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