includes/DifferenceEngine.php

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