includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @todo indicate where diff.doc can be found.
   5  * @addtogroup DifferenceEngine
   6  */
   7
   8 /**
   9  * Constant to indicate diff cache compatibility.
  10  * Bump this when changing the diff formatting in a way that
  11  * fixes important bugs or such to force cached diff views to
  12  * clear.
  13  */
  14 define( 'MW_DIFF_VERSION', '1.11a' );
  15
  16 /**
  17  * @todo document
  18  * @public
  19  * @addtogroup DifferenceEngine
  20  */
  21 class DifferenceEngine {
  22         /**#@+
  23          * @private
  24          */
  25         var $mOldid, $mNewid, $mTitle;
  26         var $mOldtitle, $mNewtitle, $mPagetitle;
  27         var $mOldtext, $mNewtext;
  28         var $mOldPage, $mNewPage;
  29         var $mRcidMarkPatrolled;
  30         var $mOldRev, $mNewRev;
  31         var $mRevisionsLoaded = false; // Have the revisions been loaded
  32         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  33         /**#@-*/
  34
  35         /**
  36          * Constructor
  37          * @param $titleObj Title object that the diff is associated with
  38          * @param $old Integer: old ID we want to show and diff with.
  39          * @param $new String: either 'prev' or 'next'.
  40          * @param $rcid Integer: ??? FIXME (default 0)
  41          * @param $refreshCache boolean If set, refreshes the diff cache
  42          */
  43         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0, $refreshCache = false ) {
  44                 $this->mTitle = $titleObj;
  45                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  46
  47                 if ( 'prev' === $new ) {
  48                         # Show diff between revision $old and the previous one.
  49                         # Get previous one from DB.
  50                         #
  51                         $this->mNewid = intval($old);
  52
  53                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  54
  55                 } elseif ( 'next' === $new ) {
  56                         # Show diff between revision $old and the previous one.
  57                         # Get previous one from DB.
  58                         #
  59                         $this->mOldid = intval($old);
  60                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  61                         if ( false === $this->mNewid ) {
  62                                 # if no result, NewId points to the newest old revision. The only newer
  63                                 # revision is cur, which is "0".
  64                                 $this->mNewid = 0;
  65                         }
  66
  67                 } else {
  68                         $this->mOldid = intval($old);
  69                         $this->mNewid = intval($new);
  70                 }
  71                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  72                 $this->mRefreshCache = $refreshCache;
  73         }
  74
  75         function showDiffPage( $diffOnly = false ) {
  76                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  77                 $fname = 'DifferenceEngine::showDiffPage';
  78                 wfProfileIn( $fname );
  79
  80                 # If external diffs are enabled both globally and for the user,
  81                 # we'll use the application/x-external-editor interface to call
  82                 # an external diff tool like kompare, kdiff3, etc.
  83                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  84                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  85                         $wgOut->disable();
  86                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  87                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  88                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  89                         $special=$wgLang->getNsText(NS_SPECIAL);
  90                         $control=<<<CONTROL
  91 [Process]
  92 Type=Diff text
  93 Engine=MediaWiki
  94 Script={$wgServer}{$wgScript}
  95 Special namespace={$special}
  96
  97 [File]
  98 Extension=wiki
  99 URL=$url1
 100
 101 [File 2]
 102 Extension=wiki
 103 URL=$url2
 104 CONTROL;
 105                         echo($control);
 106                         return;
 107                 }
 108
 109                 $wgOut->setArticleFlag( false );
 110                 if ( ! $this->loadRevisionData() ) {
 111                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, {$this->mNewid})";
 112                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 113                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 114                         $wgOut->addWikitext( $mtext );
 115                         wfProfileOut( $fname );
 116                         return;
 117                 }
 118
 119                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 120
 121                 if ( $this->mNewRev->isCurrent() ) {
 122                         $wgOut->setArticleFlag( true );
 123                 }
 124
 125                 # mOldid is false if the difference engine is called with a "vague" query for
 126                 # a diff between a version V and its previous version V' AND the version V
 127                 # is the first version of that article. In that case, V' does not exist.
 128                 if ( $this->mOldid === false ) {
 129                         $this->showFirstRevision();
 130                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 131                         wfProfileOut( $fname );
 132                         return;
 133                 }
 134
 135                 $wgOut->suppressQuickbar();
 136
 137                 $oldTitle = $this->mOldPage->getPrefixedText();
 138                 $newTitle = $this->mNewPage->getPrefixedText();
 139                 if( $oldTitle == $newTitle ) {
 140                         $wgOut->setPageTitle( $newTitle );
 141                 } else {
 142                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 143                 }
 144                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 145                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 146
 147                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 148                         $wgOut->loginToUse();
 149                         $wgOut->output();
 150                         wfProfileOut( $fname );
 151                         exit;
 152                 }
 153
 154                 $sk = $wgUser->getSkin();
 155
 156                 if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 157                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 158                 } else {
 159                         $rollback = '';
 160                 }
 161
 162                 // Prepare a change patrol link, if applicable
 163                 if( $wgUseRCPatrol && $wgUser->isAllowed( 'patrol' ) ) {
 164                         // If we've been given an explicit change identifier, use it; saves time
 165                         if( $this->mRcidMarkPatrolled ) {
 166                                 $rcid = $this->mRcidMarkPatrolled;
 167                         } else {
 168                                 // Look for an unpatrolled change corresponding to this diff
 169                                 $db = wfGetDB( DB_SLAVE );
 170                                 $change = RecentChange::newFromConds(
 171                                         array(
 172                                                 // Add redundant timestamp condition so we can use the
 173                                                 // existing index
 174                                                 'rc_timestamp' => $db->timestamp( $this->mNewRev->getTimestamp() ),
 175                                                 'rc_this_oldid' => $this->mNewid,
 176                                                 'rc_last_oldid' => $this->mOldid,
 177                                                 'rc_patrolled' => 0,
 178                                         ),
 179                                         __METHOD__
 180                                 );
 181                                 if( $change instanceof RecentChange ) {
 182                                         $rcid = $change->mAttribs['rc_id'];
 183                                 } else {
 184                                         // None found
 185                                         $rcid = 0;
 186                                 }
 187                         }
 188                         // Build the link
 189                         if( $rcid ) {
 190                                 $patrol = ' [' . $sk->makeKnownLinkObj(
 191                                         $this->mTitle,
 192                                         wfMsgHtml( 'markaspatrolleddiff' ),
 193                                         "action=markpatrolled&rcid={$rcid}"
 194                                 ) . ']';
 195                         } else {
 196                                 $patrol = '';
 197                         }
 198                 } else {
 199                         $patrol = '';
 200                 }
 201
 202                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 203                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 204                 if ( $this->mNewRev->isCurrent() ) {
 205                         $nextlink = '&nbsp;';
 206                 } else {
 207                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 208                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 209                 }
 210
 211                 $oldminor = '';
 212                 $newminor = '';
 213
 214                 if ($this->mOldRev->mMinorEdit == 1) {
 215                         $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
 216                                 wfMsg( 'minoreditletter') ) . ' ';
 217                 }
 218
 219                 if ($this->mNewRev->mMinorEdit == 1) {
 220                         $newminor = wfElement( 'span', array( 'class' => 'minor' ),
 221                         wfMsg( 'minoreditletter') ) . ' ';
 222                 }
 223
 224                 $rdel = ''; $ldel = '';
 225                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 226                         $revdel = SpecialPage::getTitleFor( 'Revisiondelete' );
 227                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 228                         // If revision was hidden from sysops
 229                                 $ldel = wfMsgHtml('rev-delundel');
 230                         } else {
 231                                 $ldel = $sk->makeKnownLinkObj( $revdel,
 232                                         wfMsgHtml('rev-delundel'),
 233                                         'target=' . urlencode( $this->mOldRev->mTitle->getPrefixedDbkey() ) .
 234                                         '&oldid=' . urlencode( $this->mOldRev->getId() ) );
 235                                 // Bolden oversighted content
 236                                 if( $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 237                                         $ldel = "<strong>$ldel</strong>";
 238                         }
 239                         $ldel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$ldel</small>)</tt> ";
 240                         // We don't currently handle well changing the top revision's settings
 241                         if( $this->mNewRev->isCurrent() ) {
 242                         // If revision was hidden from sysops
 243                                 $rdel = wfMsgHtml('rev-delundel');
 244                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 245                         // If revision was hidden from sysops
 246                                 $rdel = wfMsgHtml('rev-delundel');
 247                         } else {
 248                                 $rdel = $sk->makeKnownLinkObj( $revdel,
 249                                         wfMsgHtml('rev-delundel'),
 250                                         'target=' . urlencode( $this->mNewRev->mTitle->getPrefixedDbkey() ) .
 251                                         '&oldid=' . urlencode( $this->mNewRev->getId() ) );
 252                                 // Bolden oversighted content
 253                                 if( $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 254                                         $rdel = "<strong>$rdel</strong>";
 255                         }
 256                         $rdel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$rdel</small>)</tt> ";
 257                 }
 258
 259                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 260                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, true ) . "</div>" .
 261                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, true ) . $ldel . "</div>" .
 262                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 263                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 264                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, true ) . " $rollback</div>" .
 265                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, true ) . $rdel . "</div>" .
 266                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 267
 268                 $this->showDiff( $oldHeader, $newHeader );
 269
 270                 if ( !$diffOnly )
 271                         $this->renderNewRevision();
 272
 273                 wfProfileOut( $fname );
 274         }
 275
 276         /**
 277          * Show the new revision of the page.
 278          */
 279         function renderNewRevision() {
 280                 global $wgOut;
 281                 $fname = 'DifferenceEngine::renderNewRevision';
 282                 wfProfileIn( $fname );
 283
 284                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 285                 #add deleted rev tag if needed
 286                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 287                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-permission' ) );
 288                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 289                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-view' ) );
 290                 }
 291
 292                 if( !$this->mNewRev->isCurrent() ) {
 293                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 294                 }
 295
 296                 $this->loadNewText();
 297                 if( is_object( $this->mNewRev ) ) {
 298                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 299                 }
 300
 301                 if ($this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage()) {
 302                         // Stolen from Article::view --AG 2007-10-11
 303
 304                         // Give hooks a chance to customise the output
 305                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 306                                 // Wrap the whole lot in a <pre> and don't parse
 307                                 $m = array();
 308                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 309                                 $wgOut->addHtml( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 310                                 $wgOut->addHtml( htmlspecialchars( $this->mNewtext ) );
 311                                 $wgOut->addHtml( "\n</pre>\n" );
 312                         }
 313                 } else
 314                         $wgOut->addWikiTextTidy( $this->mNewtext );
 315
 316                 if( !$this->mNewRev->isCurrent() ) {
 317                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 318                 }
 319
 320                 wfProfileOut( $fname );
 321         }
 322
 323         /**
 324          * Show the first revision of an article. Uses normal diff headers in
 325          * contrast to normal "old revision" display style.
 326          */
 327         function showFirstRevision() {
 328                 global $wgOut, $wgUser;
 329
 330                 $fname = 'DifferenceEngine::showFirstRevision';
 331                 wfProfileIn( $fname );
 332
 333                 # Get article text from the DB
 334                 #
 335                 if ( ! $this->loadNewText() ) {
 336                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 337                           "{$this->mNewid})";
 338                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 339                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 340                         $wgOut->addWikitext( $mtext );
 341                         wfProfileOut( $fname );
 342                         return;
 343                 }
 344                 if ( $this->mNewRev->isCurrent() ) {
 345                         $wgOut->setArticleFlag( true );
 346                 }
 347
 348                 # Check if user is allowed to look at this page. If not, bail out.
 349                 #
 350                 if ( !( $this->mTitle->userCanRead() ) ) {
 351                         $wgOut->loginToUse();
 352                         $wgOut->output();
 353                         wfProfileOut( $fname );
 354                         exit;
 355                 }
 356
 357                 # Prepare the header box
 358                 #
 359                 $sk = $wgUser->getSkin();
 360
 361                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 362                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 363                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 364                         $sk->revComment( $this->mNewRev ) . "<br />" .
 365                         $nextlink . "</div>\n";
 366
 367                 $wgOut->addHTML( $header );
 368
 369                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 370                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 371
 372                 wfProfileOut( $fname );
 373         }
 374
 375         /**
 376          * Get the diff text, send it to $wgOut
 377          * Returns false if the diff could not be generated, otherwise returns true
 378          */
 379         function showDiff( $otitle, $ntitle ) {
 380                 global $wgOut;
 381                 $diff = $this->getDiff( $otitle, $ntitle );
 382                 if ( $diff === false ) {
 383                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
 384                         return false;
 385                 } else {
 386                         $this->showDiffStyle();
 387                         $wgOut->addHTML( $diff );
 388                         return true;
 389                 }
 390         }
 391
 392         /**
 393          * Add style sheets and supporting JS for diff display.
 394          */
 395         function showDiffStyle() {
 396                 global $wgStylePath, $wgStyleVersion, $wgOut;
 397                 $wgOut->addStyle( 'common/diff.css' );
 398
 399                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 400                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 401         }
 402
 403         /**
 404          * Get complete diff table, including header
 405          *
 406          * @param Title $otitle Old title
 407          * @param Title $ntitle New title
 408          * @return mixed
 409          */
 410         function getDiff( $otitle, $ntitle ) {
 411                 $body = $this->getDiffBody();
 412                 if ( $body === false ) {
 413                         return false;
 414                 } else {
 415                         $multi = $this->getMultiNotice();
 416                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 417                 }
 418         }
 419
 420         /**
 421          * Get the diff table body, without header
 422          *
 423          * @return mixed
 424          */
 425         function getDiffBody() {
 426                 global $wgMemc;
 427                 $fname = 'DifferenceEngine::getDiffBody';
 428                 wfProfileIn( $fname );
 429
 430                 // Cacheable?
 431                 $key = false;
 432                 if ( $this->mOldid && $this->mNewid ) {
 433                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 434                         // Try cache
 435                         if ( !$this->mRefreshCache ) {
 436                                 $difftext = $wgMemc->get( $key );
 437                                 if ( $difftext ) {
 438                                         wfIncrStats( 'diff_cache_hit' );
 439                                         $difftext = $this->localiseLineNumbers( $difftext );
 440                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 441                                         wfProfileOut( $fname );
 442                                         return $difftext;
 443                                 }
 444                         } // don't try to load but save the result
 445                 }
 446
 447                 // Loadtext is permission safe, this just clears out the diff
 448                 if ( !$this->loadText() ) {
 449                         wfProfileOut( $fname );
 450                         return false;
 451                 } else if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 452                         return '';
 453                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 454                         return '';
 455                 }
 456
 457                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 458
 459                 // Save to cache for 7 days
 460                 // Only do this for public revs, otherwise an admin can view the diff and a non-admin can nab it!
 461                 if ( $this->mOldRev && $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 462                         wfIncrStats( 'diff_uncacheable' );
 463                 } else if ( $this->mNewRev && $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 464                         wfIncrStats( 'diff_uncacheable' );
 465                 } else if ( $key !== false && $difftext !== false ) {
 466                         wfIncrStats( 'diff_cache_miss' );
 467                         $wgMemc->set( $key, $difftext, 7*86400 );
 468                 } else {
 469                         wfIncrStats( 'diff_uncacheable' );
 470                 }
 471                 // Replace line numbers with the text in the user's language
 472                 if ( $difftext !== false ) {
 473                         $difftext = $this->localiseLineNumbers( $difftext );
 474                 }
 475                 wfProfileOut( $fname );
 476                 return $difftext;
 477         }
 478
 479         /**
 480          * Generate a diff, no caching
 481          * $otext and $ntext must be already segmented
 482          */
 483         function generateDiffBody( $otext, $ntext ) {
 484                 global $wgExternalDiffEngine, $wgContLang;
 485                 $fname = 'DifferenceEngine::generateDiffBody';
 486
 487                 $otext = str_replace( "\r\n", "\n", $otext );
 488                 $ntext = str_replace( "\r\n", "\n", $ntext );
 489
 490                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 491                         # For historical reasons, external diff engine expects
 492                         # input text to be HTML-escaped already
 493                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 494                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 495                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 496                                 dl('php_wikidiff.so');
 497                         }
 498                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 499                 }
 500
 501                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 502                         # Better external diff engine, the 2 may some day be dropped
 503                         # This one does the escaping and segmenting itself
 504                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 505                                 wfProfileIn( "$fname-dl" );
 506                                 @dl('php_wikidiff2.so');
 507                                 wfProfileOut( "$fname-dl" );
 508                         }
 509                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 510                                 wfProfileIn( 'wikidiff2_do_diff' );
 511                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 512                                 wfProfileOut( 'wikidiff2_do_diff' );
 513                                 return $text;
 514                         }
 515                 }
 516                 if ( $wgExternalDiffEngine !== false ) {
 517                         # Diff via the shell
 518                         global $wgTmpDirectory;
 519                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 520                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 521
 522                         $tempFile1 = fopen( $tempName1, "w" );
 523                         if ( !$tempFile1 ) {
 524                                 wfProfileOut( $fname );
 525                                 return false;
 526                         }
 527                         $tempFile2 = fopen( $tempName2, "w" );
 528                         if ( !$tempFile2 ) {
 529                                 wfProfileOut( $fname );
 530                                 return false;
 531                         }
 532                         fwrite( $tempFile1, $otext );
 533                         fwrite( $tempFile2, $ntext );
 534                         fclose( $tempFile1 );
 535                         fclose( $tempFile2 );
 536                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 537                         wfProfileIn( "$fname-shellexec" );
 538                         $difftext = wfShellExec( $cmd );
 539                         wfProfileOut( "$fname-shellexec" );
 540                         unlink( $tempName1 );
 541                         unlink( $tempName2 );
 542                         return $difftext;
 543                 }
 544
 545                 # Native PHP diff
 546                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 547                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 548                 $diffs = new Diff( $ota, $nta );
 549                 $formatter = new TableDiffFormatter();
 550                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 551         }
 552
 553
 554         /**
 555          * Replace line numbers with the text in the user's language
 556          */
 557         function localiseLineNumbers( $text ) {
 558                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 559                         array( &$this, 'localiseLineNumbersCb' ), $text );
 560         }
 561
 562         function localiseLineNumbersCb( $matches ) {
 563                 global $wgLang;
 564                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 565         }
 566
 567
 568         /**
 569          * If there are revisions between the ones being compared, return a note saying so.
 570          */
 571         function getMultiNotice() {
 572                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 573                         return '';
 574
 575                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 576                         // Comparing two different pages? Count would be meaningless.
 577                         return '';
 578                 }
 579
 580                 $oldid = $this->mOldRev->getId();
 581                 $newid = $this->mNewRev->getId();
 582                 if ( $oldid > $newid ) {
 583                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 584                 }
 585
 586                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 587                 if ( !$n )
 588                         return '';
 589
 590                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 591         }
 592
 593
 594         /**
 595          * Add the header to a diff body
 596          */
 597         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 598                 global $wgOut;
 599
 600                 $header = "
 601                         <table class='diff'>
 602                         <col class='diff-marker' />
 603                         <col class='diff-content' />
 604                         <col class='diff-marker' />
 605                         <col class='diff-content' />
 606                         <tr>
 607                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 608                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 609                         </tr>
 610                 ";
 611
 612                 if ( $multi != '' )
 613                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 614
 615                 return $header . $diff . "</table>";
 616         }
 617
 618         /**
 619          * Use specified text instead of loading from the database
 620          */
 621         function setText( $oldText, $newText ) {
 622                 $this->mOldtext = $oldText;
 623                 $this->mNewtext = $newText;
 624                 $this->mTextLoaded = 2;
 625         }
 626
 627         /**
 628          * Load revision metadata for the specified articles. If newid is 0, then compare
 629          * the old article in oldid to the current article; if oldid is 0, then
 630          * compare the current article to the immediately previous one (ignoring the
 631          * value of newid).
 632          *
 633          * If oldid is false, leave the corresponding revision object set
 634          * to false. This is impossible via ordinary user input, and is provided for
 635          * API convenience.
 636          */
 637         function loadRevisionData() {
 638                 global $wgLang;
 639                 if ( $this->mRevisionsLoaded ) {
 640                         return true;
 641                 } else {
 642                         // Whether it succeeds or fails, we don't want to try again
 643                         $this->mRevisionsLoaded = true;
 644                 }
 645
 646                 // Load the new revision object
 647                 $this->mNewRev = $this->mNewid
 648                         ? Revision::newFromId( $this->mNewid )
 649                         : Revision::newFromTitle( $this->mTitle );
 650                 if( !$this->mNewRev instanceof Revision )
 651                         return false;
 652
 653                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 654                 $this->mNewid = $this->mNewRev->getId();
 655
 656                 // Set assorted variables
 657                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 658                 $this->mNewPage = $this->mNewRev->getTitle();
 659                 if( $this->mNewRev->isCurrent() ) {
 660                         $newLink = $this->mNewPage->escapeLocalUrl();
 661                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 662                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 663
 664                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
 665                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 666
 667                 } else {
 668                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 669                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 670                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 671
 672                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
 673                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 674                 }
 675                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 676                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 677                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 678                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 679                 }
 680
 681                 // Load the old revision object
 682                 $this->mOldRev = false;
 683                 if( $this->mOldid ) {
 684                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 685                 } elseif ( $this->mOldid === 0 ) {
 686                         $rev = $this->mNewRev->getPrevious();
 687                         if( $rev ) {
 688                                 $this->mOldid = $rev->getId();
 689                                 $this->mOldRev = $rev;
 690                         } else {
 691                                 // No previous revision; mark to show as first-version only.
 692                                 $this->mOldid = false;
 693                                 $this->mOldRev = false;
 694                         }
 695                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 696
 697                 if( is_null( $this->mOldRev ) ) {
 698                         return false;
 699                 }
 700
 701                 if ( $this->mOldRev ) {
 702                         $this->mOldPage = $this->mOldRev->getTitle();
 703
 704                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 705                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 706                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 707                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 708
 709                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 710                                 . " (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 711                         // Add an "undo" link
 712                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 713                         if ( $this->mNewRev->userCan(Revision::DELETED_TEXT) )
 714                                 $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 715
 716                         if ( !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 717                                 $this->mOldtitle = "<span class='history-deleted'>{$this->mOldPagetitle}</span>";
 718                         } else if ( $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 719                                 $this->mOldtitle = '<span class="history-deleted">'.$this->mOldtitle.'</span>';
 720                         }
 721                 }
 722
 723                 return true;
 724         }
 725
 726         /**
 727          * Load the text of the revisions, as well as revision data.
 728          */
 729         function loadText() {
 730                 if ( $this->mTextLoaded == 2 ) {
 731                         return true;
 732                 } else {
 733                         // Whether it succeeds or fails, we don't want to try again
 734                         $this->mTextLoaded = 2;
 735                 }
 736
 737                 if ( !$this->loadRevisionData() ) {
 738                         return false;
 739                 }
 740                 if ( $this->mOldRev ) {
 741                         $this->mOldtext = $this->mOldRev->revText();
 742                         if ( $this->mOldtext === false ) {
 743                                 return false;
 744                         }
 745                 }
 746                 if ( $this->mNewRev ) {
 747                         $this->mNewtext = $this->mNewRev->revText();
 748                         if ( $this->mNewtext === false ) {
 749                                 return false;
 750                         }
 751                 }
 752                 return true;
 753         }
 754
 755         /**
 756          * Load the text of the new revision, not the old one
 757          */
 758         function loadNewText() {
 759                 if ( $this->mTextLoaded >= 1 ) {
 760                         return true;
 761                 } else {
 762                         $this->mTextLoaded = 1;
 763                 }
 764                 if ( !$this->loadRevisionData() ) {
 765                         return false;
 766                 }
 767                 $this->mNewtext = $this->mNewRev->getText();
 768                 return true;
 769         }
 770
 771
 772 }
 773
 774 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 775 //
 776 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 777 // You may copy this code freely under the conditions of the GPL.
 778 //
 779
 780 define('USE_ASSERTS', function_exists('assert'));
 781
 782 /**
 783  * @todo document
 784  * @private
 785  * @addtogroup DifferenceEngine
 786  */
 787 class _DiffOp {
 788         var $type;
 789         var $orig;
 790         var $closing;
 791
 792         function reverse() {
 793                 trigger_error('pure virtual', E_USER_ERROR);
 794         }
 795
 796         function norig() {
 797                 return $this->orig ? sizeof($this->orig) : 0;
 798         }
 799
 800         function nclosing() {
 801                 return $this->closing ? sizeof($this->closing) : 0;
 802         }
 803 }
 804
 805 /**
 806  * @todo document
 807  * @private
 808  * @addtogroup DifferenceEngine
 809  */
 810 class _DiffOp_Copy extends _DiffOp {
 811         var $type = 'copy';
 812
 813         function _DiffOp_Copy ($orig, $closing = false) {
 814                 if (!is_array($closing))
 815                         $closing = $orig;
 816                 $this->orig = $orig;
 817                 $this->closing = $closing;
 818         }
 819
 820         function reverse() {
 821                 return new _DiffOp_Copy($this->closing, $this->orig);
 822         }
 823 }
 824
 825 /**
 826  * @todo document
 827  * @private
 828  * @addtogroup DifferenceEngine
 829  */
 830 class _DiffOp_Delete extends _DiffOp {
 831         var $type = 'delete';
 832
 833         function _DiffOp_Delete ($lines) {
 834                 $this->orig = $lines;
 835                 $this->closing = false;
 836         }
 837
 838         function reverse() {
 839                 return new _DiffOp_Add($this->orig);
 840         }
 841 }
 842
 843 /**
 844  * @todo document
 845  * @private
 846  * @addtogroup DifferenceEngine
 847  */
 848 class _DiffOp_Add extends _DiffOp {
 849         var $type = 'add';
 850
 851         function _DiffOp_Add ($lines) {
 852                 $this->closing = $lines;
 853                 $this->orig = false;
 854         }
 855
 856         function reverse() {
 857                 return new _DiffOp_Delete($this->closing);
 858         }
 859 }
 860
 861 /**
 862  * @todo document
 863  * @private
 864  * @addtogroup DifferenceEngine
 865  */
 866 class _DiffOp_Change extends _DiffOp {
 867         var $type = 'change';
 868
 869         function _DiffOp_Change ($orig, $closing) {
 870                 $this->orig = $orig;
 871                 $this->closing = $closing;
 872         }
 873
 874         function reverse() {
 875                 return new _DiffOp_Change($this->closing, $this->orig);
 876         }
 877 }
 878
 879
 880 /**
 881  * Class used internally by Diff to actually compute the diffs.
 882  *
 883  * The algorithm used here is mostly lifted from the perl module
 884  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 885  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 886  *
 887  * More ideas are taken from:
 888  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 889  *
 890  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 891  * diffutils-2.7, which can be found at:
 892  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 893  *
 894  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 895  * are my own.
 896  *
 897  * Line length limits for robustness added by Tim Starling, 2005-08-31
 898  *
 899  * @author Geoffrey T. Dairiki, Tim Starling
 900  * @private
 901  * @addtogroup DifferenceEngine
 902  */
 903 class _DiffEngine
 904 {
 905         const MAX_XREF_LENGTH =  10000;
 906
 907         function diff ($from_lines, $to_lines) {
 908                 $fname = '_DiffEngine::diff';
 909                 wfProfileIn( $fname );
 910
 911                 $n_from = sizeof($from_lines);
 912                 $n_to = sizeof($to_lines);
 913
 914                 $this->xchanged = $this->ychanged = array();
 915                 $this->xv = $this->yv = array();
 916                 $this->xind = $this->yind = array();
 917                 unset($this->seq);
 918                 unset($this->in_seq);
 919                 unset($this->lcs);
 920
 921                 // Skip leading common lines.
 922                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 923                         if ($from_lines[$skip] !== $to_lines[$skip])
 924                                 break;
 925                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 926                 }
 927                 // Skip trailing common lines.
 928                 $xi = $n_from; $yi = $n_to;
 929                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 930                         if ($from_lines[$xi] !== $to_lines[$yi])
 931                                 break;
 932                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 933                 }
 934
 935                 // Ignore lines which do not exist in both files.
 936                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 937                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 938                 }
 939
 940                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 941                         $line = $to_lines[$yi];
 942                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 943                                 continue;
 944                         $yhash[$this->_line_hash($line)] = 1;
 945                         $this->yv[] = $line;
 946                         $this->yind[] = $yi;
 947                 }
 948                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 949                         $line = $from_lines[$xi];
 950                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 951                                 continue;
 952                         $this->xv[] = $line;
 953                         $this->xind[] = $xi;
 954                 }
 955
 956                 // Find the LCS.
 957                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 958
 959                 // Merge edits when possible
 960                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 961                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 962
 963                 // Compute the edit operations.
 964                 $edits = array();
 965                 $xi = $yi = 0;
 966                 while ($xi < $n_from || $yi < $n_to) {
 967                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 968                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 969
 970                         // Skip matching "snake".
 971                         $copy = array();
 972                         while ( $xi < $n_from && $yi < $n_to
 973                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 974                                 $copy[] = $from_lines[$xi++];
 975                                 ++$yi;
 976                         }
 977                         if ($copy)
 978                                 $edits[] = new _DiffOp_Copy($copy);
 979
 980                         // Find deletes & adds.
 981                         $delete = array();
 982                         while ($xi < $n_from && $this->xchanged[$xi])
 983                                 $delete[] = $from_lines[$xi++];
 984
 985                         $add = array();
 986                         while ($yi < $n_to && $this->ychanged[$yi])
 987                                 $add[] = $to_lines[$yi++];
 988
 989                         if ($delete && $add)
 990                                 $edits[] = new _DiffOp_Change($delete, $add);
 991                         elseif ($delete)
 992                                 $edits[] = new _DiffOp_Delete($delete);
 993                         elseif ($add)
 994                                 $edits[] = new _DiffOp_Add($add);
 995                 }
 996                 wfProfileOut( $fname );
 997                 return $edits;
 998         }
 999
1000         /**
1001          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1002          */
1003         function _line_hash( $line ) {
1004                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1005                         return md5( $line );
1006                 } else {
1007                         return $line;
1008                 }
1009         }
1010
1011
1012         /* Divide the Largest Common Subsequence (LCS) of the sequences
1013          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1014          * sized segments.
1015          *
1016          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1017          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1018          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1019          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1020          * that (X0, Y0) == (XOFF, YOFF) and
1021          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1022          *
1023          * This function assumes that the first lines of the specified portions
1024          * of the two files do not match, and likewise that the last lines do not
1025          * match.  The caller must trim matching lines from the beginning and end
1026          * of the portions it is going to specify.
1027          */
1028         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1029                 $fname = '_DiffEngine::_diag';
1030                 wfProfileIn( $fname );
1031                 $flip = false;
1032
1033                 if ($xlim - $xoff > $ylim - $yoff) {
1034                         // Things seems faster (I'm not sure I understand why)
1035                                 // when the shortest sequence in X.
1036                                 $flip = true;
1037                         list ($xoff, $xlim, $yoff, $ylim)
1038                         = array( $yoff, $ylim, $xoff, $xlim);
1039                 }
1040
1041                 if ($flip)
1042                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1043                                 $ymatches[$this->xv[$i]][] = $i;
1044                 else
1045                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1046                                 $ymatches[$this->yv[$i]][] = $i;
1047
1048                 $this->lcs = 0;
1049                 $this->seq[0]= $yoff - 1;
1050                 $this->in_seq = array();
1051                 $ymids[0] = array();
1052
1053                 $numer = $xlim - $xoff + $nchunks - 1;
1054                 $x = $xoff;
1055                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1056                         wfProfileIn( "$fname-chunk" );
1057                         if ($chunk > 0)
1058                                 for ($i = 0; $i <= $this->lcs; $i++)
1059                                         $ymids[$i][$chunk-1] = $this->seq[$i];
1060
1061                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1062                         for ( ; $x < $x1; $x++) {
1063                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1064                                         if (empty($ymatches[$line]))
1065                                                 continue;
1066                                 $matches = $ymatches[$line];
1067                                 reset($matches);
1068                                 while (list ($junk, $y) = each($matches))
1069                                         if (empty($this->in_seq[$y])) {
1070                                                 $k = $this->_lcs_pos($y);
1071                                                 USE_ASSERTS && assert($k > 0);
1072                                                 $ymids[$k] = $ymids[$k-1];
1073                                                 break;
1074                                         }
1075                                 while (list ( /* $junk */, $y) = each($matches)) {
1076                                         if ($y > $this->seq[$k-1]) {
1077                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1078                                                 // Optimization: this is a common case:
1079                                                 //      next match is just replacing previous match.
1080                                                 $this->in_seq[$this->seq[$k]] = false;
1081                                                 $this->seq[$k] = $y;
1082                                                 $this->in_seq[$y] = 1;
1083                                         } else if (empty($this->in_seq[$y])) {
1084                                                 $k = $this->_lcs_pos($y);
1085                                                 USE_ASSERTS && assert($k > 0);
1086                                                 $ymids[$k] = $ymids[$k-1];
1087                                         }
1088                                 }
1089                         }
1090                         wfProfileOut( "$fname-chunk" );
1091                 }
1092
1093                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1094                 $ymid = $ymids[$this->lcs];
1095                 for ($n = 0; $n < $nchunks - 1; $n++) {
1096                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1097                         $y1 = $ymid[$n] + 1;
1098                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1099                 }
1100                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1101
1102                 wfProfileOut( $fname );
1103                 return array($this->lcs, $seps);
1104         }
1105
1106         function _lcs_pos ($ypos) {
1107                 $fname = '_DiffEngine::_lcs_pos';
1108                 wfProfileIn( $fname );
1109
1110                 $end = $this->lcs;
1111                 if ($end == 0 || $ypos > $this->seq[$end]) {
1112                         $this->seq[++$this->lcs] = $ypos;
1113                         $this->in_seq[$ypos] = 1;
1114                         wfProfileOut( $fname );
1115                         return $this->lcs;
1116                 }
1117
1118                 $beg = 1;
1119                 while ($beg < $end) {
1120                         $mid = (int)(($beg + $end) / 2);
1121                         if ( $ypos > $this->seq[$mid] )
1122                                 $beg = $mid + 1;
1123                         else
1124                                 $end = $mid;
1125                 }
1126
1127                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1128
1129                 $this->in_seq[$this->seq[$end]] = false;
1130                 $this->seq[$end] = $ypos;
1131                 $this->in_seq[$ypos] = 1;
1132                 wfProfileOut( $fname );
1133                 return $end;
1134         }
1135
1136         /* Find LCS of two sequences.
1137          *
1138          * The results are recorded in the vectors $this->{x,y}changed[], by
1139          * storing a 1 in the element for each line that is an insertion
1140          * or deletion (ie. is not in the LCS).
1141          *
1142          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1143          *
1144          * Note that XLIM, YLIM are exclusive bounds.
1145          * All line numbers are origin-0 and discarded lines are not counted.
1146          */
1147         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1148                 $fname = '_DiffEngine::_compareseq';
1149                 wfProfileIn( $fname );
1150
1151                 // Slide down the bottom initial diagonal.
1152                 while ($xoff < $xlim && $yoff < $ylim
1153                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1154                         ++$xoff;
1155                         ++$yoff;
1156                 }
1157
1158                 // Slide up the top initial diagonal.
1159                 while ($xlim > $xoff && $ylim > $yoff
1160                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1161                         --$xlim;
1162                         --$ylim;
1163                 }
1164
1165                 if ($xoff == $xlim || $yoff == $ylim)
1166                         $lcs = 0;
1167                 else {
1168                         // This is ad hoc but seems to work well.
1169                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1170                         //$nchunks = max(2,min(8,(int)$nchunks));
1171                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1172                         list ($lcs, $seps)
1173                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1174                 }
1175
1176                 if ($lcs == 0) {
1177                         // X and Y sequences have no common subsequence:
1178                         // mark all changed.
1179                         while ($yoff < $ylim)
1180                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1181                         while ($xoff < $xlim)
1182                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1183                 } else {
1184                         // Use the partitions to split this problem into subproblems.
1185                         reset($seps);
1186                         $pt1 = $seps[0];
1187                         while ($pt2 = next($seps)) {
1188                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1189                                 $pt1 = $pt2;
1190                         }
1191                 }
1192                 wfProfileOut( $fname );
1193         }
1194
1195         /* Adjust inserts/deletes of identical lines to join changes
1196          * as much as possible.
1197          *
1198          * We do something when a run of changed lines include a
1199          * line at one end and has an excluded, identical line at the other.
1200          * We are free to choose which identical line is included.
1201          * `compareseq' usually chooses the one at the beginning,
1202          * but usually it is cleaner to consider the following identical line
1203          * to be the "change".
1204          *
1205          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1206          */
1207         function _shift_boundaries ($lines, &$changed, $other_changed) {
1208                 $fname = '_DiffEngine::_shift_boundaries';
1209                 wfProfileIn( $fname );
1210                 $i = 0;
1211                 $j = 0;
1212
1213                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1214                 $len = sizeof($lines);
1215                 $other_len = sizeof($other_changed);
1216
1217                 while (1) {
1218                         /*
1219                          * Scan forwards to find beginning of another run of changes.
1220                          * Also keep track of the corresponding point in the other file.
1221                          *
1222                          * Throughout this code, $i and $j are adjusted together so that
1223                          * the first $i elements of $changed and the first $j elements
1224                          * of $other_changed both contain the same number of zeros
1225                          * (unchanged lines).
1226                          * Furthermore, $j is always kept so that $j == $other_len or
1227                          * $other_changed[$j] == false.
1228                          */
1229                         while ($j < $other_len && $other_changed[$j])
1230                                 $j++;
1231
1232                         while ($i < $len && ! $changed[$i]) {
1233                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1234                                 $i++; $j++;
1235                                 while ($j < $other_len && $other_changed[$j])
1236                                         $j++;
1237                         }
1238
1239                         if ($i == $len)
1240                                 break;
1241
1242                         $start = $i;
1243
1244                         // Find the end of this run of changes.
1245                         while (++$i < $len && $changed[$i])
1246                                 continue;
1247
1248                         do {
1249                                 /*
1250                                  * Record the length of this run of changes, so that
1251                                  * we can later determine whether the run has grown.
1252                                  */
1253                                 $runlength = $i - $start;
1254
1255                                 /*
1256                                  * Move the changed region back, so long as the
1257                                  * previous unchanged line matches the last changed one.
1258                                  * This merges with previous changed regions.
1259                                  */
1260                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1261                                         $changed[--$start] = 1;
1262                                         $changed[--$i] = false;
1263                                         while ($start > 0 && $changed[$start - 1])
1264                                                 $start--;
1265                                         USE_ASSERTS && assert('$j > 0');
1266                                         while ($other_changed[--$j])
1267                                                 continue;
1268                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1269                                 }
1270
1271                                 /*
1272                                  * Set CORRESPONDING to the end of the changed run, at the last
1273                                  * point where it corresponds to a changed run in the other file.
1274                                  * CORRESPONDING == LEN means no such point has been found.
1275                                  */
1276                                 $corresponding = $j < $other_len ? $i : $len;
1277
1278                                 /*
1279                                  * Move the changed region forward, so long as the
1280                                  * first changed line matches the following unchanged one.
1281                                  * This merges with following changed regions.
1282                                  * Do this second, so that if there are no merges,
1283                                  * the changed region is moved forward as far as possible.
1284                                  */
1285                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1286                                         $changed[$start++] = false;
1287                                         $changed[$i++] = 1;
1288                                         while ($i < $len && $changed[$i])
1289                                                 $i++;
1290
1291                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1292                                         $j++;
1293                                         if ($j < $other_len && $other_changed[$j]) {
1294                                                 $corresponding = $i;
1295                                                 while ($j < $other_len && $other_changed[$j])
1296                                                         $j++;
1297                                         }
1298                                 }
1299                         } while ($runlength != $i - $start);
1300
1301                         /*
1302                          * If possible, move the fully-merged run of changes
1303                          * back to a corresponding run in the other file.
1304                          */
1305                         while ($corresponding < $i) {
1306                                 $changed[--$start] = 1;
1307                                 $changed[--$i] = 0;
1308                                 USE_ASSERTS && assert('$j > 0');
1309                                 while ($other_changed[--$j])
1310                                         continue;
1311                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1312                         }
1313                 }
1314                 wfProfileOut( $fname );
1315         }
1316 }
1317
1318 /**
1319  * Class representing a 'diff' between two sequences of strings.
1320  * @todo document
1321  * @private
1322  * @addtogroup DifferenceEngine
1323  */
1324 class Diff
1325 {
1326         var $edits;
1327
1328         /**
1329          * Constructor.
1330          * Computes diff between sequences of strings.
1331          *
1332          * @param $from_lines array An array of strings.
1333          *                (Typically these are lines from a file.)
1334          * @param $to_lines array An array of strings.
1335          */
1336         function Diff($from_lines, $to_lines) {
1337                 $eng = new _DiffEngine;
1338                 $this->edits = $eng->diff($from_lines, $to_lines);
1339                 //$this->_check($from_lines, $to_lines);
1340         }
1341
1342         /**
1343          * Compute reversed Diff.
1344          *
1345          * SYNOPSIS:
1346          *
1347          *      $diff = new Diff($lines1, $lines2);
1348          *      $rev = $diff->reverse();
1349          * @return object A Diff object representing the inverse of the
1350          *                                original diff.
1351          */
1352         function reverse () {
1353                 $rev = $this;
1354                 $rev->edits = array();
1355                 foreach ($this->edits as $edit) {
1356                         $rev->edits[] = $edit->reverse();
1357                 }
1358                 return $rev;
1359         }
1360
1361         /**
1362          * Check for empty diff.
1363          *
1364          * @return bool True iff two sequences were identical.
1365          */
1366         function isEmpty () {
1367                 foreach ($this->edits as $edit) {
1368                         if ($edit->type != 'copy')
1369                                 return false;
1370                 }
1371                 return true;
1372         }
1373
1374         /**
1375          * Compute the length of the Longest Common Subsequence (LCS).
1376          *
1377          * This is mostly for diagnostic purposed.
1378          *
1379          * @return int The length of the LCS.
1380          */
1381         function lcs () {
1382                 $lcs = 0;
1383                 foreach ($this->edits as $edit) {
1384                         if ($edit->type == 'copy')
1385                                 $lcs += sizeof($edit->orig);
1386                 }
1387                 return $lcs;
1388         }
1389
1390         /**
1391          * Get the original set of lines.
1392          *
1393          * This reconstructs the $from_lines parameter passed to the
1394          * constructor.
1395          *
1396          * @return array The original sequence of strings.
1397          */
1398         function orig() {
1399                 $lines = array();
1400
1401                 foreach ($this->edits as $edit) {
1402                         if ($edit->orig)
1403                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1404                 }
1405                 return $lines;
1406         }
1407
1408         /**
1409          * Get the closing set of lines.
1410          *
1411          * This reconstructs the $to_lines parameter passed to the
1412          * constructor.
1413          *
1414          * @return array The sequence of strings.
1415          */
1416         function closing() {
1417                 $lines = array();
1418
1419                 foreach ($this->edits as $edit) {
1420                         if ($edit->closing)
1421                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1422                 }
1423                 return $lines;
1424         }
1425
1426         /**
1427          * Check a Diff for validity.
1428          *
1429          * This is here only for debugging purposes.
1430          */
1431         function _check ($from_lines, $to_lines) {
1432                 $fname = 'Diff::_check';
1433                 wfProfileIn( $fname );
1434                 if (serialize($from_lines) != serialize($this->orig()))
1435                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1436                 if (serialize($to_lines) != serialize($this->closing()))
1437                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1438
1439                 $rev = $this->reverse();
1440                 if (serialize($to_lines) != serialize($rev->orig()))
1441                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1442                 if (serialize($from_lines) != serialize($rev->closing()))
1443                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1444
1445
1446                 $prevtype = 'none';
1447                 foreach ($this->edits as $edit) {
1448                         if ( $prevtype == $edit->type )
1449                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1450                         $prevtype = $edit->type;
1451                 }
1452
1453                 $lcs = $this->lcs();
1454                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1455                 wfProfileOut( $fname );
1456         }
1457 }
1458
1459 /**
1460  * @todo document, bad name.
1461  * @private
1462  * @addtogroup DifferenceEngine
1463  */
1464 class MappedDiff extends Diff
1465 {
1466         /**
1467          * Constructor.
1468          *
1469          * Computes diff between sequences of strings.
1470          *
1471          * This can be used to compute things like
1472          * case-insensitve diffs, or diffs which ignore
1473          * changes in white-space.
1474          *
1475          * @param $from_lines array An array of strings.
1476          *      (Typically these are lines from a file.)
1477          *
1478          * @param $to_lines array An array of strings.
1479          *
1480          * @param $mapped_from_lines array This array should
1481          *      have the same size number of elements as $from_lines.
1482          *      The elements in $mapped_from_lines and
1483          *      $mapped_to_lines are what is actually compared
1484          *      when computing the diff.
1485          *
1486          * @param $mapped_to_lines array This array should
1487          *      have the same number of elements as $to_lines.
1488          */
1489         function MappedDiff($from_lines, $to_lines,
1490                                                 $mapped_from_lines, $mapped_to_lines) {
1491                 $fname = 'MappedDiff::MappedDiff';
1492                 wfProfileIn( $fname );
1493
1494                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1495                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1496
1497                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1498
1499                 $xi = $yi = 0;
1500                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1501                         $orig = &$this->edits[$i]->orig;
1502                         if (is_array($orig)) {
1503                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1504                                 $xi += sizeof($orig);
1505                         }
1506
1507                         $closing = &$this->edits[$i]->closing;
1508                         if (is_array($closing)) {
1509                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1510                                 $yi += sizeof($closing);
1511                         }
1512                 }
1513                 wfProfileOut( $fname );
1514         }
1515 }
1516
1517 /**
1518  * A class to format Diffs
1519  *
1520  * This class formats the diff in classic diff format.
1521  * It is intended that this class be customized via inheritance,
1522  * to obtain fancier outputs.
1523  * @todo document
1524  * @private
1525  * @addtogroup DifferenceEngine
1526  */
1527 class DiffFormatter
1528 {
1529         /**
1530          * Number of leading context "lines" to preserve.
1531          *
1532          * This should be left at zero for this class, but subclasses
1533          * may want to set this to other values.
1534          */
1535         var $leading_context_lines = 0;
1536
1537         /**
1538          * Number of trailing context "lines" to preserve.
1539          *
1540          * This should be left at zero for this class, but subclasses
1541          * may want to set this to other values.
1542          */
1543         var $trailing_context_lines = 0;
1544
1545         /**
1546          * Format a diff.
1547          *
1548          * @param $diff object A Diff object.
1549          * @return string The formatted output.
1550          */
1551         function format($diff) {
1552                 $fname = 'DiffFormatter::format';
1553                 wfProfileIn( $fname );
1554
1555                 $xi = $yi = 1;
1556                 $block = false;
1557                 $context = array();
1558
1559                 $nlead = $this->leading_context_lines;
1560                 $ntrail = $this->trailing_context_lines;
1561
1562                 $this->_start_diff();
1563
1564                 foreach ($diff->edits as $edit) {
1565                         if ($edit->type == 'copy') {
1566                                 if (is_array($block)) {
1567                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1568                                                 $block[] = $edit;
1569                                         }
1570                                         else{
1571                                                 if ($ntrail) {
1572                                                         $context = array_slice($edit->orig, 0, $ntrail);
1573                                                         $block[] = new _DiffOp_Copy($context);
1574                                                 }
1575                                                 $this->_block($x0, $ntrail + $xi - $x0,
1576                                                                           $y0, $ntrail + $yi - $y0,
1577                                                                           $block);
1578                                                 $block = false;
1579                                         }
1580                                 }
1581                                 $context = $edit->orig;
1582                         }
1583                         else {
1584                                 if (! is_array($block)) {
1585                                         $context = array_slice($context, sizeof($context) - $nlead);
1586                                         $x0 = $xi - sizeof($context);
1587                                         $y0 = $yi - sizeof($context);
1588                                         $block = array();
1589                                         if ($context)
1590                                                 $block[] = new _DiffOp_Copy($context);
1591                                 }
1592                                 $block[] = $edit;
1593                         }
1594
1595                         if ($edit->orig)
1596                                 $xi += sizeof($edit->orig);
1597                         if ($edit->closing)
1598                                 $yi += sizeof($edit->closing);
1599                 }
1600
1601                 if (is_array($block))
1602                         $this->_block($x0, $xi - $x0,
1603                                                   $y0, $yi - $y0,
1604                                                   $block);
1605
1606                 $end = $this->_end_diff();
1607                 wfProfileOut( $fname );
1608                 return $end;
1609         }
1610
1611         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1612                 $fname = 'DiffFormatter::_block';
1613                 wfProfileIn( $fname );
1614                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1615                 foreach ($edits as $edit) {
1616                         if ($edit->type == 'copy')
1617                                 $this->_context($edit->orig);
1618                         elseif ($edit->type == 'add')
1619                                 $this->_added($edit->closing);
1620                         elseif ($edit->type == 'delete')
1621                                 $this->_deleted($edit->orig);
1622                         elseif ($edit->type == 'change')
1623                                 $this->_changed($edit->orig, $edit->closing);
1624                         else
1625                                 trigger_error('Unknown edit type', E_USER_ERROR);
1626                 }
1627                 $this->_end_block();
1628                 wfProfileOut( $fname );
1629         }
1630
1631         function _start_diff() {
1632                 ob_start();
1633         }
1634
1635         function _end_diff() {
1636                 $val = ob_get_contents();
1637                 ob_end_clean();
1638                 return $val;
1639         }
1640
1641         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1642                 if ($xlen > 1)
1643                         $xbeg .= "," . ($xbeg + $xlen - 1);
1644                 if ($ylen > 1)
1645                         $ybeg .= "," . ($ybeg + $ylen - 1);
1646
1647                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1648         }
1649
1650         function _start_block($header) {
1651                 echo $header . "\n";
1652         }
1653
1654         function _end_block() {
1655         }
1656
1657         function _lines($lines, $prefix = ' ') {
1658                 foreach ($lines as $line)
1659                         echo "$prefix $line\n";
1660         }
1661
1662         function _context($lines) {
1663                 $this->_lines($lines);
1664         }
1665
1666         function _added($lines) {
1667                 $this->_lines($lines, '>');
1668         }
1669         function _deleted($lines) {
1670                 $this->_lines($lines, '<');
1671         }
1672
1673         function _changed($orig, $closing) {
1674                 $this->_deleted($orig);
1675                 echo "---\n";
1676                 $this->_added($closing);
1677         }
1678 }
1679
1680 /**
1681  * A formatter that outputs unified diffs
1682  * @addtogroup DifferenceEngine
1683  */
1684
1685 class UnifiedDiffFormatter extends DiffFormatter
1686 {
1687         var $leading_context_lines = 2;
1688         var $trailing_context_lines = 2;
1689
1690         function _added($lines) {
1691                 $this->_lines($lines, '+');
1692         }
1693         function _deleted($lines) {
1694                 $this->_lines($lines, '-');
1695         }
1696         function _changed($orig, $closing) {
1697                 $this->_deleted($orig);
1698                 $this->_added($closing);
1699         }
1700         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1701                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1702         }
1703 }
1704
1705 /**
1706  * A pseudo-formatter that just passes along the Diff::$edits array
1707  * @addtogroup DifferenceEngine
1708  */
1709 class ArrayDiffFormatter extends DiffFormatter
1710 {
1711         function format($diff)
1712         {
1713                 $oldline = 1;
1714                 $newline = 1;
1715                 $retval = array();
1716                 foreach($diff->edits as $edit)
1717                         switch($edit->type)
1718                         {
1719                                 case 'add':
1720                                         foreach($edit->closing as $l)
1721                                         {
1722                                                 $retval[] = array(
1723                                                         'action' => 'add',
1724                                                         'new'=> $l,
1725                                                         'newline' => $newline++
1726                                                 );
1727                                         }
1728                                         break;
1729                                 case 'delete':
1730                                         foreach($edit->orig as $l)
1731                                         {
1732                                                 $retval[] = array(
1733                                                         'action' => 'delete',
1734                                                         'old' => $l,
1735                                                         'oldline' => $oldline++,
1736                                                 );
1737                                         }
1738                                         break;
1739                                 case 'change':
1740                                         foreach($edit->orig as $i => $l)
1741                                         {
1742                                                 $retval[] = array(
1743                                                         'action' => 'change',
1744                                                         'old' => $l,
1745                                                         'new' => @$edit->closing[$i],
1746                                                         'oldline' => $oldline++,
1747                                                         'newline' => $newline++,
1748                                                 );
1749                                         }
1750                                         break;
1751                                 case 'copy':
1752                                         $oldline += count($edit->orig);
1753                                         $newline += count($edit->orig);
1754                         }
1755                 return $retval;
1756         }
1757 }
1758
1759 /**
1760  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1761  *
1762  */
1763
1764 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1765
1766 /**
1767  * @todo document
1768  * @private
1769  * @addtogroup DifferenceEngine
1770  */
1771 class _HWLDF_WordAccumulator {
1772         function _HWLDF_WordAccumulator () {
1773                 $this->_lines = array();
1774                 $this->_line = '';
1775                 $this->_group = '';
1776                 $this->_tag = '';
1777         }
1778
1779         function _flushGroup ($new_tag) {
1780                 if ($this->_group !== '') {
1781                         if ($this->_tag == 'ins')
1782                                 $this->_line .= '<ins class="diffchange">' .
1783                                         htmlspecialchars ( $this->_group ) . '</ins>';
1784                         elseif ($this->_tag == 'del')
1785                                 $this->_line .= '<del class="diffchange">' .
1786                                         htmlspecialchars ( $this->_group ) . '</del>';
1787                         else
1788                                 $this->_line .= htmlspecialchars ( $this->_group );
1789                 }
1790                 $this->_group = '';
1791                 $this->_tag = $new_tag;
1792         }
1793
1794         function _flushLine ($new_tag) {
1795                 $this->_flushGroup($new_tag);
1796                 if ($this->_line != '')
1797                         array_push ( $this->_lines, $this->_line );
1798                 else
1799                         # make empty lines visible by inserting an NBSP
1800                         array_push ( $this->_lines, NBSP );
1801                 $this->_line = '';
1802         }
1803
1804         function addWords ($words, $tag = '') {
1805                 if ($tag != $this->_tag)
1806                         $this->_flushGroup($tag);
1807
1808                 foreach ($words as $word) {
1809                         // new-line should only come as first char of word.
1810                         if ($word == '')
1811                                 continue;
1812                         if ($word[0] == "\n") {
1813                                 $this->_flushLine($tag);
1814                                 $word = substr($word, 1);
1815                         }
1816                         assert(!strstr($word, "\n"));
1817                         $this->_group .= $word;
1818                 }
1819         }
1820
1821         function getLines() {
1822                 $this->_flushLine('~done');
1823                 return $this->_lines;
1824         }
1825 }
1826
1827 /**
1828  * @todo document
1829  * @private
1830  * @addtogroup DifferenceEngine
1831  */
1832 class WordLevelDiff extends MappedDiff
1833 {
1834         const MAX_LINE_LENGTH = 10000;
1835
1836         function WordLevelDiff ($orig_lines, $closing_lines) {
1837                 $fname = 'WordLevelDiff::WordLevelDiff';
1838                 wfProfileIn( $fname );
1839
1840                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1841                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1842
1843                 $this->MappedDiff($orig_words, $closing_words,
1844                                                   $orig_stripped, $closing_stripped);
1845                 wfProfileOut( $fname );
1846         }
1847
1848         function _split($lines) {
1849                 $fname = 'WordLevelDiff::_split';
1850                 wfProfileIn( $fname );
1851
1852                 $words = array();
1853                 $stripped = array();
1854                 $first = true;
1855                 foreach ( $lines as $line ) {
1856                         # If the line is too long, just pretend the entire line is one big word
1857                         # This prevents resource exhaustion problems
1858                         if ( $first ) {
1859                                 $first = false;
1860                         } else {
1861                                 $words[] = "\n";
1862                                 $stripped[] = "\n";
1863                         }
1864                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1865                                 $words[] = $line;
1866                                 $stripped[] = $line;
1867                         } else {
1868                                 $m = array();
1869                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1870                                         $line, $m))
1871                                 {
1872                                         $words = array_merge( $words, $m[0] );
1873                                         $stripped = array_merge( $stripped, $m[1] );
1874                                 }
1875                         }
1876                 }
1877                 wfProfileOut( $fname );
1878                 return array($words, $stripped);
1879         }
1880
1881         function orig () {
1882                 $fname = 'WordLevelDiff::orig';
1883                 wfProfileIn( $fname );
1884                 $orig = new _HWLDF_WordAccumulator;
1885
1886                 foreach ($this->edits as $edit) {
1887                         if ($edit->type == 'copy')
1888                                 $orig->addWords($edit->orig);
1889                         elseif ($edit->orig)
1890                                 $orig->addWords($edit->orig, 'del');
1891                 }
1892                 $lines = $orig->getLines();
1893                 wfProfileOut( $fname );
1894                 return $lines;
1895         }
1896
1897         function closing () {
1898                 $fname = 'WordLevelDiff::closing';
1899                 wfProfileIn( $fname );
1900                 $closing = new _HWLDF_WordAccumulator;
1901
1902                 foreach ($this->edits as $edit) {
1903                         if ($edit->type == 'copy')
1904                                 $closing->addWords($edit->closing);
1905                         elseif ($edit->closing)
1906                                 $closing->addWords($edit->closing, 'ins');
1907                 }
1908                 $lines = $closing->getLines();
1909                 wfProfileOut( $fname );
1910                 return $lines;
1911         }
1912 }
1913
1914 /**
1915  *      Wikipedia Table style diff formatter.
1916  * @todo document
1917  * @private
1918  * @addtogroup DifferenceEngine
1919  */
1920 class TableDiffFormatter extends DiffFormatter
1921 {
1922         function TableDiffFormatter() {
1923                 $this->leading_context_lines = 2;
1924                 $this->trailing_context_lines = 2;
1925         }
1926
1927         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1928                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1929                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1930                 return $r;
1931         }
1932
1933         function _start_block( $header ) {
1934                 echo $header;
1935         }
1936
1937         function _end_block() {
1938         }
1939
1940         function _lines( $lines, $prefix=' ', $color='white' ) {
1941         }
1942
1943         # HTML-escape parameter before calling this
1944         function addedLine( $line ) {
1945                 return $this->wrapLine( '+', 'diff-addedline', $line );
1946         }
1947
1948         # HTML-escape parameter before calling this
1949         function deletedLine( $line ) {
1950                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1951         }
1952
1953         # HTML-escape parameter before calling this
1954         function contextLine( $line ) {
1955                 return $this->wrapLine( ' ', 'diff-context', $line );
1956         }
1957
1958         private function wrapLine( $marker, $class, $line ) {
1959                 if( $line !== '' ) {
1960                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1961                         $line = "<div>$line</div>";
1962                 }
1963                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1964         }
1965
1966         function emptyLine() {
1967                 return '<td colspan="2">&nbsp;</td>';
1968         }
1969
1970         function _added( $lines ) {
1971                 foreach ($lines as $line) {
1972                         echo '<tr>' . $this->emptyLine() .
1973                                 $this->addedLine( '<ins class="diffchange">' .
1974                                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1975                 }
1976         }
1977
1978         function _deleted($lines) {
1979                 foreach ($lines as $line) {
1980                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1981                                 htmlspecialchars ( $line ) . '</del>' ) .
1982                           $this->emptyLine() . "</tr>\n";
1983                 }
1984         }
1985
1986         function _context( $lines ) {
1987                 foreach ($lines as $line) {
1988                         echo '<tr>' .
1989                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1990                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1991                 }
1992         }
1993
1994         function _changed( $orig, $closing ) {
1995                 $fname = 'TableDiffFormatter::_changed';
1996                 wfProfileIn( $fname );
1997
1998                 $diff = new WordLevelDiff( $orig, $closing );
1999                 $del = $diff->orig();
2000                 $add = $diff->closing();
2001
2002                 # Notice that WordLevelDiff returns HTML-escaped output.
2003                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2004
2005                 while ( $line = array_shift( $del ) ) {
2006                         $aline = array_shift( $add );
2007                         echo '<tr>' . $this->deletedLine( $line ) .
2008                                 $this->addedLine( $aline ) . "</tr>\n";
2009                 }
2010                 foreach ($add as $line) {       # If any leftovers
2011                         echo '<tr>' . $this->emptyLine() .
2012                                 $this->addedLine( $line ) . "</tr>\n";
2013                 }
2014                 wfProfileOut( $fname );
2015         }
2016 }
2017
2018
2019