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() . ' ' . wfMsgExt( 'missingarticle-diff', array( 'escape' ), $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() . ' ' . wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 330                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 331                         $wgOut->addWikiMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 332                         wfProfileOut( __METHOD__ );
 333                         return;
 334                 }
 335                 if ( $this->mNewRev->isCurrent() ) {
 336                         $wgOut->setArticleFlag( true );
 337                 }
 338
 339                 # Check if user is allowed to look at this page. If not, bail out.
 340                 #
 341                 if ( !( $this->mTitle->userCanRead() ) ) {
 342                         $wgOut->loginToUse();
 343                         $wgOut->output();
 344                         wfProfileOut( __METHOD__ );
 345                         exit;
 346                 }
 347
 348                 # Prepare the header box
 349                 #
 350                 $sk = $wgUser->getSkin();
 351
 352                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 353                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 354                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 355                         $sk->revComment( $this->mNewRev ) . "<br />" .
 356                         $nextlink . "</div>\n";
 357
 358                 $wgOut->addHTML( $header );
 359
 360                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 361                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 362
 363                 wfProfileOut( __METHOD__ );
 364         }
 365
 366         /**
 367          * Get the diff text, send it to $wgOut
 368          * Returns false if the diff could not be generated, otherwise returns true
 369          */
 370         function showDiff( $otitle, $ntitle ) {
 371                 global $wgOut;
 372                 $diff = $this->getDiff( $otitle, $ntitle );
 373                 if ( $diff === false ) {
 374                         $wgOut->addWikiMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" );
 375                         return false;
 376                 } else {
 377                         $this->showDiffStyle();
 378                         $wgOut->addHTML( $diff );
 379                         return true;
 380                 }
 381         }
 382
 383         /**
 384          * Add style sheets and supporting JS for diff display.
 385          */
 386         function showDiffStyle() {
 387                 global $wgStylePath, $wgStyleVersion, $wgOut;
 388                 $wgOut->addStyle( 'common/diff.css' );
 389
 390                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 391                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 392         }
 393
 394         /**
 395          * Get complete diff table, including header
 396          *
 397          * @param Title $otitle Old title
 398          * @param Title $ntitle New title
 399          * @return mixed
 400          */
 401         function getDiff( $otitle, $ntitle ) {
 402                 $body = $this->getDiffBody();
 403                 if ( $body === false ) {
 404                         return false;
 405                 } else {
 406                         $multi = $this->getMultiNotice();
 407                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 408                 }
 409         }
 410
 411         /**
 412          * Get the diff table body, without header
 413          *
 414          * @return mixed
 415          */
 416         function getDiffBody() {
 417                 global $wgMemc;
 418                 wfProfileIn( __METHOD__ );
 419                 // Check if the diff should be hidden from this user
 420                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 421                         return '';
 422                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 423                         return '';
 424                 }
 425                 // Cacheable?
 426                 $key = false;
 427                 if ( $this->mOldid && $this->mNewid ) {
 428                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 429                         // Try cache
 430                         if ( !$this->mRefreshCache ) {
 431                                 $difftext = $wgMemc->get( $key );
 432                                 if ( $difftext ) {
 433                                         wfIncrStats( 'diff_cache_hit' );
 434                                         $difftext = $this->localiseLineNumbers( $difftext );
 435                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 436                                         wfProfileOut( __METHOD__ );
 437                                         return $difftext;
 438                                 }
 439                         } // don't try to load but save the result
 440                 }
 441
 442                 // Loadtext is permission safe, this just clears out the diff
 443                 if ( !$this->loadText() ) {
 444                         wfProfileOut( __METHOD__ );
 445                         return false;
 446                 }
 447
 448                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 449
 450                 // Save to cache for 7 days
 451                 if ( $key !== false && $difftext !== false ) {
 452                         wfIncrStats( 'diff_cache_miss' );
 453                         $wgMemc->set( $key, $difftext, 7*86400 );
 454                 } else {
 455                         wfIncrStats( 'diff_uncacheable' );
 456                 }
 457                 // Replace line numbers with the text in the user's language
 458                 if ( $difftext !== false ) {
 459                         $difftext = $this->localiseLineNumbers( $difftext );
 460                 }
 461                 wfProfileOut( __METHOD__ );
 462                 return $difftext;
 463         }
 464
 465         /**
 466          * Generate a diff, no caching
 467          * $otext and $ntext must be already segmented
 468          */
 469         function generateDiffBody( $otext, $ntext ) {
 470                 global $wgExternalDiffEngine, $wgContLang;
 471
 472                 $otext = str_replace( "\r\n", "\n", $otext );
 473                 $ntext = str_replace( "\r\n", "\n", $ntext );
 474
 475                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 476                         # For historical reasons, external diff engine expects
 477                         # input text to be HTML-escaped already
 478                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 479                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 480                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 481                                 dl('php_wikidiff.so');
 482                         }
 483                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 484                                 $this->debug( 'wikidiff1' );
 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                                 $text .= $this->debug( 'wikidiff2' );
 499                                 wfProfileOut( 'wikidiff2_do_diff' );
 500                                 return $text;
 501                         }
 502                 }
 503                 if ( $wgExternalDiffEngine !== false ) {
 504                         # Diff via the shell
 505                         global $wgTmpDirectory;
 506                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 507                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 508
 509                         $tempFile1 = fopen( $tempName1, "w" );
 510                         if ( !$tempFile1 ) {
 511                                 wfProfileOut( __METHOD__ );
 512                                 return false;
 513                         }
 514                         $tempFile2 = fopen( $tempName2, "w" );
 515                         if ( !$tempFile2 ) {
 516                                 wfProfileOut( __METHOD__ );
 517                                 return false;
 518                         }
 519                         fwrite( $tempFile1, $otext );
 520                         fwrite( $tempFile2, $ntext );
 521                         fclose( $tempFile1 );
 522                         fclose( $tempFile2 );
 523                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 524                         wfProfileIn( __METHOD__ . "-shellexec" );
 525                         $difftext = wfShellExec( $cmd );
 526                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 527                         wfProfileOut( __METHOD__ . "-shellexec" );
 528                         unlink( $tempName1 );
 529                         unlink( $tempName2 );
 530                         return $difftext;
 531                 }
 532
 533                 # Native PHP diff
 534                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 535                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 536                 $diffs = new Diff( $ota, $nta );
 537                 $formatter = new TableDiffFormatter();
 538                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 539                         $this->debug();
 540         }
 541
 542         /**
 543          * Generate a debug comment indicating diff generating time,
 544          * server node, and generator backend.
 545          */
 546         protected function debug( $generator="internal" ) {
 547                 global $wgShowHostnames, $wgNodeName;
 548                 $data = array( $generator );
 549                 if( $wgShowHostnames ) {
 550                         $data[] = $wgNodeName;
 551                 }
 552                 $data[] = wfTimestamp( TS_DB );
 553                 return "<!-- diff generator: " .
 554                         implode( " ",
 555                                 array_map(
 556                                         "htmlspecialchars",
 557                                                 $data ) ) .
 558                         " -->\n";
 559         }
 560
 561         /**
 562          * Replace line numbers with the text in the user's language
 563          */
 564         function localiseLineNumbers( $text ) {
 565                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 566                         array( &$this, 'localiseLineNumbersCb' ), $text );
 567         }
 568
 569         function localiseLineNumbersCb( $matches ) {
 570                 global $wgLang;
 571                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 572         }
 573
 574
 575         /**
 576          * If there are revisions between the ones being compared, return a note saying so.
 577          */
 578         function getMultiNotice() {
 579                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 580                         return '';
 581
 582                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 583                         // Comparing two different pages? Count would be meaningless.
 584                         return '';
 585                 }
 586
 587                 $oldid = $this->mOldRev->getId();
 588                 $newid = $this->mNewRev->getId();
 589                 if ( $oldid > $newid ) {
 590                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 591                 }
 592
 593                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 594                 if ( !$n )
 595                         return '';
 596
 597                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 598         }
 599
 600
 601         /**
 602          * Add the header to a diff body
 603          */
 604         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 605                 global $wgOut;
 606
 607                 $header = "
 608                         <table class='diff'>
 609                         <col class='diff-marker' />
 610                         <col class='diff-content' />
 611                         <col class='diff-marker' />
 612                         <col class='diff-content' />
 613                         <tr valign='top'>
 614                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 615                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 616                         </tr>
 617                 ";
 618
 619                 if ( $multi != '' )
 620                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 621
 622                 return $header . $diff . "</table>";
 623         }
 624
 625         /**
 626          * Use specified text instead of loading from the database
 627          */
 628         function setText( $oldText, $newText ) {
 629                 $this->mOldtext = $oldText;
 630                 $this->mNewtext = $newText;
 631                 $this->mTextLoaded = 2;
 632         }
 633
 634         /**
 635          * Load revision metadata for the specified articles. If newid is 0, then compare
 636          * the old article in oldid to the current article; if oldid is 0, then
 637          * compare the current article to the immediately previous one (ignoring the
 638          * value of newid).
 639          *
 640          * If oldid is false, leave the corresponding revision object set
 641          * to false. This is impossible via ordinary user input, and is provided for
 642          * API convenience.
 643          */
 644         function loadRevisionData() {
 645                 global $wgLang;
 646                 if ( $this->mRevisionsLoaded ) {
 647                         return true;
 648                 } else {
 649                         // Whether it succeeds or fails, we don't want to try again
 650                         $this->mRevisionsLoaded = true;
 651                 }
 652
 653                 // Load the new revision object
 654                 $this->mNewRev = $this->mNewid
 655                         ? Revision::newFromId( $this->mNewid )
 656                         : Revision::newFromTitle( $this->mTitle );
 657                 if( !$this->mNewRev instanceof Revision )
 658                         return false;
 659
 660                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 661                 $this->mNewid = $this->mNewRev->getId();
 662
 663                 // Check if page is editable
 664                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 665
 666                 // Set assorted variables
 667                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 668                 $this->mNewPage = $this->mNewRev->getTitle();
 669                 if( $this->mNewRev->isCurrent() ) {
 670                         $newLink = $this->mNewPage->escapeLocalUrl();
 671                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 672                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 673
 674                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)";
 675                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 676
 677                 } else {
 678                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 679                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 680                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 681
 682                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 683                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 684                 }
 685                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 686                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 687                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 688                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 689                 }
 690
 691                 // Load the old revision object
 692                 $this->mOldRev = false;
 693                 if( $this->mOldid ) {
 694                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 695                 } elseif ( $this->mOldid === 0 ) {
 696                         $rev = $this->mNewRev->getPrevious();
 697                         if( $rev ) {
 698                                 $this->mOldid = $rev->getId();
 699                                 $this->mOldRev = $rev;
 700                         } else {
 701                                 // No previous revision; mark to show as first-version only.
 702                                 $this->mOldid = false;
 703                                 $this->mOldRev = false;
 704                         }
 705                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 706
 707                 if( is_null( $this->mOldRev ) ) {
 708                         return false;
 709                 }
 710
 711                 if ( $this->mOldRev ) {
 712                         $this->mOldPage = $this->mOldRev->getTitle();
 713
 714                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 715                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 716                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 717                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 718
 719                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 720                                 . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 721                         // Add an "undo" link
 722                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 723                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 724                                 $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 725                         }
 726
 727                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 728                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 729                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 730                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 731                         }
 732                 }
 733
 734                 return true;
 735         }
 736
 737         /**
 738          * Load the text of the revisions, as well as revision data.
 739          */
 740         function loadText() {
 741                 if ( $this->mTextLoaded == 2 ) {
 742                         return true;
 743                 } else {
 744                         // Whether it succeeds or fails, we don't want to try again
 745                         $this->mTextLoaded = 2;
 746                 }
 747
 748                 if ( !$this->loadRevisionData() ) {
 749                         return false;
 750                 }
 751                 if ( $this->mOldRev ) {
 752                         $this->mOldtext = $this->mOldRev->revText();
 753                         if ( $this->mOldtext === false ) {
 754                                 return false;
 755                         }
 756                 }
 757                 if ( $this->mNewRev ) {
 758                         $this->mNewtext = $this->mNewRev->revText();
 759                         if ( $this->mNewtext === false ) {
 760                                 return false;
 761                         }
 762                 }
 763                 return true;
 764         }
 765
 766         /**
 767          * Load the text of the new revision, not the old one
 768          */
 769         function loadNewText() {
 770                 if ( $this->mTextLoaded >= 1 ) {
 771                         return true;
 772                 } else {
 773                         $this->mTextLoaded = 1;
 774                 }
 775                 if ( !$this->loadRevisionData() ) {
 776                         return false;
 777                 }
 778                 $this->mNewtext = $this->mNewRev->getText();
 779                 return true;
 780         }
 781
 782
 783 }
 784
 785 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 786 //
 787 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 788 // You may copy this code freely under the conditions of the GPL.
 789 //
 790
 791 define('USE_ASSERTS', function_exists('assert'));
 792
 793 /**
 794  * @todo document
 795  * @private
 796  * @addtogroup DifferenceEngine
 797  */
 798 class _DiffOp {
 799         var $type;
 800         var $orig;
 801         var $closing;
 802
 803         function reverse() {
 804                 trigger_error('pure virtual', E_USER_ERROR);
 805         }
 806
 807         function norig() {
 808                 return $this->orig ? sizeof($this->orig) : 0;
 809         }
 810
 811         function nclosing() {
 812                 return $this->closing ? sizeof($this->closing) : 0;
 813         }
 814 }
 815
 816 /**
 817  * @todo document
 818  * @private
 819  * @addtogroup DifferenceEngine
 820  */
 821 class _DiffOp_Copy extends _DiffOp {
 822         var $type = 'copy';
 823
 824         function _DiffOp_Copy ($orig, $closing = false) {
 825                 if (!is_array($closing))
 826                         $closing = $orig;
 827                 $this->orig = $orig;
 828                 $this->closing = $closing;
 829         }
 830
 831         function reverse() {
 832                 return new _DiffOp_Copy($this->closing, $this->orig);
 833         }
 834 }
 835
 836 /**
 837  * @todo document
 838  * @private
 839  * @addtogroup DifferenceEngine
 840  */
 841 class _DiffOp_Delete extends _DiffOp {
 842         var $type = 'delete';
 843
 844         function _DiffOp_Delete ($lines) {
 845                 $this->orig = $lines;
 846                 $this->closing = false;
 847         }
 848
 849         function reverse() {
 850                 return new _DiffOp_Add($this->orig);
 851         }
 852 }
 853
 854 /**
 855  * @todo document
 856  * @private
 857  * @addtogroup DifferenceEngine
 858  */
 859 class _DiffOp_Add extends _DiffOp {
 860         var $type = 'add';
 861
 862         function _DiffOp_Add ($lines) {
 863                 $this->closing = $lines;
 864                 $this->orig = false;
 865         }
 866
 867         function reverse() {
 868                 return new _DiffOp_Delete($this->closing);
 869         }
 870 }
 871
 872 /**
 873  * @todo document
 874  * @private
 875  * @addtogroup DifferenceEngine
 876  */
 877 class _DiffOp_Change extends _DiffOp {
 878         var $type = 'change';
 879
 880         function _DiffOp_Change ($orig, $closing) {
 881                 $this->orig = $orig;
 882                 $this->closing = $closing;
 883         }
 884
 885         function reverse() {
 886                 return new _DiffOp_Change($this->closing, $this->orig);
 887         }
 888 }
 889
 890
 891 /**
 892  * Class used internally by Diff to actually compute the diffs.
 893  *
 894  * The algorithm used here is mostly lifted from the perl module
 895  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 896  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 897  *
 898  * More ideas are taken from:
 899  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 900  *
 901  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 902  * diffutils-2.7, which can be found at:
 903  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 904  *
 905  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 906  * are my own.
 907  *
 908  * Line length limits for robustness added by Tim Starling, 2005-08-31
 909  *
 910  * @author Geoffrey T. Dairiki, Tim Starling
 911  * @private
 912  * @addtogroup DifferenceEngine
 913  */
 914 class _DiffEngine {
 915         const MAX_XREF_LENGTH =  10000;
 916
 917         function diff ($from_lines, $to_lines) {
 918                 wfProfileIn( __METHOD__ );
 919
 920                 $n_from = sizeof($from_lines);
 921                 $n_to = sizeof($to_lines);
 922
 923                 $this->xchanged = $this->ychanged = array();
 924                 $this->xv = $this->yv = array();
 925                 $this->xind = $this->yind = array();
 926                 unset($this->seq);
 927                 unset($this->in_seq);
 928                 unset($this->lcs);
 929
 930                 // Skip leading common lines.
 931                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 932                         if ($from_lines[$skip] !== $to_lines[$skip])
 933                                 break;
 934                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 935                 }
 936                 // Skip trailing common lines.
 937                 $xi = $n_from; $yi = $n_to;
 938                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 939                         if ($from_lines[$xi] !== $to_lines[$yi])
 940                                 break;
 941                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 942                 }
 943
 944                 // Ignore lines which do not exist in both files.
 945                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 946                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 947                 }
 948
 949                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 950                         $line = $to_lines[$yi];
 951                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 952                                 continue;
 953                         $yhash[$this->_line_hash($line)] = 1;
 954                         $this->yv[] = $line;
 955                         $this->yind[] = $yi;
 956                 }
 957                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 958                         $line = $from_lines[$xi];
 959                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 960                                 continue;
 961                         $this->xv[] = $line;
 962                         $this->xind[] = $xi;
 963                 }
 964
 965                 // Find the LCS.
 966                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 967
 968                 // Merge edits when possible
 969                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 970                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 971
 972                 // Compute the edit operations.
 973                 $edits = array();
 974                 $xi = $yi = 0;
 975                 while ($xi < $n_from || $yi < $n_to) {
 976                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 977                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 978
 979                         // Skip matching "snake".
 980                         $copy = array();
 981                         while ( $xi < $n_from && $yi < $n_to
 982                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 983                                 $copy[] = $from_lines[$xi++];
 984                                 ++$yi;
 985                         }
 986                         if ($copy)
 987                                 $edits[] = new _DiffOp_Copy($copy);
 988
 989                         // Find deletes & adds.
 990                         $delete = array();
 991                         while ($xi < $n_from && $this->xchanged[$xi])
 992                                 $delete[] = $from_lines[$xi++];
 993
 994                         $add = array();
 995                         while ($yi < $n_to && $this->ychanged[$yi])
 996                                 $add[] = $to_lines[$yi++];
 997
 998                         if ($delete && $add)
 999                                 $edits[] = new _DiffOp_Change($delete, $add);
1000                         elseif ($delete)
1001                                 $edits[] = new _DiffOp_Delete($delete);
1002                         elseif ($add)
1003                                 $edits[] = new _DiffOp_Add($add);
1004                 }
1005                 wfProfileOut( __METHOD__ );
1006                 return $edits;
1007         }
1008
1009         /**
1010          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1011          */
1012         function _line_hash( $line ) {
1013                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1014                         return md5( $line );
1015                 } else {
1016                         return $line;
1017                 }
1018         }
1019
1020
1021         /* Divide the Largest Common Subsequence (LCS) of the sequences
1022          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1023          * sized segments.
1024          *
1025          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1026          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1027          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1028          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1029          * that (X0, Y0) == (XOFF, YOFF) and
1030          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1031          *
1032          * This function assumes that the first lines of the specified portions
1033          * of the two files do not match, and likewise that the last lines do not
1034          * match.  The caller must trim matching lines from the beginning and end
1035          * of the portions it is going to specify.
1036          */
1037         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1038                 wfProfileIn( __METHOD__ );
1039                 $flip = false;
1040
1041                 if ($xlim - $xoff > $ylim - $yoff) {
1042                         // Things seems faster (I'm not sure I understand why)
1043                                 // when the shortest sequence in X.
1044                                 $flip = true;
1045                         list ($xoff, $xlim, $yoff, $ylim)
1046                         = array( $yoff, $ylim, $xoff, $xlim);
1047                 }
1048
1049                 if ($flip)
1050                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1051                                 $ymatches[$this->xv[$i]][] = $i;
1052                 else
1053                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1054                                 $ymatches[$this->yv[$i]][] = $i;
1055
1056                 $this->lcs = 0;
1057                 $this->seq[0]= $yoff - 1;
1058                 $this->in_seq = array();
1059                 $ymids[0] = array();
1060
1061                 $numer = $xlim - $xoff + $nchunks - 1;
1062                 $x = $xoff;
1063                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1064                         wfProfileIn( __METHOD__ . "-chunk" );
1065                         if ($chunk > 0)
1066                                 for ($i = 0; $i <= $this->lcs; $i++)
1067                                         $ymids[$i][$chunk-1] = $this->seq[$i];
1068
1069                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1070                         for ( ; $x < $x1; $x++) {
1071                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1072                                         if (empty($ymatches[$line]))
1073                                                 continue;
1074                                 $matches = $ymatches[$line];
1075                                 reset($matches);
1076                                 while (list ($junk, $y) = each($matches))
1077                                         if (empty($this->in_seq[$y])) {
1078                                                 $k = $this->_lcs_pos($y);
1079                                                 USE_ASSERTS && assert($k > 0);
1080                                                 $ymids[$k] = $ymids[$k-1];
1081                                                 break;
1082                                         }
1083                                 while (list ( /* $junk */, $y) = each($matches)) {
1084                                         if ($y > $this->seq[$k-1]) {
1085                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1086                                                 // Optimization: this is a common case:
1087                                                 //      next match is just replacing previous match.
1088                                                 $this->in_seq[$this->seq[$k]] = false;
1089                                                 $this->seq[$k] = $y;
1090                                                 $this->in_seq[$y] = 1;
1091                                         } else if (empty($this->in_seq[$y])) {
1092                                                 $k = $this->_lcs_pos($y);
1093                                                 USE_ASSERTS && assert($k > 0);
1094                                                 $ymids[$k] = $ymids[$k-1];
1095                                         }
1096                                 }
1097                         }
1098                         wfProfileOut( __METHOD__ . "-chunk" );
1099                 }
1100
1101                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1102                 $ymid = $ymids[$this->lcs];
1103                 for ($n = 0; $n < $nchunks - 1; $n++) {
1104                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1105                         $y1 = $ymid[$n] + 1;
1106                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1107                 }
1108                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1109
1110                 wfProfileOut( __METHOD__ );
1111                 return array($this->lcs, $seps);
1112         }
1113
1114         function _lcs_pos ($ypos) {
1115                 wfProfileIn( __METHOD__ );
1116
1117                 $end = $this->lcs;
1118                 if ($end == 0 || $ypos > $this->seq[$end]) {
1119                         $this->seq[++$this->lcs] = $ypos;
1120                         $this->in_seq[$ypos] = 1;
1121                         wfProfileOut( __METHOD__ );
1122                         return $this->lcs;
1123                 }
1124
1125                 $beg = 1;
1126                 while ($beg < $end) {
1127                         $mid = (int)(($beg + $end) / 2);
1128                         if ( $ypos > $this->seq[$mid] )
1129                                 $beg = $mid + 1;
1130                         else
1131                                 $end = $mid;
1132                 }
1133
1134                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1135
1136                 $this->in_seq[$this->seq[$end]] = false;
1137                 $this->seq[$end] = $ypos;
1138                 $this->in_seq[$ypos] = 1;
1139                 wfProfileOut( __METHOD__ );
1140                 return $end;
1141         }
1142
1143         /* Find LCS of two sequences.
1144          *
1145          * The results are recorded in the vectors $this->{x,y}changed[], by
1146          * storing a 1 in the element for each line that is an insertion
1147          * or deletion (ie. is not in the LCS).
1148          *
1149          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1150          *
1151          * Note that XLIM, YLIM are exclusive bounds.
1152          * All line numbers are origin-0 and discarded lines are not counted.
1153          */
1154         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1155                 wfProfileIn( __METHOD__ );
1156
1157                 // Slide down the bottom initial diagonal.
1158                 while ($xoff < $xlim && $yoff < $ylim
1159                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1160                         ++$xoff;
1161                         ++$yoff;
1162                 }
1163
1164                 // Slide up the top initial diagonal.
1165                 while ($xlim > $xoff && $ylim > $yoff
1166                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1167                         --$xlim;
1168                         --$ylim;
1169                 }
1170
1171                 if ($xoff == $xlim || $yoff == $ylim)
1172                         $lcs = 0;
1173                 else {
1174                         // This is ad hoc but seems to work well.
1175                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1176                         //$nchunks = max(2,min(8,(int)$nchunks));
1177                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1178                         list ($lcs, $seps)
1179                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1180                 }
1181
1182                 if ($lcs == 0) {
1183                         // X and Y sequences have no common subsequence:
1184                         // mark all changed.
1185                         while ($yoff < $ylim)
1186                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1187                         while ($xoff < $xlim)
1188                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1189                 } else {
1190                         // Use the partitions to split this problem into subproblems.
1191                         reset($seps);
1192                         $pt1 = $seps[0];
1193                         while ($pt2 = next($seps)) {
1194                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1195                                 $pt1 = $pt2;
1196                         }
1197                 }
1198                 wfProfileOut( __METHOD__ );
1199         }
1200
1201         /* Adjust inserts/deletes of identical lines to join changes
1202          * as much as possible.
1203          *
1204          * We do something when a run of changed lines include a
1205          * line at one end and has an excluded, identical line at the other.
1206          * We are free to choose which identical line is included.
1207          * `compareseq' usually chooses the one at the beginning,
1208          * but usually it is cleaner to consider the following identical line
1209          * to be the "change".
1210          *
1211          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1212          */
1213         function _shift_boundaries ($lines, &$changed, $other_changed) {
1214                 wfProfileIn( __METHOD__ );
1215                 $i = 0;
1216                 $j = 0;
1217
1218                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1219                 $len = sizeof($lines);
1220                 $other_len = sizeof($other_changed);
1221
1222                 while (1) {
1223                         /*
1224                          * Scan forwards to find beginning of another run of changes.
1225                          * Also keep track of the corresponding point in the other file.
1226                          *
1227                          * Throughout this code, $i and $j are adjusted together so that
1228                          * the first $i elements of $changed and the first $j elements
1229                          * of $other_changed both contain the same number of zeros
1230                          * (unchanged lines).
1231                          * Furthermore, $j is always kept so that $j == $other_len or
1232                          * $other_changed[$j] == false.
1233                          */
1234                         while ($j < $other_len && $other_changed[$j])
1235                                 $j++;
1236
1237                         while ($i < $len && ! $changed[$i]) {
1238                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1239                                 $i++; $j++;
1240                                 while ($j < $other_len && $other_changed[$j])
1241                                         $j++;
1242                         }
1243
1244                         if ($i == $len)
1245                                 break;
1246
1247                         $start = $i;
1248
1249                         // Find the end of this run of changes.
1250                         while (++$i < $len && $changed[$i])
1251                                 continue;
1252
1253                         do {
1254                                 /*
1255                                  * Record the length of this run of changes, so that
1256                                  * we can later determine whether the run has grown.
1257                                  */
1258                                 $runlength = $i - $start;
1259
1260                                 /*
1261                                  * Move the changed region back, so long as the
1262                                  * previous unchanged line matches the last changed one.
1263                                  * This merges with previous changed regions.
1264                                  */
1265                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1266                                         $changed[--$start] = 1;
1267                                         $changed[--$i] = false;
1268                                         while ($start > 0 && $changed[$start - 1])
1269                                                 $start--;
1270                                         USE_ASSERTS && assert('$j > 0');
1271                                         while ($other_changed[--$j])
1272                                                 continue;
1273                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1274                                 }
1275
1276                                 /*
1277                                  * Set CORRESPONDING to the end of the changed run, at the last
1278                                  * point where it corresponds to a changed run in the other file.
1279                                  * CORRESPONDING == LEN means no such point has been found.
1280                                  */
1281                                 $corresponding = $j < $other_len ? $i : $len;
1282
1283                                 /*
1284                                  * Move the changed region forward, so long as the
1285                                  * first changed line matches the following unchanged one.
1286                                  * This merges with following changed regions.
1287                                  * Do this second, so that if there are no merges,
1288                                  * the changed region is moved forward as far as possible.
1289                                  */
1290                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1291                                         $changed[$start++] = false;
1292                                         $changed[$i++] = 1;
1293                                         while ($i < $len && $changed[$i])
1294                                                 $i++;
1295
1296                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1297                                         $j++;
1298                                         if ($j < $other_len && $other_changed[$j]) {
1299                                                 $corresponding = $i;
1300                                                 while ($j < $other_len && $other_changed[$j])
1301                                                         $j++;
1302                                         }
1303                                 }
1304                         } while ($runlength != $i - $start);
1305
1306                         /*
1307                          * If possible, move the fully-merged run of changes
1308                          * back to a corresponding run in the other file.
1309                          */
1310                         while ($corresponding < $i) {
1311                                 $changed[--$start] = 1;
1312                                 $changed[--$i] = 0;
1313                                 USE_ASSERTS && assert('$j > 0');
1314                                 while ($other_changed[--$j])
1315                                         continue;
1316                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1317                         }
1318                 }
1319                 wfProfileOut( __METHOD__ );
1320         }
1321 }
1322
1323 /**
1324  * Class representing a 'diff' between two sequences of strings.
1325  * @todo document
1326  * @private
1327  * @addtogroup DifferenceEngine
1328  */
1329 class Diff
1330 {
1331         var $edits;
1332
1333         /**
1334          * Constructor.
1335          * Computes diff between sequences of strings.
1336          *
1337          * @param $from_lines array An array of strings.
1338          *                (Typically these are lines from a file.)
1339          * @param $to_lines array An array of strings.
1340          */
1341         function Diff($from_lines, $to_lines) {
1342                 $eng = new _DiffEngine;
1343                 $this->edits = $eng->diff($from_lines, $to_lines);
1344                 //$this->_check($from_lines, $to_lines);
1345         }
1346
1347         /**
1348          * Compute reversed Diff.
1349          *
1350          * SYNOPSIS:
1351          *
1352          *      $diff = new Diff($lines1, $lines2);
1353          *      $rev = $diff->reverse();
1354          * @return object A Diff object representing the inverse of the
1355          *                                original diff.
1356          */
1357         function reverse () {
1358                 $rev = $this;
1359                 $rev->edits = array();
1360                 foreach ($this->edits as $edit) {
1361                         $rev->edits[] = $edit->reverse();
1362                 }
1363                 return $rev;
1364         }
1365
1366         /**
1367          * Check for empty diff.
1368          *
1369          * @return bool True iff two sequences were identical.
1370          */
1371         function isEmpty () {
1372                 foreach ($this->edits as $edit) {
1373                         if ($edit->type != 'copy')
1374                                 return false;
1375                 }
1376                 return true;
1377         }
1378
1379         /**
1380          * Compute the length of the Longest Common Subsequence (LCS).
1381          *
1382          * This is mostly for diagnostic purposed.
1383          *
1384          * @return int The length of the LCS.
1385          */
1386         function lcs () {
1387                 $lcs = 0;
1388                 foreach ($this->edits as $edit) {
1389                         if ($edit->type == 'copy')
1390                                 $lcs += sizeof($edit->orig);
1391                 }
1392                 return $lcs;
1393         }
1394
1395         /**
1396          * Get the original set of lines.
1397          *
1398          * This reconstructs the $from_lines parameter passed to the
1399          * constructor.
1400          *
1401          * @return array The original sequence of strings.
1402          */
1403         function orig() {
1404                 $lines = array();
1405
1406                 foreach ($this->edits as $edit) {
1407                         if ($edit->orig)
1408                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1409                 }
1410                 return $lines;
1411         }
1412
1413         /**
1414          * Get the closing set of lines.
1415          *
1416          * This reconstructs the $to_lines parameter passed to the
1417          * constructor.
1418          *
1419          * @return array The sequence of strings.
1420          */
1421         function closing() {
1422                 $lines = array();
1423
1424                 foreach ($this->edits as $edit) {
1425                         if ($edit->closing)
1426                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1427                 }
1428                 return $lines;
1429         }
1430
1431         /**
1432          * Check a Diff for validity.
1433          *
1434          * This is here only for debugging purposes.
1435          */
1436         function _check ($from_lines, $to_lines) {
1437                 wfProfileIn( __METHOD__ );
1438                 if (serialize($from_lines) != serialize($this->orig()))
1439                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1440                 if (serialize($to_lines) != serialize($this->closing()))
1441                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1442
1443                 $rev = $this->reverse();
1444                 if (serialize($to_lines) != serialize($rev->orig()))
1445                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1446                 if (serialize($from_lines) != serialize($rev->closing()))
1447                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1448
1449
1450                 $prevtype = 'none';
1451                 foreach ($this->edits as $edit) {
1452                         if ( $prevtype == $edit->type )
1453                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1454                         $prevtype = $edit->type;
1455                 }
1456
1457                 $lcs = $this->lcs();
1458                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1459                 wfProfileOut( __METHOD__ );
1460         }
1461 }
1462
1463 /**
1464  * @todo document, bad name.
1465  * @private
1466  * @addtogroup DifferenceEngine
1467  */
1468 class MappedDiff extends Diff
1469 {
1470         /**
1471          * Constructor.
1472          *
1473          * Computes diff between sequences of strings.
1474          *
1475          * This can be used to compute things like
1476          * case-insensitve diffs, or diffs which ignore
1477          * changes in white-space.
1478          *
1479          * @param $from_lines array An array of strings.
1480          *      (Typically these are lines from a file.)
1481          *
1482          * @param $to_lines array An array of strings.
1483          *
1484          * @param $mapped_from_lines array This array should
1485          *      have the same size number of elements as $from_lines.
1486          *      The elements in $mapped_from_lines and
1487          *      $mapped_to_lines are what is actually compared
1488          *      when computing the diff.
1489          *
1490          * @param $mapped_to_lines array This array should
1491          *      have the same number of elements as $to_lines.
1492          */
1493         function MappedDiff($from_lines, $to_lines,
1494                 $mapped_from_lines, $mapped_to_lines) {
1495                 wfProfileIn( __METHOD__ );
1496
1497                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1498                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1499
1500                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1501
1502                 $xi = $yi = 0;
1503                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1504                         $orig = &$this->edits[$i]->orig;
1505                         if (is_array($orig)) {
1506                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1507                                 $xi += sizeof($orig);
1508                         }
1509
1510                         $closing = &$this->edits[$i]->closing;
1511                         if (is_array($closing)) {
1512                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1513                                 $yi += sizeof($closing);
1514                         }
1515                 }
1516                 wfProfileOut( __METHOD__ );
1517         }
1518 }
1519
1520 /**
1521  * A class to format Diffs
1522  *
1523  * This class formats the diff in classic diff format.
1524  * It is intended that this class be customized via inheritance,
1525  * to obtain fancier outputs.
1526  * @todo document
1527  * @private
1528  * @addtogroup DifferenceEngine
1529  */
1530 class DiffFormatter {
1531         /**
1532          * Number of leading context "lines" to preserve.
1533          *
1534          * This should be left at zero for this class, but subclasses
1535          * may want to set this to other values.
1536          */
1537         var $leading_context_lines = 0;
1538
1539         /**
1540          * Number of trailing context "lines" to preserve.
1541          *
1542          * This should be left at zero for this class, but subclasses
1543          * may want to set this to other values.
1544          */
1545         var $trailing_context_lines = 0;
1546
1547         /**
1548          * Format a diff.
1549          *
1550          * @param $diff object A Diff object.
1551          * @return string The formatted output.
1552          */
1553         function format($diff) {
1554                 wfProfileIn( __METHOD__ );
1555
1556                 $xi = $yi = 1;
1557                 $block = false;
1558                 $context = array();
1559
1560                 $nlead = $this->leading_context_lines;
1561                 $ntrail = $this->trailing_context_lines;
1562
1563                 $this->_start_diff();
1564
1565                 foreach ($diff->edits as $edit) {
1566                         if ($edit->type == 'copy') {
1567                                 if (is_array($block)) {
1568                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1569                                                 $block[] = $edit;
1570                                         }
1571                                         else{
1572                                                 if ($ntrail) {
1573                                                         $context = array_slice($edit->orig, 0, $ntrail);
1574                                                         $block[] = new _DiffOp_Copy($context);
1575                                                 }
1576                                                 $this->_block($x0, $ntrail + $xi - $x0,
1577                                                                           $y0, $ntrail + $yi - $y0,
1578                                                                           $block);
1579                                                 $block = false;
1580                                         }
1581                                 }
1582                                 $context = $edit->orig;
1583                         }
1584                         else {
1585                                 if (! is_array($block)) {
1586                                         $context = array_slice($context, sizeof($context) - $nlead);
1587                                         $x0 = $xi - sizeof($context);
1588                                         $y0 = $yi - sizeof($context);
1589                                         $block = array();
1590                                         if ($context)
1591                                                 $block[] = new _DiffOp_Copy($context);
1592                                 }
1593                                 $block[] = $edit;
1594                         }
1595
1596                         if ($edit->orig)
1597                                 $xi += sizeof($edit->orig);
1598                         if ($edit->closing)
1599                                 $yi += sizeof($edit->closing);
1600                 }
1601
1602                 if (is_array($block))
1603                         $this->_block($x0, $xi - $x0,
1604                                                   $y0, $yi - $y0,
1605                                                   $block);
1606
1607                 $end = $this->_end_diff();
1608                 wfProfileOut( __METHOD__ );
1609                 return $end;
1610         }
1611
1612         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1613                 wfProfileIn( __METHOD__ );
1614                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1615                 foreach ($edits as $edit) {
1616                         if ($edit->type == 'copy')
1617                                 $this->_context($edit->orig);
1618                         elseif ($edit->type == 'add')
1619                                 $this->_added($edit->closing);
1620                         elseif ($edit->type == 'delete')
1621                                 $this->_deleted($edit->orig);
1622                         elseif ($edit->type == 'change')
1623                                 $this->_changed($edit->orig, $edit->closing);
1624                         else
1625                                 trigger_error('Unknown edit type', E_USER_ERROR);
1626                 }
1627                 $this->_end_block();
1628                 wfProfileOut( __METHOD__ );
1629         }
1630
1631         function _start_diff() {
1632                 ob_start();
1633         }
1634
1635         function _end_diff() {
1636                 $val = ob_get_contents();
1637                 ob_end_clean();
1638                 return $val;
1639         }
1640
1641         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1642                 if ($xlen > 1)
1643                         $xbeg .= "," . ($xbeg + $xlen - 1);
1644                 if ($ylen > 1)
1645                         $ybeg .= "," . ($ybeg + $ylen - 1);
1646
1647                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1648         }
1649
1650         function _start_block($header) {
1651                 echo $header . "\n";
1652         }
1653
1654         function _end_block() {
1655         }
1656
1657         function _lines($lines, $prefix = ' ') {
1658                 foreach ($lines as $line)
1659                         echo "$prefix $line\n";
1660         }
1661
1662         function _context($lines) {
1663                 $this->_lines($lines);
1664         }
1665
1666         function _added($lines) {
1667                 $this->_lines($lines, '>');
1668         }
1669         function _deleted($lines) {
1670                 $this->_lines($lines, '<');
1671         }
1672
1673         function _changed($orig, $closing) {
1674                 $this->_deleted($orig);
1675                 echo "---\n";
1676                 $this->_added($closing);
1677         }
1678 }
1679
1680 /**
1681  * A formatter that outputs unified diffs
1682  * @addtogroup DifferenceEngine
1683  */
1684
1685 class UnifiedDiffFormatter extends DiffFormatter {
1686         var $leading_context_lines = 2;
1687         var $trailing_context_lines = 2;
1688
1689         function _added($lines) {
1690                 $this->_lines($lines, '+');
1691         }
1692         function _deleted($lines) {
1693                 $this->_lines($lines, '-');
1694         }
1695         function _changed($orig, $closing) {
1696                 $this->_deleted($orig);
1697                 $this->_added($closing);
1698         }
1699         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1700                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1701         }
1702 }
1703
1704 /**
1705  * A pseudo-formatter that just passes along the Diff::$edits array
1706  * @addtogroup DifferenceEngine
1707  */
1708 class ArrayDiffFormatter extends DiffFormatter {
1709         function format($diff) {
1710                 $oldline = 1;
1711                 $newline = 1;
1712                 $retval = array();
1713                 foreach($diff->edits as $edit)
1714                         switch($edit->type) {
1715                                 case 'add':
1716                                         foreach($edit->closing as $l) {
1717                                                 $retval[] = array(
1718                                                         'action' => 'add',
1719                                                         'new'=> $l,
1720                                                         'newline' => $newline++
1721                                                 );
1722                                         }
1723                                         break;
1724                                 case 'delete':
1725                                         foreach($edit->orig as $l) {
1726                                                 $retval[] = array(
1727                                                         'action' => 'delete',
1728                                                         'old' => $l,
1729                                                         'oldline' => $oldline++,
1730                                                 );
1731                                         }
1732                                         break;
1733                                 case 'change':
1734                                         foreach($edit->orig as $i => $l) {
1735                                                 $retval[] = array(
1736                                                         'action' => 'change',
1737                                                         'old' => $l,
1738                                                         'new' => @$edit->closing[$i],
1739                                                         'oldline' => $oldline++,
1740                                                         'newline' => $newline++,
1741                                                 );
1742                                         }
1743                                         break;
1744                                 case 'copy':
1745                                         $oldline += count($edit->orig);
1746                                         $newline += count($edit->orig);
1747                         }
1748                 return $retval;
1749         }
1750 }
1751
1752 /**
1753  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1754  *
1755  */
1756
1757 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1758
1759 /**
1760  * @todo document
1761  * @private
1762  * @addtogroup DifferenceEngine
1763  */
1764 class _HWLDF_WordAccumulator {
1765         function _HWLDF_WordAccumulator () {
1766                 $this->_lines = array();
1767                 $this->_line = '';
1768                 $this->_group = '';
1769                 $this->_tag = '';
1770         }
1771
1772         function _flushGroup ($new_tag) {
1773                 if ($this->_group !== '') {
1774                         if ($this->_tag == 'ins')
1775                                 $this->_line .= '<ins class="diffchange diffchange-inline">' .
1776                                         htmlspecialchars ( $this->_group ) . '</ins>';
1777                         elseif ($this->_tag == 'del')
1778                                 $this->_line .= '<del class="diffchange diffchange-inline">' .
1779                                         htmlspecialchars ( $this->_group ) . '</del>';
1780                         else
1781                                 $this->_line .= htmlspecialchars ( $this->_group );
1782                 }
1783                 $this->_group = '';
1784                 $this->_tag = $new_tag;
1785         }
1786
1787         function _flushLine ($new_tag) {
1788                 $this->_flushGroup($new_tag);
1789                 if ($this->_line != '')
1790                         array_push ( $this->_lines, $this->_line );
1791                 else
1792                         # make empty lines visible by inserting an NBSP
1793                         array_push ( $this->_lines, NBSP );
1794                 $this->_line = '';
1795         }
1796
1797         function addWords ($words, $tag = '') {
1798                 if ($tag != $this->_tag)
1799                         $this->_flushGroup($tag);
1800
1801                 foreach ($words as $word) {
1802                         // new-line should only come as first char of word.
1803                         if ($word == '')
1804                                 continue;
1805                         if ($word[0] == "\n") {
1806                                 $this->_flushLine($tag);
1807                                 $word = substr($word, 1);
1808                         }
1809                         assert(!strstr($word, "\n"));
1810                         $this->_group .= $word;
1811                 }
1812         }
1813
1814         function getLines() {
1815                 $this->_flushLine('~done');
1816                 return $this->_lines;
1817         }
1818 }
1819
1820 /**
1821  * @todo document
1822  * @private
1823  * @addtogroup DifferenceEngine
1824  */
1825 class WordLevelDiff extends MappedDiff {
1826         const MAX_LINE_LENGTH = 10000;
1827
1828         function WordLevelDiff ($orig_lines, $closing_lines) {
1829                 wfProfileIn( __METHOD__ );
1830
1831                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1832                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1833
1834                 $this->MappedDiff($orig_words, $closing_words,
1835                         $orig_stripped, $closing_stripped);
1836                 wfProfileOut( __METHOD__ );
1837         }
1838
1839         function _split($lines) {
1840                 wfProfileIn( __METHOD__ );
1841
1842                 $words = array();
1843                 $stripped = array();
1844                 $first = true;
1845                 foreach ( $lines as $line ) {
1846                         # If the line is too long, just pretend the entire line is one big word
1847                         # This prevents resource exhaustion problems
1848                         if ( $first ) {
1849                                 $first = false;
1850                         } else {
1851                                 $words[] = "\n";
1852                                 $stripped[] = "\n";
1853                         }
1854                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1855                                 $words[] = $line;
1856                                 $stripped[] = $line;
1857                         } else {
1858                                 $m = array();
1859                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1860                                         $line, $m))
1861                                 {
1862                                         $words = array_merge( $words, $m[0] );
1863                                         $stripped = array_merge( $stripped, $m[1] );
1864                                 }
1865                         }
1866                 }
1867                 wfProfileOut( __METHOD__ );
1868                 return array($words, $stripped);
1869         }
1870
1871         function orig () {
1872                 wfProfileIn( __METHOD__ );
1873                 $orig = new _HWLDF_WordAccumulator;
1874
1875                 foreach ($this->edits as $edit) {
1876                         if ($edit->type == 'copy')
1877                                 $orig->addWords($edit->orig);
1878                         elseif ($edit->orig)
1879                                 $orig->addWords($edit->orig, 'del');
1880                 }
1881                 $lines = $orig->getLines();
1882                 wfProfileOut( __METHOD__ );
1883                 return $lines;
1884         }
1885
1886         function closing () {
1887                 wfProfileIn( __METHOD__ );
1888                 $closing = new _HWLDF_WordAccumulator;
1889
1890                 foreach ($this->edits as $edit) {
1891                         if ($edit->type == 'copy')
1892                                 $closing->addWords($edit->closing);
1893                         elseif ($edit->closing)
1894                                 $closing->addWords($edit->closing, 'ins');
1895                 }
1896                 $lines = $closing->getLines();
1897                 wfProfileOut( __METHOD__ );
1898                 return $lines;
1899         }
1900 }
1901
1902 /**
1903  * Wikipedia Table style diff formatter.
1904  * @todo document
1905  * @private
1906  * @addtogroup DifferenceEngine
1907  */
1908 class TableDiffFormatter extends DiffFormatter {
1909         function TableDiffFormatter() {
1910                 $this->leading_context_lines = 2;
1911                 $this->trailing_context_lines = 2;
1912         }
1913
1914         public static function escapeWhiteSpace( $msg ) {
1915                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
1916                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
1917                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
1918                 return $msg;
1919         }
1920
1921         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1922                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1923                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1924                 return $r;
1925         }
1926
1927         function _start_block( $header ) {
1928                 echo $header;
1929         }
1930
1931         function _end_block() {
1932         }
1933
1934         function _lines( $lines, $prefix=' ', $color='white' ) {
1935         }
1936
1937         # HTML-escape parameter before calling this
1938         function addedLine( $line ) {
1939                 return $this->wrapLine( '+', 'diff-addedline', $line );
1940         }
1941
1942         # HTML-escape parameter before calling this
1943         function deletedLine( $line ) {
1944                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1945         }
1946
1947         # HTML-escape parameter before calling this
1948         function contextLine( $line ) {
1949                 return $this->wrapLine( ' ', 'diff-context', $line );
1950         }
1951
1952         private function wrapLine( $marker, $class, $line ) {
1953                 if( $line !== '' ) {
1954                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1955                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1956                 }
1957                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1958         }
1959
1960         function emptyLine() {
1961                 return '<td colspan="2">&nbsp;</td>';
1962         }
1963
1964         function _added( $lines ) {
1965                 foreach ($lines as $line) {
1966                         echo '<tr>' . $this->emptyLine() .
1967                                 $this->addedLine( '<ins class="diffchange">' .
1968                                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1969                 }
1970         }
1971
1972         function _deleted($lines) {
1973                 foreach ($lines as $line) {
1974                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1975                                 htmlspecialchars ( $line ) . '</del>' ) .
1976                           $this->emptyLine() . "</tr>\n";
1977                 }
1978         }
1979
1980         function _context( $lines ) {
1981                 foreach ($lines as $line) {
1982                         echo '<tr>' .
1983                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1984                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1985                 }
1986         }
1987
1988         function _changed( $orig, $closing ) {
1989                 wfProfileIn( __METHOD__ );
1990
1991                 $diff = new WordLevelDiff( $orig, $closing );
1992                 $del = $diff->orig();
1993                 $add = $diff->closing();
1994
1995                 # Notice that WordLevelDiff returns HTML-escaped output.
1996                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1997
1998                 while ( $line = array_shift( $del ) ) {
1999                         $aline = array_shift( $add );
2000                         echo '<tr>' . $this->deletedLine( $line ) .
2001                                 $this->addedLine( $aline ) . "</tr>\n";
2002                 }
2003                 foreach ($add as $line) {       # If any leftovers
2004                         echo '<tr>' . $this->emptyLine() .
2005                                 $this->addedLine( $line ) . "</tr>\n";
2006                 }
2007                 wfProfileOut( __METHOD__ );
2008         }
2009 }