each line that is an insertion
 672          * or deletion (ie. is not in the LCS).
 673          *
 674          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 675          *
 676          * Note that XLIM, YLIM are exclusive bounds.
 677          * All line numbers are origin-0 and discarded lines are not counted.
 678          */
 679         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 680                 $fname = '_DiffEngine::_compareseq';
 681                 wfProfileIn( $fname );
 682
 683                 // Slide down the bottom initial diagonal.
 684                 while ($xoff < $xlim && $yoff < $ylim
 685                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 686                         ++$xoff;
 687                         ++$yoff;
 688                 }
 689
 690                 // Slide up the top initial diagonal.
 691                 while ($xlim > $xoff && $ylim > $yoff
 692                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 693                         --$xlim;
 694                         --$ylim;
 695                 }
 696
 697                 if ($xoff == $xlim || $yoff == $ylim)
 698                         $lcs = 0;
 699                 else {
 700                         // This is ad hoc but seems to work well.
 701                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 702                         //$nchunks = max(2,min(8,(int)$nchunks));
 703                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 704                         list ($lcs, $seps)
 705                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 706                 }
 707
 708                 if ($lcs == 0) {
 709                         // X and Y sequences have no common subsequence:
 710                         // mark all changed.
 711                         while ($yoff < $ylim)
 712                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 713                         while ($xoff < $xlim)
 714                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 715                 } else {
 716                         // Use the partitions to split this problem into subproblems.
 717                         reset($seps);
 718                         $pt1 = $seps[0];
 719                         while ($pt2 = next($seps)) {
 720                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 721                                 $pt1 = $pt2;
 722                         }
 723                 }
 724                 wfProfileOut( $fname );
 725         }
 726
 727         /* Adjust inserts/deletes of identical lines to join changes
 728          * as much as possible.
 729          *
 730          * We do something when a run of changed lines include a
 731          * line at one end and has an excluded, identical line at the other.
 732          * We are free to choose which identical line is included.
 733          * `compareseq' usually chooses the one at the beginning,
 734          * but usually it is cleaner to consider the following identical line
 735          * to be the "change".
 736          *
 737          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 738          */
 739         function _shift_boundaries ($lines, &$changed, $other_changed) {
 740                 $fname = '_DiffEngine::_shift_boundaries';
 741                 wfProfileIn( $fname );
 742                 $i = 0;
 743                 $j = 0;
 744
 745                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 746                 $len = sizeof($lines);
 747                 $other_len = sizeof($other_changed);
 748
 749                 while (1) {
 750                         /*
 751                          * Scan forwards to find beginning of another run of changes.
 752                          * Also keep track of the corresponding point in the other file.
 753                          *
 754                          * Throughout this code, $i and $j are adjusted together so that
 755                          * the first $i elements of $changed and the first $j elements
 756                          * of $other_changed both contain the same number of zeros
 757                          * (unchanged lines).
 758                          * Furthermore, $j is always kept so that $j == $other_len or
 759                          * $other_changed[$j] == false.
 760                          */
 761                         while ($j < $other_len && $other_changed[$j])
 762                                 $j++;
 763
 764                         while ($i < $len && ! $changed[$i]) {
 765                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 766                                 $i++; $j++;
 767                                 while ($j < $other_len && $other_changed[$j])
 768                                         $j++;
 769                         }
 770
 771                         if ($i == $len)
 772                                 break;
 773
 774                         $start = $i;
 775
 776                         // Find the end of this run of changes.
 777                         while (++$i < $len && $changed[$i])
 778                                 continue;
 779
 780                         do {
 781                                 /*
 782                                  * Record the length of this run of changes, so that
 783                                  * we can later determine whether the run has grown.
 784                                  */
 785                                 $runlength = $i - $start;
 786
 787                                 /*
 788                                  * Move the changed region back, so long as the
 789                                  * previous unchanged line matches the last changed one.
 790                                  * This merges with previous changed regions.
 791                                  */
 792                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 793                                         $changed[--$start] = 1;
 794                                         $changed[--$i] = false;
 795                                         while ($start > 0 && $changed[$start - 1])
 796                                                 $start--;
 797                                         USE_ASSERTS && assert('$j > 0');
 798                                         while ($other_changed[--$j])
 799                                                 continue;
 800                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 801                                 }
 802
 803                                 /*
 804                                  * Set CORRESPONDING to the end of the changed run, at the last
 805                                  * point where it corresponds to a changed run in the other file.
 806                                  * CORRESPONDING == LEN means no such point has been found.
 807                                  */
 808                                 $corresponding = $j < $other_len ? $i : $len;
 809
 810                                 /*
 811                                  * Move the changed region forward, so long as the
 812                                  * first changed line matches the following unchanged one.
 813                                  * This merges with following changed regions.
 814                                  * Do this second, so that if there are no merges,
 815                                  * the changed region is moved forward as far as possible.
 816                                  */
 817                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 818                                         $changed[$start++] = false;
 819                                         $changed[$i++] = 1;
 820                                         while ($i < $len && $changed[$i])
 821                                                 $i++;
 822
 823                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 824                                         $j++;
 825                                         if ($j < $other_len && $other_changed[$j]) {
 826                                                 $corresponding = $i;
 827                                                 while ($j < $other_len && $other_changed[$j])
 828                                                         $j++;
 829                                         }
 830                                 }
 831                         } while ($runlength != $i - $start);
 832
 833                         /*
 834                          * If possible, move the fully-merged run of changes
 835                          * back to a corresponding run in the other file.
 836                          */
 837                         while ($corresponding < $i) {
 838                                 $changed[--$start] = 1;
 839                                 $changed[--$i] = 0;
 840                                 USE_ASSERTS && assert('$j > 0');
 841                                 while ($other_changed[--$j])
 842                                         continue;
 843                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 844                         }
 845                 }
 846                 wfProfileOut( $fname );
 847         }
 848 }
 849
 850 /**
 851  * Class representing a 'diff' between two sequences of strings.
 852  * @todo document
 853  * @access private
 854  */
 855 class Diff
 856 {
 857         var $edits;
 858
 859         /**
 860          * Constructor.
 861          * Computes diff between sequences of strings.
 862          *
 863          * @param $from_lines array An array of strings.
 864          *                (Typically these are lines from a file.)
 865          * @param $to_lines array An array of strings.
 866          */
 867         function Diff($from_lines, $to_lines) {
 868                 $eng = new _DiffEngine;
 869                 $this->edits = $eng->diff($from_lines, $to_lines);
 870                 //$this->_check($from_lines, $to_lines);
 871         }
 872
 873         /**
 874          * Compute reversed Diff.
 875          *
 876          * SYNOPSIS:
 877          *
 878          *      $diff = new Diff($lines1, $lines2);
 879          *      $rev = $diff->reverse();
 880          * @return object A Diff object representing the inverse of the
 881          *                                original diff.
 882          */
 883         function reverse () {
 884                 $rev = $this;
 885                 $rev->edits = array();
 886                 foreach ($this->edits as $edit) {
 887                         $rev->edits[] = $edit->reverse();
 888                 }
 889                 return $rev;
 890         }
 891
 892         /**
 893          * Check for empty diff.
 894          *
 895          * @return bool True iff two sequences were identical.
 896          */
 897         function isEmpty () {
 898                 foreach ($this->edits as $edit) {
 899                         if ($edit->type != 'copy')
 900                                 return false;
 901                 }
 902                 return true;
 903         }
 904
 905         /**
 906          * Compute the length of the Longest Common Subsequence (LCS).
 907          *
 908          * This is mostly for diagnostic purposed.
 909          *
 910          * @return int The length of the LCS.
 911          */
 912         function lcs () {
 913                 $lcs = 0;
 914                 foreach ($this->edits as $edit) {
 915                         if ($edit->type == 'copy')
 916                                 $lcs += sizeof($edit->orig);
 917                 }
 918                 return $lcs;
 919         }
 920
 921         /**
 922          * Get the original set of lines.
 923          *
 924          * This reconstructs the $from_lines parameter passed to the
 925          * constructor.
 926          *
 927          * @return array The original sequence of strings.
 928          */
 929         function orig() {
 930                 $lines = array();
 931
 932                 foreach ($this->edits as $edit) {
 933                         if ($edit->orig)
 934                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 935                 }
 936                 return $lines;
 937         }
 938
 939         /**
 940          * Get the closing set of lines.
 941          *
 942          * This reconstructs the $to_lines parameter passed to the
 943          * constructor.
 944          *
 945          * @return array The sequence of strings.
 946          */
 947         function closing() {
 948                 $lines = array();
 949
 950                 foreach ($this->edits as $edit) {
 951                         if ($edit->closing)
 952                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 953                 }
 954                 return $lines;
 955         }
 956
 957         /**
 958          * Check a Diff for validity.
 959          *
 960          * This is here only for debugging purposes.
 961          */
 962         function _check ($from_lines, $to_lines) {
 963                 $fname = 'Diff::_check';
 964                 wfProfileIn( $fname );
 965                 if (serialize($from_lines) != serialize($this->orig()))
 966                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 967                 if (serialize($to_lines) != serialize($this->closing()))
 968                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 969
 970                 $rev = $this->reverse();
 971                 if (serialize($to_lines) != serialize($rev->orig()))
 972                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 973                 if (serialize($from_lines) != serialize($rev->closing()))
 974                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 975
 976
 977                 $prevtype = 'none';
 978                 foreach ($this->edits as $edit) {
 979                         if ( $prevtype == $edit->type )
 980                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 981                         $prevtype = $edit->type;
 982                 }
 983
 984                 $lcs = $this->lcs();
 985                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 986                 wfProfileOut( $fname );
 987         }
 988 }
 989
 990 /**
 991  * FIXME: bad name.
 992  * @todo document
 993  * @access private
 994  */
 995 class MappedDiff extends Diff
 996 {
 997         /**
 998          * Constructor.
 999          *
1000          * Computes diff between sequences of strings.
1001          *
1002          * This can be used to compute things like
1003          * case-insensitve diffs, or diffs which ignore
1004          * changes in white-space.
1005          *
1006          * @param $from_lines array An array of strings.
1007          *      (Typically these are lines from a file.)
1008          *
1009          * @param $to_lines array An array of strings.
1010          *
1011          * @param $mapped_from_lines array This array should
1012          *      have the same size number of elements as $from_lines.
1013          *      The elements in $mapped_from_lines and
1014          *      $mapped_to_lines are what is actually compared
1015          *      when computing the diff.
1016          *
1017          * @param $mapped_to_lines array This array should
1018          *      have the same number of elements as $to_lines.
1019          */
1020         function MappedDiff($from_lines, $to_lines,
1021                                                 $mapped_from_lines, $mapped_to_lines) {
1022                 $fname = 'MappedDiff::MappedDiff';
1023                 wfProfileIn( $fname );
1024
1025                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1026                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1027
1028                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1029
1030                 $xi = $yi = 0;
1031                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1032                         $orig = &$this->edits[$i]->orig;
1033                         if (is_array($orig)) {
1034                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1035                                 $xi += sizeof($orig);
1036                         }
1037
1038                         $closing = &$this->edits[$i]->closing;
1039                         if (is_array($closing)) {
1040                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1041                                 $yi += sizeof($closing);
1042                         }
1043                 }
1044                 wfProfileOut( $fname );
1045         }
1046 }
1047
1048 /**
1049  * A class to format Diffs
1050  *
1051  * This class formats the diff in classic diff format.
1052  * It is intended that this class be customized via inheritance,
1053  * to obtain fancier outputs.
1054  * @todo document
1055  * @access private
1056  */
1057 class DiffFormatter
1058 {
1059         /**
1060          * Number of leading context "lines" to preserve.
1061          *
1062          * This should be left at zero for this class, but subclasses
1063          * may want to set this to other values.
1064          */
1065         var $leading_context_lines = 0;
1066
1067         /**
1068          * Number of trailing context "lines" to preserve.
1069          *
1070          * This should be left at zero for this class, but subclasses
1071          * may want to set this to other values.
1072          */
1073         var $trailing_context_lines = 0;
1074
1075         /**
1076          * Format a diff.
1077          *
1078          * @param $diff object A Diff object.
1079          * @return string The formatted output.
1080          */
1081         function format($diff) {
1082                 $fname = 'DiffFormatter::format';
1083                 wfProfileIn( $fname );
1084
1085                 $xi = $yi = 1;
1086                 $block = false;
1087                 $context = array();
1088
1089                 $nlead = $this->leading_context_lines;
1090                 $ntrail = $this->trailing_context_lines;
1091
1092                 $this->_start_diff();
1093
1094                 foreach ($diff->edits as $edit) {
1095                         if ($edit->type == 'copy') {
1096                                 if (is_array($block)) {
1097                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1098                                                 $block[] = $edit;
1099                                         }
1100                                         else{
1101                                                 if ($ntrail) {
1102                                                         $context = array_slice($edit->orig, 0, $ntrail);
1103                                                         $block[] = new _DiffOp_Copy($context);
1104                                                 }
1105                                                 $this->_block($x0, $ntrail + $xi - $x0,
1106                                                                           $y0, $ntrail + $yi - $y0,
1107                                                                           $block);
1108                                                 $block = false;
1109                                         }
1110                                 }
1111                                 $context = $edit->orig;
1112                         }
1113                         else {
1114                                 if (! is_array($block)) {
1115                                         $context = array_slice($context, sizeof($context) - $nlead);
1116                                         $x0 = $xi - sizeof($context);
1117                                         $y0 = $yi - sizeof($context);
1118                                         $block = array();
1119                                         if ($context)
1120                                                 $block[] = new _DiffOp_Copy($context);
1121                                 }
1122                                 $block[] = $edit;
1123                         }
1124
1125                         if ($edit->orig)
1126                                 $xi += sizeof($edit->orig);
1127                         if ($edit->closing)
1128                                 $yi += sizeof($edit->closing);
1129                 }
1130
1131                 if (is_array($block))
1132                         $this->_block($x0, $xi - $x0,
1133                                                   $y0, $yi - $y0,
1134                                                   $block);
1135
1136                 $end = $this->_end_diff();
1137                 wfProfileOut( $fname );
1138                 return $end;
1139         }
1140
1141         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1142                 $fname = 'DiffFormatter::_block';
1143                 wfProfileIn( $fname );
1144                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1145                 foreach ($edits as $edit) {
1146                         if ($edit->type == 'copy')
1147                                 $this->_context($edit->orig);
1148                         elseif ($edit->type == 'add')
1149                                 $this->_added($edit->closing);
1150                         elseif ($edit->type == 'delete')
1151                                 $this->_deleted($edit->orig);
1152                         elseif ($edit->type == 'change')
1153                                 $this->_changed($edit->orig, $edit->closing);
1154                         else
1155                                 trigger_error('Unknown edit type', E_USER_ERROR);
1156                 }
1157                 $this->_end_block();
1158                 wfProfileOut( $fname );
1159         }
1160
1161         function _start_diff() {
1162                 ob_start();
1163         }
1164
1165         function _end_diff() {
1166                 $val = ob_get_contents();
1167                 ob_end_clean();
1168                 return $val;
1169         }
1170
1171         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1172                 if ($xlen > 1)
1173                         $xbeg .= "," . ($xbeg + $xlen - 1);
1174                 if ($ylen > 1)
1175                         $ybeg .= "," . ($ybeg + $ylen - 1);
1176
1177                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1178         }
1179
1180         function _start_block($header) {
1181                 echo $header;
1182         }
1183
1184         function _end_block() {
1185         }
1186
1187         function _lines($lines, $prefix = ' ') {
1188                 foreach ($lines as $line)
1189                         echo "$prefix $line\n";
1190         }
1191
1192         function _context($lines) {
1193                 $this->_lines($lines);
1194         }
1195
1196         function _added($lines) {
1197                 $this->_lines($lines, '>');
1198         }
1199         function _deleted($lines) {
1200                 $this->_lines($lines, '<');
1201         }
1202
1203         function _changed($orig, $closing) {
1204                 $this->_deleted($orig);
1205                 echo "---\n";
1206                 $this->_added($closing);
1207         }
1208 }
1209
1210
1211 /**
1212  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1213  *
1214  */
1215
1216 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1217
1218 /**
1219  * @todo document
1220  * @access private
1221  */
1222 class _HWLDF_WordAccumulator {
1223         function _HWLDF_WordAccumulator () {
1224                 $this->_lines = array();
1225                 $this->_line = '';
1226                 $this->_group = '';
1227                 $this->_tag = '';
1228         }
1229
1230         function _flushGroup ($new_tag) {
1231                 if ($this->_group !== '') {
1232                         if ($this->_tag == 'mark')
1233                                 $this->_line .= '<span class="diffchange">' .
1234                                         htmlspecialchars ( $this->_group ) . '</span>';
1235                         else
1236                                 $this->_line .= htmlspecialchars ( $this->_group );
1237                 }
1238                 $this->_group = '';
1239                 $this->_tag = $new_tag;
1240         }
1241
1242         function _flushLine ($new_tag) {
1243                 $this->_flushGroup($new_tag);
1244                 if ($this->_line != '')
1245                         array_push ( $this->_lines, $this->_line );
1246                 else
1247                         # make empty lines visible by inserting an NBSP
1248                         array_push ( $this->_lines, NBSP );
1249                 $this->_line = '';
1250         }
1251
1252         function addWords ($words, $tag = '') {
1253                 if ($tag != $this->_tag)
1254                         $this->_flushGroup($tag);
1255
1256                 foreach ($words as $word) {
1257                         // new-line should only come as first char of word.
1258                         if ($word == '')
1259                                 continue;
1260                         if ($word[0] == "\n") {
1261                                 $this->_flushLine($tag);
1262                                 $word = substr($word, 1);
1263                         }
1264                         assert(!strstr($word, "\n"));
1265                         $this->_group .= $word;
1266                 }
1267         }
1268
1269         function getLines() {
1270                 $this->_flushLine('~done');
1271                 return $this->_lines;
1272         }
1273 }
1274
1275 /**
1276  * @todo document
1277  * @access private
1278  */
1279 class WordLevelDiff extends MappedDiff
1280 {
1281         function WordLevelDiff ($orig_lines, $closing_lines) {
1282                 $fname = 'WordLevelDiff::WordLevelDiff';
1283                 wfProfileIn( $fname );
1284
1285                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1286                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1287
1288
1289                 $this->MappedDiff($orig_words, $closing_words,
1290                                                   $orig_stripped, $closing_stripped);
1291                 wfProfileOut( $fname );
1292         }
1293
1294         function _split($lines) {
1295                 $fname = 'WordLevelDiff::_split';
1296                 wfProfileIn( $fname );
1297                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1298                                                         implode("\n", $lines),
1299                                                         $m)) {
1300                         wfProfileOut( $fname );
1301                         return array(array(''), array(''));
1302                 }
1303                 wfProfileOut( $fname );
1304                 return array($m[0], $m[1]);
1305         }
1306
1307         function orig () {
1308                 $fname = 'WordLevelDiff::orig';
1309                 wfProfileIn( $fname );
1310                 $orig = new _HWLDF_WordAccumulator;
1311
1312                 foreach ($this->edits as $edit) {
1313                         if ($edit->type == 'copy')
1314                                 $orig->addWords($edit->orig);
1315                         elseif ($edit->orig)
1316                                 $orig->addWords($edit->orig, 'mark');
1317                 }
1318                 $lines = $orig->getLines();
1319                 wfProfileOut( $fname );
1320                 return $lines;
1321         }
1322
1323         function closing () {
1324                 $fname = 'WordLevelDiff::closing';
1325                 wfProfileIn( $fname );
1326                 $closing = new _HWLDF_WordAccumulator;
1327
1328                 foreach ($this->edits as $edit) {
1329                         if ($edit->type == 'copy')
1330                                 $closing->addWords($edit->closing);
1331                         elseif ($edit->closing)
1332                                 $closing->addWords($edit->closing, 'mark');
1333                 }
1334                 $lines = $closing->getLines();
1335                 wfProfileOut( $fname );
1336                 return $lines;
1337         }
1338 }
1339
1340 /**
1341  *      Wikipedia Table style diff formatter.
1342  * @todo document
1343  * @access private
1344  */
1345 class TableDiffFormatter extends DiffFormatter
1346 {
1347         function TableDiffFormatter() {
1348                 $this->leading_context_lines = 2;
1349                 $this->trailing_context_lines = 2;
1350         }
1351
1352         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1353                 $l1 = wfMsg( 'lineno', $xbeg );
1354                 $l2 = wfMsg( 'lineno', $ybeg );
1355
1356                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1357                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1358                 return $r;
1359         }
1360
1361         function _start_block( $header ) {
1362                 global $wgOut;
1363                 echo $header;
1364         }
1365
1366         function _end_block() {
1367         }
1368
1369         function _lines( $lines, $prefix=' ', $color='white' ) {
1370         }
1371
1372         # HTML-escape parameter before calling this
1373         function addedLine( $line ) {
1374                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1375         }
1376
1377         # HTML-escape parameter before calling this
1378         function deletedLine( $line ) {
1379                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1380         }
1381
1382         # HTML-escape parameter before calling this
1383         function contextLine( $line ) {
1384                 return "<td> </td><td class='diff-context'>{$line}</td>";
1385         }
1386
1387         function emptyLine() {
1388                 return '<td colspan="2">&nbsp;</td>';
1389         }
1390
1391         function _added( $lines ) {
1392                 foreach ($lines as $line) {
1393                         echo '<tr>' . $this->emptyLine() .
1394                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1395                 }
1396         }
1397
1398         function _deleted($lines) {
1399                 foreach ($lines as $line) {
1400                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1401                           $this->emptyLine() . "</tr>\n";
1402                 }
1403         }
1404
1405         function _context( $lines ) {
1406                 foreach ($lines as $line) {
1407                         echo '<tr>' .
1408                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1409                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1410                 }
1411         }
1412
1413         function _changed( $orig, $closing ) {
1414                 $fname = 'TableDiffFormatter::_changed';
1415                 wfProfileIn( $fname );
1416
1417                 $diff = new WordLevelDiff( $orig, $closing );
1418                 $del = $diff->orig();
1419                 $add = $diff->closing();
1420
1421                 # Notice that WordLevelDiff returns HTML-escaped output.
1422                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1423
1424                 while ( $line = array_shift( $del ) ) {
1425                         $aline = array_shift( $add );
1426                         echo '<tr>' . $this->deletedLine( $line ) .
1427                                 $this->addedLine( $aline ) . "</tr>\n";
1428                 }
1429                 foreach ($add as $line) {       # If any leftovers
1430                         echo '<tr>' . $this->emptyLine() .
1431                                 $this->addedLine( $line ) . "</tr>\n";
1432                 }
1433                 wfProfileOut( $fname );
1434         }
1435 }
1436
1437 ?>