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