includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * @defgroup DifferenceEngine DifferenceEngine
   4  */
   5
   6 /**
   7  * Constant to indicate diff cache compatibility.
   8  * Bump this when changing the diff formatting in a way that
   9  * fixes important bugs or such to force cached diff views to
  10  * clear.
  11  */
  12 define( 'MW_DIFF_VERSION', '1.11a' );
  13
  14 /**
  15  * @todo document
  16  * @ingroup DifferenceEngine
  17  */
  18 class DifferenceEngine {
  19         /**#@+
  20          * @private
  21          */
  22         var $mOldid, $mNewid, $mTitle;
  23         var $mOldtitle, $mNewtitle, $mPagetitle;
  24         var $mOldtext, $mNewtext;
  25         var $mOldPage, $mNewPage;
  26         var $mRcidMarkPatrolled;
  27         var $mOldRev, $mNewRev;
  28         var $mRevisionsLoaded = false; // Have the revisions been loaded
  29         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  30         /**#@-*/
  31
  32         /**
  33          * Constructor
  34          * @param $titleObj Title object that the diff is associated with
  35          * @param $old Integer: old ID we want to show and diff with.
  36          * @param $new String: either 'prev' or 'next'.
  37          * @param $rcid Integer: ??? FIXME (default 0)
  38          * @param $refreshCache boolean If set, refreshes the diff cache
  39          */
  40         function __construct( $titleObj = null, $old = 0, $new = 0, $rcid = 0, $refreshCache = false ) {
  41                 $this->mTitle = $titleObj;
  42                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  43
  44                 if ( 'prev' === $new ) {
  45                         # Show diff between revision $old and the previous one.
  46                         # Get previous one from DB.
  47                         #
  48                         $this->mNewid = intval($old);
  49
  50                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  51
  52                 } elseif ( 'next' === $new ) {
  53                         # Show diff between revision $old and the previous one.
  54                         # Get previous one from DB.
  55                         #
  56                         $this->mOldid = intval($old);
  57                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  58                         if ( false === $this->mNewid ) {
  59                                 # if no result, NewId points to the newest old revision. The only newer
  60                                 # revision is cur, which is "0".
  61                                 $this->mNewid = 0;
  62                         }
  63
  64                 } else {
  65                         $this->mOldid = intval($old);
  66                         $this->mNewid = intval($new);
  67                 }
  68                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  69                 $this->mRefreshCache = $refreshCache;
  70         }
  71
  72         function getTitle() {
  73                 return $this->mTitle;
  74         }
  75
  76         function showDiffPage( $diffOnly = false ) {
  77                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  78                 wfProfileIn( __METHOD__ );
  79
  80                 # If external diffs are enabled both globally and for the user,
  81                 # we'll use the application/x-external-editor interface to call
  82                 # an external diff tool like kompare, kdiff3, etc.
  83                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  84                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  85                         $wgOut->disable();
  86                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  87                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  88                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  89                         $special=$wgLang->getNsText(NS_SPECIAL);
  90                         $control=<<<CONTROL
  91                         [Process]
  92                         Type=Diff text
  93                         Engine=MediaWiki
  94                         Script={$wgServer}{$wgScript}
  95                         Special namespace={$special}
  96
  97                         [File]
  98                         Extension=wiki
  99                         URL=$url1
 100
 101                         [File 2]
 102                         Extension=wiki
 103                         URL=$url2
 104 CONTROL;
 105                         echo($control);
 106                         return;
 107                 }
 108
 109                 $wgOut->setArticleFlag( false );
 110                 if ( ! $this->loadRevisionData() ) {
 111                         $t = $this->mTitle->getPrefixedText();
 112                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 113                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 114                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 115                         wfProfileOut( __METHOD__ );
 116                         return;
 117                 }
 118
 119                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 120
 121                 if ( $this->mNewRev->isCurrent() ) {
 122                         $wgOut->setArticleFlag( true );
 123                 }
 124
 125                 # mOldid is false if the difference engine is called with a "vague" query for
 126                 # a diff between a version V and its previous version V' AND the version V
 127                 # is the first version of that article. In that case, V' does not exist.
 128                 if ( $this->mOldid === false ) {
 129                         $this->showFirstRevision();
 130                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 131                         wfProfileOut( __METHOD__ );
 132                         return;
 133                 }
 134
 135                 $wgOut->suppressQuickbar();
 136
 137                 $oldTitle = $this->mOldPage->getPrefixedText();
 138                 $newTitle = $this->mNewPage->getPrefixedText();
 139                 if( $oldTitle == $newTitle ) {
 140                         $wgOut->setPageTitle( $newTitle );
 141                 } else {
 142                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 143                 }
 144                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 145                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 146
 147                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 148                         $wgOut->loginToUse();
 149                         $wgOut->output();
 150                         wfProfileOut( __METHOD__ );
 151                         exit;
 152                 }
 153
 154                 $sk = $wgUser->getSkin();
 155
 156                 // Check if page is editable
 157                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 158                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed( 'rollback' ) ) {
 159                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 160                 } else {
 161                         $rollback = '';
 162                 }
 163
 164                 // Prepare a change patrol link, if applicable
 165                 if( $wgUseRCPatrol && $wgUser->isAllowed( 'patrol' ) ) {
 166                         // If we've been given an explicit change identifier, use it; saves time
 167                         if( $this->mRcidMarkPatrolled ) {
 168                                 $rcid = $this->mRcidMarkPatrolled;
 169                         } else {
 170                                 // Look for an unpatrolled change corresponding to this diff
 171                                 $db = wfGetDB( DB_SLAVE );
 172                                 $change = RecentChange::newFromConds(
 173                                 array(
 174                                 // Add redundant user,timestamp condition so we can use the existing index
 175                                                 'rc_user_text'  => $this->mNewRev->getRawUserText(),
 176                                                 'rc_timestamp' => $db->timestamp( $this->mNewRev->getTimestamp() ),
 177                                                 'rc_this_oldid' => $this->mNewid,
 178                                                 'rc_last_oldid' => $this->mOldid,
 179                                                 'rc_patrolled' => 0
 180                                 ),
 181                                 __METHOD__
 182                                 );
 183                                 if( $change instanceof RecentChange ) {
 184                                         $rcid = $change->mAttribs['rc_id'];
 185                                 } else {
 186                                         // None found
 187                                         $rcid = 0;
 188                                 }
 189                         }
 190                         // Build the link
 191                         if( $rcid ) {
 192                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj(
 193                                 $this->mTitle,
 194                                 wfMsgHtml( 'markaspatrolleddiff' ),
 195                                         "action=markpatrolled&rcid={$rcid}"
 196                                 ) . ']</span>';
 197                         } else {
 198                                 $patrol = '';
 199                         }
 200                 } else {
 201                         $patrol = '';
 202                 }
 203
 204                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 205                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 206                 if ( $this->mNewRev->isCurrent() ) {
 207                         $nextlink = '&nbsp;';
 208                 } else {
 209                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 210                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 211                 }
 212
 213                 $oldminor = '';
 214                 $newminor = '';
 215
 216                 if ($this->mOldRev->mMinorEdit == 1) {
 217                         $oldminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 218                 }
 219
 220                 if ($this->mNewRev->mMinorEdit == 1) {
 221                         $newminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 222                 }
 223
 224                 $rdel = ''; $ldel = '';
 225                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 226                         $revdel = SpecialPage::getTitleFor( 'Revisiondelete' );
 227                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 228                                 // If revision was hidden from sysops
 229                                 $ldel = wfMsgHtml('rev-delundel');
 230                         } else {
 231                                 $ldel = $sk->makeKnownLinkObj( $revdel,
 232                                 wfMsgHtml('rev-delundel'),
 233                                         'target=' . urlencode( $this->mOldRev->mTitle->getPrefixedDbkey() ) .
 234                                         '&oldid=' . urlencode( $this->mOldRev->getId() ) );
 235                                 // Bolden oversighted content
 236                                 if( $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 237                                 $ldel = "<strong>$ldel</strong>";
 238                         }
 239                         $ldel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$ldel</small>)</tt> ";
 240                         // We don't currently handle well changing the top revision's settings
 241                         if( $this->mNewRev->isCurrent() ) {
 242                                 // If revision was hidden from sysops
 243                                 $rdel = wfMsgHtml('rev-delundel');
 244                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 245                                 // If revision was hidden from sysops
 246                                 $rdel = wfMsgHtml('rev-delundel');
 247                         } else {
 248                                 $rdel = $sk->makeKnownLinkObj( $revdel,
 249                                 wfMsgHtml('rev-delundel'),
 250                                         'target=' . urlencode( $this->mNewRev->mTitle->getPrefixedDbkey() ) .
 251                                         '&oldid=' . urlencode( $this->mNewRev->getId() ) );
 252                                 // Bolden oversighted content
 253                                 if( $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 254                                 $rdel = "<strong>$rdel</strong>";
 255                         }
 256                         $rdel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$rdel</small>)</tt> ";
 257                 }
 258
 259                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 260                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, true ) . "</div>" .
 261                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, true ) . $ldel . "</div>" .
 262                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 263                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 264                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, true ) . " $rollback</div>" .
 265                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, true ) . $rdel . "</div>" .
 266                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 267
 268                 $this->showDiff( $oldHeader, $newHeader );
 269
 270                 if ( !$diffOnly )
 271                 $this->renderNewRevision();
 272
 273                 wfProfileOut( __METHOD__ );
 274         }
 275
 276         /**
 277          * Show the new revision of the page.
 278          */
 279         function renderNewRevision() {
 280                 global $wgOut;
 281                 wfProfileIn( __METHOD__ );
 282
 283                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 284                 #add deleted rev tag if needed
 285                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 286                         $wgOut->addWikiMsg( 'rev-deleted-text-permission' );
 287                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 288                         $wgOut->addWikiMsg( 'rev-deleted-text-view' );
 289                 }
 290
 291                 if( !$this->mNewRev->isCurrent() ) {
 292                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 293                 }
 294
 295                 $this->loadNewText();
 296                 if( is_object( $this->mNewRev ) ) {
 297                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 298                 }
 299
 300                 if ($this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage()) {
 301                         // Stolen from Article::view --AG 2007-10-11
 302
 303                         // Give hooks a chance to customise the output
 304                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 305                                 // Wrap the whole lot in a <pre> and don't parse
 306                                 $m = array();
 307                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 308                                 $wgOut->addHtml( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 309                                 $wgOut->addHtml( htmlspecialchars( $this->mNewtext ) );
 310                                 $wgOut->addHtml( "\n</pre>\n" );
 311                         }
 312                 } else
 313                 $wgOut->addWikiTextTidy( $this->mNewtext );
 314
 315                 if( !$this->mNewRev->isCurrent() ) {
 316                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 317                 }
 318
 319                 wfProfileOut( __METHOD__ );
 320         }
 321
 322         /**
 323          * Show the first revision of an article. Uses normal diff headers in
 324          * contrast to normal "old revision" display style.
 325          */
 326         function showFirstRevision() {
 327                 global $wgOut, $wgUser;
 328                 wfProfileIn( __METHOD__ );
 329
 330                 # Get article text from the DB
 331                 #
 332                 if ( ! $this->loadNewText() ) {
 333                         $t = $this->mTitle->getPrefixedText();
 334                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 335                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 336                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 337                         wfProfileOut( __METHOD__ );
 338                         return;
 339                 }
 340                 if ( $this->mNewRev->isCurrent() ) {
 341                         $wgOut->setArticleFlag( true );
 342                 }
 343
 344                 # Check if user is allowed to look at this page. If not, bail out.
 345                 #
 346                 if ( !( $this->mTitle->userCanRead() ) ) {
 347                         $wgOut->loginToUse();
 348                         $wgOut->output();
 349                         wfProfileOut( __METHOD__ );
 350                         exit;
 351                 }
 352
 353                 # Prepare the header box
 354                 #
 355                 $sk = $wgUser->getSkin();
 356
 357                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 358                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 359                 $sk->revUserTools( $this->mNewRev ) . "<br />" .
 360                 $sk->revComment( $this->mNewRev ) . "<br />" .
 361                 $nextlink . "</div>\n";
 362
 363                 $wgOut->addHTML( $header );
 364
 365                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 366                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 367
 368                 wfProfileOut( __METHOD__ );
 369         }
 370
 371         /**
 372          * Get the diff text, send it to $wgOut
 373          * Returns false if the diff could not be generated, otherwise returns true
 374          */
 375         function showDiff( $otitle, $ntitle ) {
 376                 global $wgOut;
 377                 $diff = $this->getDiff( $otitle, $ntitle );
 378                 if ( $diff === false ) {
 379                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 380                         return false;
 381                 } else {
 382                         $this->showDiffStyle();
 383                         $wgOut->addHTML( $diff );
 384                         return true;
 385                 }
 386         }
 387
 388         /**
 389          * Add style sheets and supporting JS for diff display.
 390          */
 391         function showDiffStyle() {
 392                 global $wgStylePath, $wgStyleVersion, $wgOut;
 393                 $wgOut->addStyle( 'common/diff.css' );
 394
 395                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 396                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 397         }
 398
 399         /**
 400          * Get complete diff table, including header
 401          *
 402          * @param Title $otitle Old title
 403          * @param Title $ntitle New title
 404          * @return mixed
 405          */
 406         function getDiff( $otitle, $ntitle ) {
 407                 $body = $this->getDiffBody();
 408                 if ( $body === false ) {
 409                         return false;
 410                 } else {
 411                         $multi = $this->getMultiNotice();
 412                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 413                 }
 414         }
 415
 416         /**
 417          * Get the diff table body, without header
 418          *
 419          * @return mixed
 420          */
 421         function getDiffBody() {
 422                 global $wgMemc;
 423                 wfProfileIn( __METHOD__ );
 424                 // Check if the diff should be hidden from this user
 425                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 426                         return '';
 427                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 428                         return '';
 429                 }
 430                 // Cacheable?
 431                 $key = false;
 432                 if ( $this->mOldid && $this->mNewid ) {
 433                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 434                         // Try cache
 435                         if ( !$this->mRefreshCache ) {
 436                                 $difftext = $wgMemc->get( $key );
 437                                 if ( $difftext ) {
 438                                         wfIncrStats( 'diff_cache_hit' );
 439                                         $difftext = $this->localiseLineNumbers( $difftext );
 440                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 441                                         wfProfileOut( __METHOD__ );
 442                                         return $difftext;
 443                                 }
 444                         } // don't try to load but save the result
 445                 }
 446
 447                 // Loadtext is permission safe, this just clears out the diff
 448                 if ( !$this->loadText() ) {
 449                         wfProfileOut( __METHOD__ );
 450                         return false;
 451                 }
 452
 453                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 454
 455                 // Save to cache for 7 days
 456                 if ( $key !== false && $difftext !== false ) {
 457                         wfIncrStats( 'diff_cache_miss' );
 458                         $wgMemc->set( $key, $difftext, 7*86400 );
 459                 } else {
 460                         wfIncrStats( 'diff_uncacheable' );
 461                 }
 462                 // Replace line numbers with the text in the user's language
 463                 if ( $difftext !== false ) {
 464                         $difftext = $this->localiseLineNumbers( $difftext );
 465                 }
 466                 wfProfileOut( __METHOD__ );
 467                 return $difftext;
 468         }
 469
 470         /**
 471          * Generate a diff, no caching
 472          * $otext and $ntext must be already segmented
 473          */
 474         function generateDiffBody( $otext, $ntext ) {
 475                 global $wgExternalDiffEngine, $wgContLang;
 476
 477                 $otext = str_replace( "\r\n", "\n", $otext );
 478                 $ntext = str_replace( "\r\n", "\n", $ntext );
 479
 480                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 481                         # For historical reasons, external diff engine expects
 482                         # input text to be HTML-escaped already
 483                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 484                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 485                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 486                                 dl('php_wikidiff.so');
 487                         }
 488                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 489                         $this->debug( 'wikidiff1' );
 490                 }
 491
 492                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 493                         # Better external diff engine, the 2 may some day be dropped
 494                         # This one does the escaping and segmenting itself
 495                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 496                                 wfProfileIn( __METHOD__ . "-dl" );
 497                                 @dl('php_wikidiff2.so');
 498                                 wfProfileOut( __METHOD__ . "-dl" );
 499                         }
 500                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 501                                 wfProfileIn( 'wikidiff2_do_diff' );
 502                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 503                                 $text .= $this->debug( 'wikidiff2' );
 504                                 wfProfileOut( 'wikidiff2_do_diff' );
 505                                 return $text;
 506                         }
 507                 }
 508                 if ( $wgExternalDiffEngine != 'wikidiff3' && $wgExternalDiffEngine !== false ) {
 509                         # Diff via the shell
 510                         global $wgTmpDirectory;
 511                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 512                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 513
 514                         $tempFile1 = fopen( $tempName1, "w" );
 515                         if ( !$tempFile1 ) {
 516                                 wfProfileOut( __METHOD__ );
 517                                 return false;
 518                         }
 519                         $tempFile2 = fopen( $tempName2, "w" );
 520                         if ( !$tempFile2 ) {
 521                                 wfProfileOut( __METHOD__ );
 522                                 return false;
 523                         }
 524                         fwrite( $tempFile1, $otext );
 525                         fwrite( $tempFile2, $ntext );
 526                         fclose( $tempFile1 );
 527                         fclose( $tempFile2 );
 528                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 529                         wfProfileIn( __METHOD__ . "-shellexec" );
 530                         $difftext = wfShellExec( $cmd );
 531                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 532                         wfProfileOut( __METHOD__ . "-shellexec" );
 533                         unlink( $tempName1 );
 534                         unlink( $tempName2 );
 535                         return $difftext;
 536                 }
 537
 538                 # Native PHP diff
 539                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 540                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 541                 $diffs = new Diff( $ota, $nta );
 542                 $formatter = new TableDiffFormatter();
 543                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 544                 $this->debug();
 545         }
 546
 547         /**
 548          * Generate a debug comment indicating diff generating time,
 549          * server node, and generator backend.
 550          */
 551         protected function debug( $generator="internal" ) {
 552                 global $wgShowHostnames, $wgNodeName;
 553                 $data = array( $generator );
 554                 if( $wgShowHostnames ) {
 555                         $data[] = $wgNodeName;
 556                 }
 557                 $data[] = wfTimestamp( TS_DB );
 558                 return "<!-- diff generator: " .
 559                 implode( " ",
 560                 array_map(
 561                                         "htmlspecialchars",
 562                 $data ) ) .
 563                         " -->\n";
 564         }
 565
 566         /**
 567          * Replace line numbers with the text in the user's language
 568          */
 569         function localiseLineNumbers( $text ) {
 570                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 571                 array( &$this, 'localiseLineNumbersCb' ), $text );
 572         }
 573
 574         function localiseLineNumbersCb( $matches ) {
 575                 global $wgLang;
 576                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 577         }
 578
 579
 580         /**
 581          * If there are revisions between the ones being compared, return a note saying so.
 582          */
 583         function getMultiNotice() {
 584                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 585                 return '';
 586
 587                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 588                         // Comparing two different pages? Count would be meaningless.
 589                         return '';
 590                 }
 591
 592                 $oldid = $this->mOldRev->getId();
 593                 $newid = $this->mNewRev->getId();
 594                 if ( $oldid > $newid ) {
 595                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 596                 }
 597
 598                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 599                 if ( !$n )
 600                 return '';
 601
 602                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 603         }
 604
 605
 606         /**
 607          * Add the header to a diff body
 608          */
 609         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 610                 global $wgOut;
 611
 612                 $header = "
 613                 <table class='diff'>
 614                 <col class='diff-marker' />
 615                 <col class='diff-content' />
 616                 <col class='diff-marker' />
 617                 <col class='diff-content' />
 618                 <tr valign='top'>
 619                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 620                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 621                 </tr>
 622                 ";
 623
 624                 if ( $multi != '' )
 625                 $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 626
 627                 return $header . $diff . "</table>";
 628         }
 629
 630         /**
 631          * Use specified text instead of loading from the database
 632          */
 633         function setText( $oldText, $newText ) {
 634                 $this->mOldtext = $oldText;
 635                 $this->mNewtext = $newText;
 636                 $this->mTextLoaded = 2;
 637         }
 638
 639         /**
 640          * Load revision metadata for the specified articles. If newid is 0, then compare
 641          * the old article in oldid to the current article; if oldid is 0, then
 642          * compare the current article to the immediately previous one (ignoring the
 643          * value of newid).
 644          *
 645          * If oldid is false, leave the corresponding revision object set
 646          * to false. This is impossible via ordinary user input, and is provided for
 647          * API convenience.
 648          */
 649         function loadRevisionData() {
 650                 global $wgLang;
 651                 if ( $this->mRevisionsLoaded ) {
 652                         return true;
 653                 } else {
 654                         // Whether it succeeds or fails, we don't want to try again
 655                         $this->mRevisionsLoaded = true;
 656                 }
 657
 658                 // Load the new revision object
 659                 $this->mNewRev = $this->mNewid
 660                 ? Revision::newFromId( $this->mNewid )
 661                 : Revision::newFromTitle( $this->mTitle );
 662                 if( !$this->mNewRev instanceof Revision )
 663                 return false;
 664
 665                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 666                 $this->mNewid = $this->mNewRev->getId();
 667
 668                 // Check if page is editable
 669                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 670
 671                 // Set assorted variables
 672                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 673                 $this->mNewPage = $this->mNewRev->getTitle();
 674                 if( $this->mNewRev->isCurrent() ) {
 675                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 676                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 677                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 678
 679                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)";
 680                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 681
 682                 } else {
 683                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 684                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 685                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 686
 687                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 688                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 689                 }
 690                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 691                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 692                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 693                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 694                 }
 695
 696                 // Load the old revision object
 697                 $this->mOldRev = false;
 698                 if( $this->mOldid ) {
 699                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 700                 } elseif ( $this->mOldid === 0 ) {
 701                         $rev = $this->mNewRev->getPrevious();
 702                         if( $rev ) {
 703                                 $this->mOldid = $rev->getId();
 704                                 $this->mOldRev = $rev;
 705                         } else {
 706                                 // No previous revision; mark to show as first-version only.
 707                                 $this->mOldid = false;
 708                                 $this->mOldRev = false;
 709                         }
 710                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 711
 712                 if( is_null( $this->mOldRev ) ) {
 713                         return false;
 714                 }
 715
 716                 if ( $this->mOldRev ) {
 717                         $this->mOldPage = $this->mOldRev->getTitle();
 718
 719                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 720                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 721                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 722                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 723
 724                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 725                         . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 726                         // Add an "undo" link
 727                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 728                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 729                                 $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 730                         }
 731
 732                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 733                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 734                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 735                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 736                         }
 737                 }
 738
 739                 return true;
 740         }
 741
 742         /**
 743          * Load the text of the revisions, as well as revision data.
 744          */
 745         function loadText() {
 746                 if ( $this->mTextLoaded == 2 ) {
 747                         return true;
 748                 } else {
 749                         // Whether it succeeds or fails, we don't want to try again
 750                         $this->mTextLoaded = 2;
 751                 }
 752
 753                 if ( !$this->loadRevisionData() ) {
 754                         return false;
 755                 }
 756                 if ( $this->mOldRev ) {
 757                         $this->mOldtext = $this->mOldRev->revText();
 758                         if ( $this->mOldtext === false ) {
 759                                 return false;
 760                         }
 761                 }
 762                 if ( $this->mNewRev ) {
 763                         $this->mNewtext = $this->mNewRev->revText();
 764                         if ( $this->mNewtext === false ) {
 765                                 return false;
 766                         }
 767                 }
 768                 return true;
 769         }
 770
 771         /**
 772          * Load the text of the new revision, not the old one
 773          */
 774         function loadNewText() {
 775                 if ( $this->mTextLoaded >= 1 ) {
 776                         return true;
 777                 } else {
 778                         $this->mTextLoaded = 1;
 779                 }
 780                 if ( !$this->loadRevisionData() ) {
 781                         return false;
 782                 }
 783                 $this->mNewtext = $this->mNewRev->getText();
 784                 return true;
 785         }
 786
 787
 788 }
 789
 790 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 791 //
 792 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 793 // You may copy this code freely under the conditions of the GPL.
 794 //
 795
 796 define('USE_ASSERTS', function_exists('assert'));
 797
 798 /**
 799  * @todo document
 800  * @private
 801  * @ingroup DifferenceEngine
 802  */
 803 class _DiffOp {
 804         var $type;
 805         var $orig;
 806         var $closing;
 807
 808         function reverse() {
 809                 trigger_error('pure virtual', E_USER_ERROR);
 810         }
 811
 812         function norig() {
 813                 return $this->orig ? sizeof($this->orig) : 0;
 814         }
 815
 816         function nclosing() {
 817                 return $this->closing ? sizeof($this->closing) : 0;
 818         }
 819 }
 820
 821 /**
 822  * @todo document
 823  * @private
 824  * @ingroup DifferenceEngine
 825  */
 826 class _DiffOp_Copy extends _DiffOp {
 827         var $type = 'copy';
 828
 829         function _DiffOp_Copy ($orig, $closing = false) {
 830                 if (!is_array($closing))
 831                 $closing = $orig;
 832                 $this->orig = $orig;
 833                 $this->closing = $closing;
 834         }
 835
 836         function reverse() {
 837                 return new _DiffOp_Copy($this->closing, $this->orig);
 838         }
 839 }
 840
 841 /**
 842  * @todo document
 843  * @private
 844  * @ingroup DifferenceEngine
 845  */
 846 class _DiffOp_Delete extends _DiffOp {
 847         var $type = 'delete';
 848
 849         function _DiffOp_Delete ($lines) {
 850                 $this->orig = $lines;
 851                 $this->closing = false;
 852         }
 853
 854         function reverse() {
 855                 return new _DiffOp_Add($this->orig);
 856         }
 857 }
 858
 859 /**
 860  * @todo document
 861  * @private
 862  * @ingroup DifferenceEngine
 863  */
 864 class _DiffOp_Add extends _DiffOp {
 865         var $type = 'add';
 866
 867         function _DiffOp_Add ($lines) {
 868                 $this->closing = $lines;
 869                 $this->orig = false;
 870         }
 871
 872         function reverse() {
 873                 return new _DiffOp_Delete($this->closing);
 874         }
 875 }
 876
 877 /**
 878  * @todo document
 879  * @private
 880  * @ingroup DifferenceEngine
 881  */
 882 class _DiffOp_Change extends _DiffOp {
 883         var $type = 'change';
 884
 885         function _DiffOp_Change ($orig, $closing) {
 886                 $this->orig = $orig;
 887                 $this->closing = $closing;
 888         }
 889
 890         function reverse() {
 891                 return new _DiffOp_Change($this->closing, $this->orig);
 892         }
 893 }
 894
 895 /**
 896  * Class used internally by Diff to actually compute the diffs.
 897  *
 898  * The algorithm used here is mostly lifted from the perl module
 899  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 900  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 901  *
 902  * More ideas are taken from:
 903  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 904  *
 905  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 906  * diffutils-2.7, which can be found at:
 907  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 908  *
 909  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 910  * are my own.
 911  *
 912  * Line length limits for robustness added by Tim Starling, 2005-08-31
 913  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
 914  *
 915  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
 916  * @private
 917  * @ingroup DifferenceEngine
 918  */
 919 class _DiffEngine {
 920
 921         const MAX_XREF_LENGTH =  10000;
 922
 923         function diff ($from_lines, $to_lines) {
 924                 wfProfileIn( __METHOD__ );
 925
 926                 global $wgExternalDiffEngine;
 927
 928                 $n_from = sizeof($from_lines);
 929                 $n_to = sizeof($to_lines);
 930
 931                 if($wgExternalDiffEngine == 'wikidiff3'){
 932                         // wikidiff3
 933                         global $IP;
 934                         require_once( "$IP/includes/Diff.php" );
 935                         list($this->xchanged, $this->ychanged) = wikidiff3_diff($from_lines, $to_lines, TRUE, 100000);
 936                 }else{
 937                         // old diff
 938                         wfProfileIn( __METHOD__ ." - basicdiff");
 939                         $this->xchanged = $this->ychanged = array();
 940                         $this->xv = $this->yv = array();
 941                         $this->xind = $this->yind = array();
 942                         unset($this->seq);
 943                         unset($this->in_seq);
 944                         unset($this->lcs);
 945
 946                         // Skip leading common lines.
 947                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 948                                 if ($from_lines[$skip] !== $to_lines[$skip])
 949                                 break;
 950                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 951                         }
 952                         // Skip trailing common lines.
 953                         $xi = $n_from; $yi = $n_to;
 954                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 955                                 if ($from_lines[$xi] !== $to_lines[$yi])
 956                                 break;
 957                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 958                         }
 959
 960                         // Ignore lines which do not exist in both files.
 961                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 962                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 963                         }
 964
 965                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 966                                 $line = $to_lines[$yi];
 967                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 968                                 continue;
 969                                 $yhash[$this->_line_hash($line)] = 1;
 970                                 $this->yv[] = $line;
 971                                 $this->yind[] = $yi;
 972                         }
 973                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 974                                 $line = $from_lines[$xi];
 975                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 976                                 continue;
 977                                 $this->xv[] = $line;
 978                                 $this->xind[] = $xi;
 979                         }
 980
 981                         // Find the LCS.
 982                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 983                         wfProfileOut( __METHOD__ ." - basicdiff");
 984                 }
 985
 986                 // Merge edits when possible
 987                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 988                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 989
 990                 // Compute the edit operations.
 991                 $edits = array();
 992                 $xi = $yi = 0;
 993                 while ($xi < $n_from || $yi < $n_to) {
 994                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 995                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 996
 997                         // Skip matching "snake".
 998                         $copy = array();
 999                         while ( $xi < $n_from && $yi < $n_to
1000                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
1001                                 $copy[] = $from_lines[$xi++];
1002                                 ++$yi;
1003                         }
1004                         if ($copy)
1005                         $edits[] = new _DiffOp_Copy($copy);
1006
1007                         // Find deletes & adds.
1008                         $delete = array();
1009                         while ($xi < $n_from && $this->xchanged[$xi])
1010                         $delete[] = $from_lines[$xi++];
1011
1012                         $add = array();
1013                         while ($yi < $n_to && $this->ychanged[$yi])
1014                         $add[] = $to_lines[$yi++];
1015
1016                         if ($delete && $add)
1017                         $edits[] = new _DiffOp_Change($delete, $add);
1018                         elseif ($delete)
1019                         $edits[] = new _DiffOp_Delete($delete);
1020                         elseif ($add)
1021                         $edits[] = new _DiffOp_Add($add);
1022                 }
1023                 wfProfileOut( __METHOD__ );
1024                 return $edits;
1025         }
1026
1027         /**
1028          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1029          */
1030         function _line_hash( $line ) {
1031                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1032                         return md5( $line );
1033                 } else {
1034                         return $line;
1035                 }
1036         }
1037
1038         /* Divide the Largest Common Subsequence (LCS) of the sequences
1039          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1040          * sized segments.
1041          *
1042          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1043          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1044          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1045          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1046          * that (X0, Y0) == (XOFF, YOFF) and
1047          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1048          *
1049          * This function assumes that the first lines of the specified portions
1050          * of the two files do not match, and likewise that the last lines do not
1051          * match.  The caller must trim matching lines from the beginning and end
1052          * of the portions it is going to specify.
1053          */
1054         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1055                 $flip = false;
1056
1057                 if ($xlim - $xoff > $ylim - $yoff) {
1058                         // Things seems faster (I'm not sure I understand why)
1059                         // when the shortest sequence in X.
1060                         $flip = true;
1061                         list ($xoff, $xlim, $yoff, $ylim)
1062                         = array( $yoff, $ylim, $xoff, $xlim);
1063                 }
1064
1065                 if ($flip)
1066                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1067                 $ymatches[$this->xv[$i]][] = $i;
1068                 else
1069                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1070                 $ymatches[$this->yv[$i]][] = $i;
1071
1072                 $this->lcs = 0;
1073                 $this->seq[0]= $yoff - 1;
1074                 $this->in_seq = array();
1075                 $ymids[0] = array();
1076
1077                 $numer = $xlim - $xoff + $nchunks - 1;
1078                 $x = $xoff;
1079                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1080                         wfProfileIn( __METHOD__ . "-chunk" );
1081                         if ($chunk > 0)
1082                         for ($i = 0; $i <= $this->lcs; $i++)
1083                         $ymids[$i][$chunk-1] = $this->seq[$i];
1084
1085                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1086                         for ( ; $x < $x1; $x++) {
1087                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1088                                 if (empty($ymatches[$line]))
1089                                 continue;
1090                                 $matches = $ymatches[$line];
1091                                 reset($matches);
1092                                 while (list ($junk, $y) = each($matches))
1093                                 if (empty($this->in_seq[$y])) {
1094                                         $k = $this->_lcs_pos($y);
1095                                         USE_ASSERTS && assert($k > 0);
1096                                         $ymids[$k] = $ymids[$k-1];
1097                                         break;
1098                                 }
1099                                 while (list ( /* $junk */, $y) = each($matches)) {
1100                                         if ($y > $this->seq[$k-1]) {
1101                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1102                                                 // Optimization: this is a common case:
1103                                                 //      next match is just replacing previous match.
1104                                                 $this->in_seq[$this->seq[$k]] = false;
1105                                                 $this->seq[$k] = $y;
1106                                                 $this->in_seq[$y] = 1;
1107                                         } else if (empty($this->in_seq[$y])) {
1108                                                 $k = $this->_lcs_pos($y);
1109                                                 USE_ASSERTS && assert($k > 0);
1110                                                 $ymids[$k] = $ymids[$k-1];
1111                                         }
1112                                 }
1113                         }
1114                 }
1115
1116                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1117                 $ymid = $ymids[$this->lcs];
1118                 for ($n = 0; $n < $nchunks - 1; $n++) {
1119                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1120                         $y1 = $ymid[$n] + 1;
1121                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1122                 }
1123                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1124
1125                 return array($this->lcs, $seps);
1126         }
1127
1128         function _lcs_pos ($ypos) {
1129                 $end = $this->lcs;
1130                 if ($end == 0 || $ypos > $this->seq[$end]) {
1131                         $this->seq[++$this->lcs] = $ypos;
1132                         $this->in_seq[$ypos] = 1;
1133                         wfProfileOut( __METHOD__ );
1134                         return $this->lcs;
1135                 }
1136
1137                 $beg = 1;
1138                 while ($beg < $end) {
1139                         $mid = (int)(($beg + $end) / 2);
1140                         if ( $ypos > $this->seq[$mid] )
1141                         $beg = $mid + 1;
1142                         else
1143                         $end = $mid;
1144                 }
1145
1146                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1147
1148                 $this->in_seq[$this->seq[$end]] = false;
1149                 $this->seq[$end] = $ypos;
1150                 $this->in_seq[$ypos] = 1;
1151                 return $end;
1152         }
1153
1154         /* Find LCS of two sequences.
1155          *
1156          * The results are recorded in the vectors $this->{x,y}changed[], by
1157          * storing a 1 in the element for each line that is an insertion
1158          * or deletion (ie. is not in the LCS).
1159          *
1160          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1161          *
1162          * Note that XLIM, YLIM are exclusive bounds.
1163          * All line numbers are origin-0 and discarded lines are not counted.
1164          */
1165         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1166                 // Slide down the bottom initial diagonal.
1167                 while ($xoff < $xlim && $yoff < $ylim
1168                 && $this->xv[$xoff] == $this->yv[$yoff]) {
1169                         ++$xoff;
1170                         ++$yoff;
1171                 }
1172
1173                 // Slide up the top initial diagonal.
1174                 while ($xlim > $xoff && $ylim > $yoff
1175                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1176                         --$xlim;
1177                         --$ylim;
1178                 }
1179
1180                 if ($xoff == $xlim || $yoff == $ylim)
1181                 $lcs = 0;
1182                 else {
1183                         // This is ad hoc but seems to work well.
1184                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1185                         //$nchunks = max(2,min(8,(int)$nchunks));
1186                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1187                         list ($lcs, $seps)
1188                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1189                 }
1190
1191                 if ($lcs == 0) {
1192                         // X and Y sequences have no common subsequence:
1193                         // mark all changed.
1194                         while ($yoff < $ylim)
1195                         $this->ychanged[$this->yind[$yoff++]] = 1;
1196                         while ($xoff < $xlim)
1197                         $this->xchanged[$this->xind[$xoff++]] = 1;
1198                 } else {
1199                         // Use the partitions to split this problem into subproblems.
1200                         reset($seps);
1201                         $pt1 = $seps[0];
1202                         while ($pt2 = next($seps)) {
1203                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1204                                 $pt1 = $pt2;
1205                         }
1206                 }
1207         }
1208
1209         /* Adjust inserts/deletes of identical lines to join changes
1210          * as much as possible.
1211          *
1212          * We do something when a run of changed lines include a
1213          * line at one end and has an excluded, identical line at the other.
1214          * We are free to choose which identical line is included.
1215          * `compareseq' usually chooses the one at the beginning,
1216          * but usually it is cleaner to consider the following identical line
1217          * to be the "change".
1218          *
1219          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1220          */
1221         function _shift_boundaries ($lines, &$changed, $other_changed) {
1222                 wfProfileIn( __METHOD__ );
1223                 $i = 0;
1224                 $j = 0;
1225
1226                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1227                 $len = sizeof($lines);
1228                 $other_len = sizeof($other_changed);
1229
1230                 while (1) {
1231                         /*
1232                          * Scan forwards to find beginning of another run of changes.
1233                          * Also keep track of the corresponding point in the other file.
1234                          *
1235                          * Throughout this code, $i and $j are adjusted together so that
1236                          * the first $i elements of $changed and the first $j elements
1237                          * of $other_changed both contain the same number of zeros
1238                          * (unchanged lines).
1239                          * Furthermore, $j is always kept so that $j == $other_len or
1240                          * $other_changed[$j] == false.
1241                          */
1242                         while ($j < $other_len && $other_changed[$j])
1243                         $j++;
1244
1245                         while ($i < $len && ! $changed[$i]) {
1246                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1247                                 $i++; $j++;
1248                                 while ($j < $other_len && $other_changed[$j])
1249                                 $j++;
1250                         }
1251
1252                         if ($i == $len)
1253                         break;
1254
1255                         $start = $i;
1256
1257                         // Find the end of this run of changes.
1258                         while (++$i < $len && $changed[$i])
1259                         continue;
1260
1261                         do {
1262                                 /*
1263                                  * Record the length of this run of changes, so that
1264                                  * we can later determine whether the run has grown.
1265                                  */
1266                                 $runlength = $i - $start;
1267
1268                                 /*
1269                                  * Move the changed region back, so long as the
1270                                  * previous unchanged line matches the last changed one.
1271                                  * This merges with previous changed regions.
1272                                  */
1273                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1274                                         $changed[--$start] = 1;
1275                                         $changed[--$i] = false;
1276                                         while ($start > 0 && $changed[$start - 1])
1277                                         $start--;
1278                                         USE_ASSERTS && assert('$j > 0');
1279                                         while ($other_changed[--$j])
1280                                         continue;
1281                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1282                                 }
1283
1284                                 /*
1285                                  * Set CORRESPONDING to the end of the changed run, at the last
1286                                  * point where it corresponds to a changed run in the other file.
1287                                  * CORRESPONDING == LEN means no such point has been found.
1288                                  */
1289                                 $corresponding = $j < $other_len ? $i : $len;
1290
1291                                 /*
1292                                  * Move the changed region forward, so long as the
1293                                  * first changed line matches the following unchanged one.
1294                                  * This merges with following changed regions.
1295                                  * Do this second, so that if there are no merges,
1296                                  * the changed region is moved forward as far as possible.
1297                                  */
1298                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1299                                         $changed[$start++] = false;
1300                                         $changed[$i++] = 1;
1301                                         while ($i < $len && $changed[$i])
1302                                         $i++;
1303
1304                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1305                                         $j++;
1306                                         if ($j < $other_len && $other_changed[$j]) {
1307                                                 $corresponding = $i;
1308                                                 while ($j < $other_len && $other_changed[$j])
1309                                                 $j++;
1310                                         }
1311                                 }
1312                         } while ($runlength != $i - $start);
1313
1314                         /*
1315                          * If possible, move the fully-merged run of changes
1316                          * back to a corresponding run in the other file.
1317                          */
1318                         while ($corresponding < $i) {
1319                                 $changed[--$start] = 1;
1320                                 $changed[--$i] = 0;
1321                                 USE_ASSERTS && assert('$j > 0');
1322                                 while ($other_changed[--$j])
1323                                 continue;
1324                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1325                         }
1326                 }
1327                 wfProfileOut( __METHOD__ );
1328         }
1329 }
1330
1331 /**
1332  * Class representing a 'diff' between two sequences of strings.
1333  * @todo document
1334  * @private
1335  * @ingroup DifferenceEngine
1336  */
1337 class Diff
1338 {
1339         var $edits;
1340
1341         /**
1342          * Constructor.
1343          * Computes diff between sequences of strings.
1344          *
1345          * @param $from_lines array An array of strings.
1346          *                (Typically these are lines from a file.)
1347          * @param $to_lines array An array of strings.
1348          */
1349         function Diff($from_lines, $to_lines) {
1350                 $eng = new _DiffEngine;
1351                 $this->edits = $eng->diff($from_lines, $to_lines);
1352                 //$this->_check($from_lines, $to_lines);
1353         }
1354
1355         /**
1356          * Compute reversed Diff.
1357          *
1358          * SYNOPSIS:
1359          *
1360          *      $diff = new Diff($lines1, $lines2);
1361          *      $rev = $diff->reverse();
1362          * @return object A Diff object representing the inverse of the
1363          *                                original diff.
1364          */
1365         function reverse () {
1366                 $rev = $this;
1367                 $rev->edits = array();
1368                 foreach ($this->edits as $edit) {
1369                         $rev->edits[] = $edit->reverse();
1370                 }
1371                 return $rev;
1372         }
1373
1374         /**
1375          * Check for empty diff.
1376          *
1377          * @return bool True iff two sequences were identical.
1378          */
1379         function isEmpty () {
1380                 foreach ($this->edits as $edit) {
1381                         if ($edit->type != 'copy')
1382                         return false;
1383                 }
1384                 return true;
1385         }
1386
1387         /**
1388          * Compute the length of the Longest Common Subsequence (LCS).
1389          *
1390          * This is mostly for diagnostic purposed.
1391          *
1392          * @return int The length of the LCS.
1393          */
1394         function lcs () {
1395                 $lcs = 0;
1396                 foreach ($this->edits as $edit) {
1397                         if ($edit->type == 'copy')
1398                         $lcs += sizeof($edit->orig);
1399                 }
1400                 return $lcs;
1401         }
1402
1403         /**
1404          * Get the original set of lines.
1405          *
1406          * This reconstructs the $from_lines parameter passed to the
1407          * constructor.
1408          *
1409          * @return array The original sequence of strings.
1410          */
1411         function orig() {
1412                 $lines = array();
1413
1414                 foreach ($this->edits as $edit) {
1415                         if ($edit->orig)
1416                         array_splice($lines, sizeof($lines), 0, $edit->orig);
1417                 }
1418                 return $lines;
1419         }
1420
1421         /**
1422          * Get the closing set of lines.
1423          *
1424          * This reconstructs the $to_lines parameter passed to the
1425          * constructor.
1426          *
1427          * @return array The sequence of strings.
1428          */
1429         function closing() {
1430                 $lines = array();
1431
1432                 foreach ($this->edits as $edit) {
1433                         if ($edit->closing)
1434                         array_splice($lines, sizeof($lines), 0, $edit->closing);
1435                 }
1436                 return $lines;
1437         }
1438
1439         /**
1440          * Check a Diff for validity.
1441          *
1442          * This is here only for debugging purposes.
1443          */
1444         function _check ($from_lines, $to_lines) {
1445                 wfProfileIn( __METHOD__ );
1446                 if (serialize($from_lines) != serialize($this->orig()))
1447                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1448                 if (serialize($to_lines) != serialize($this->closing()))
1449                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1450
1451                 $rev = $this->reverse();
1452                 if (serialize($to_lines) != serialize($rev->orig()))
1453                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
1454                 if (serialize($from_lines) != serialize($rev->closing()))
1455                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1456
1457
1458                 $prevtype = 'none';
1459                 foreach ($this->edits as $edit) {
1460                         if ( $prevtype == $edit->type )
1461                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1462                         $prevtype = $edit->type;
1463                 }
1464
1465                 $lcs = $this->lcs();
1466                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1467                 wfProfileOut( __METHOD__ );
1468         }
1469 }
1470
1471 /**
1472  * @todo document, bad name.
1473  * @private
1474  * @ingroup DifferenceEngine
1475  */
1476 class MappedDiff extends Diff
1477 {
1478         /**
1479          * Constructor.
1480          *
1481          * Computes diff between sequences of strings.
1482          *
1483          * This can be used to compute things like
1484          * case-insensitve diffs, or diffs which ignore
1485          * changes in white-space.
1486          *
1487          * @param $from_lines array An array of strings.
1488          *      (Typically these are lines from a file.)
1489          *
1490          * @param $to_lines array An array of strings.
1491          *
1492          * @param $mapped_from_lines array This array should
1493          *      have the same size number of elements as $from_lines.
1494          *      The elements in $mapped_from_lines and
1495          *      $mapped_to_lines are what is actually compared
1496          *      when computing the diff.
1497          *
1498          * @param $mapped_to_lines array This array should
1499          *      have the same number of elements as $to_lines.
1500          */
1501         function MappedDiff($from_lines, $to_lines,
1502         $mapped_from_lines, $mapped_to_lines) {
1503                 wfProfileIn( __METHOD__ );
1504
1505                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1506                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1507
1508                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1509
1510                 $xi = $yi = 0;
1511                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1512                         $orig = &$this->edits[$i]->orig;
1513                         if (is_array($orig)) {
1514                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1515                                 $xi += sizeof($orig);
1516                         }
1517
1518                         $closing = &$this->edits[$i]->closing;
1519                         if (is_array($closing)) {
1520                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1521                                 $yi += sizeof($closing);
1522                         }
1523                 }
1524                 wfProfileOut( __METHOD__ );
1525         }
1526 }
1527
1528 /**
1529  * A class to format Diffs
1530  *
1531  * This class formats the diff in classic diff format.
1532  * It is intended that this class be customized via inheritance,
1533  * to obtain fancier outputs.
1534  * @todo document
1535  * @private
1536  * @ingroup DifferenceEngine
1537  */
1538 class DiffFormatter {
1539         /**
1540          * Number of leading 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 $leading_context_lines = 0;
1546
1547         /**
1548          * Number of trailing context "lines" to preserve.
1549          *
1550          * This should be left at zero for this class, but subclasses
1551          * may want to set this to other values.
1552          */
1553         var $trailing_context_lines = 0;
1554
1555         /**
1556          * Format a diff.
1557          *
1558          * @param $diff object A Diff object.
1559          * @return string The formatted output.
1560          */
1561         function format($diff) {
1562                 wfProfileIn( __METHOD__ );
1563
1564                 $xi = $yi = 1;
1565                 $block = false;
1566                 $context = array();
1567
1568                 $nlead = $this->leading_context_lines;
1569                 $ntrail = $this->trailing_context_lines;
1570
1571                 $this->_start_diff();
1572
1573                 foreach ($diff->edits as $edit) {
1574                         if ($edit->type == 'copy') {
1575                                 if (is_array($block)) {
1576                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1577                                                 $block[] = $edit;
1578                                         }
1579                                         else{
1580                                                 if ($ntrail) {
1581                                                         $context = array_slice($edit->orig, 0, $ntrail);
1582                                                         $block[] = new _DiffOp_Copy($context);
1583                                                 }
1584                                                 $this->_block($x0, $ntrail + $xi - $x0,
1585                                                 $y0, $ntrail + $yi - $y0,
1586                                                 $block);
1587                                                 $block = false;
1588                                         }
1589                                 }
1590                                 $context = $edit->orig;
1591                         }
1592                         else {
1593                                 if (! is_array($block)) {
1594                                         $context = array_slice($context, sizeof($context) - $nlead);
1595                                         $x0 = $xi - sizeof($context);
1596                                         $y0 = $yi - sizeof($context);
1597                                         $block = array();
1598                                         if ($context)
1599                                         $block[] = new _DiffOp_Copy($context);
1600                                 }
1601                                 $block[] = $edit;
1602                         }
1603
1604                         if ($edit->orig)
1605                         $xi += sizeof($edit->orig);
1606                         if ($edit->closing)
1607                         $yi += sizeof($edit->closing);
1608                 }
1609
1610                 if (is_array($block))
1611                 $this->_block($x0, $xi - $x0,
1612                 $y0, $yi - $y0,
1613                 $block);
1614
1615                 $end = $this->_end_diff();
1616                 wfProfileOut( __METHOD__ );
1617                 return $end;
1618         }
1619
1620         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1621                 wfProfileIn( __METHOD__ );
1622                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1623                 foreach ($edits as $edit) {
1624                         if ($edit->type == 'copy')
1625                         $this->_context($edit->orig);
1626                         elseif ($edit->type == 'add')
1627                         $this->_added($edit->closing);
1628                         elseif ($edit->type == 'delete')
1629                         $this->_deleted($edit->orig);
1630                         elseif ($edit->type == 'change')
1631                         $this->_changed($edit->orig, $edit->closing);
1632                         else
1633                         trigger_error('Unknown edit type', E_USER_ERROR);
1634                 }
1635                 $this->_end_block();
1636                 wfProfileOut( __METHOD__ );
1637         }
1638
1639         function _start_diff() {
1640                 ob_start();
1641         }
1642
1643         function _end_diff() {
1644                 $val = ob_get_contents();
1645                 ob_end_clean();
1646                 return $val;
1647         }
1648
1649         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1650                 if ($xlen > 1)
1651                 $xbeg .= "," . ($xbeg + $xlen - 1);
1652                 if ($ylen > 1)
1653                 $ybeg .= "," . ($ybeg + $ylen - 1);
1654
1655                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1656         }
1657
1658         function _start_block($header) {
1659                 echo $header . "\n";
1660         }
1661
1662         function _end_block() {
1663         }
1664
1665         function _lines($lines, $prefix = ' ') {
1666                 foreach ($lines as $line)
1667                 echo "$prefix $line\n";
1668         }
1669
1670         function _context($lines) {
1671                 $this->_lines($lines);
1672         }
1673
1674         function _added($lines) {
1675                 $this->_lines($lines, '>');
1676         }
1677         function _deleted($lines) {
1678                 $this->_lines($lines, '<');
1679         }
1680
1681         function _changed($orig, $closing) {
1682                 $this->_deleted($orig);
1683                 echo "---\n";
1684                 $this->_added($closing);
1685         }
1686 }
1687
1688 /**
1689  * A formatter that outputs unified diffs
1690  * @ingroup DifferenceEngine
1691  */
1692
1693 class UnifiedDiffFormatter extends DiffFormatter {
1694         var $leading_context_lines = 2;
1695         var $trailing_context_lines = 2;
1696
1697         function _added($lines) {
1698                 $this->_lines($lines, '+');
1699         }
1700         function _deleted($lines) {
1701                 $this->_lines($lines, '-');
1702         }
1703         function _changed($orig, $closing) {
1704                 $this->_deleted($orig);
1705                 $this->_added($closing);
1706         }
1707         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1708                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1709         }
1710 }
1711
1712 /**
1713  * A pseudo-formatter that just passes along the Diff::$edits array
1714  * @ingroup DifferenceEngine
1715  */
1716 class ArrayDiffFormatter extends DiffFormatter {
1717         function format($diff) {
1718                 $oldline = 1;
1719                 $newline = 1;
1720                 $retval = array();
1721                 foreach($diff->edits as $edit)
1722                 switch($edit->type) {
1723                         case 'add':
1724                                 foreach($edit->closing as $l) {
1725                                         $retval[] = array(
1726                                                         'action' => 'add',
1727                                                         'new'=> $l,
1728                                                         'newline' => $newline++
1729                                         );
1730                                 }
1731                                 break;
1732                         case 'delete':
1733                                 foreach($edit->orig as $l) {
1734                                         $retval[] = array(
1735                                                         'action' => 'delete',
1736                                                         'old' => $l,
1737                                                         'oldline' => $oldline++,
1738                                         );
1739                                 }
1740                                 break;
1741                         case 'change':
1742                                 foreach($edit->orig as $i => $l) {
1743                                         $retval[] = array(
1744                                                         'action' => 'change',
1745                                                         'old' => $l,
1746                                                         'new' => @$edit->closing[$i],
1747                                                         'oldline' => $oldline++,
1748                                                         'newline' => $newline++,
1749                                         );
1750                                 }
1751                                 break;
1752                         case 'copy':
1753                                 $oldline += count($edit->orig);
1754                                 $newline += count($edit->orig);
1755                 }
1756                 return $retval;
1757         }
1758 }
1759
1760 /**
1761  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1762  *
1763  */
1764
1765 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1766
1767 /**
1768  * @todo document
1769  * @private
1770  * @ingroup DifferenceEngine
1771  */
1772 class _HWLDF_WordAccumulator {
1773         function _HWLDF_WordAccumulator () {
1774                 $this->_lines = array();
1775                 $this->_line = '';
1776                 $this->_group = '';
1777                 $this->_tag = '';
1778         }
1779
1780         function _flushGroup ($new_tag) {
1781                 if ($this->_group !== '') {
1782                         if ($this->_tag == 'ins')
1783                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1784                         htmlspecialchars ( $this->_group ) . '</ins>';
1785                         elseif ($this->_tag == 'del')
1786                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1787                         htmlspecialchars ( $this->_group ) . '</del>';
1788                         else
1789                         $this->_line .= htmlspecialchars ( $this->_group );
1790                 }
1791                 $this->_group = '';
1792                 $this->_tag = $new_tag;
1793         }
1794
1795         function _flushLine ($new_tag) {
1796                 $this->_flushGroup($new_tag);
1797                 if ($this->_line != '')
1798                 array_push ( $this->_lines, $this->_line );
1799                 else
1800                 # make empty lines visible by inserting an NBSP
1801                 array_push ( $this->_lines, NBSP );
1802                 $this->_line = '';
1803         }
1804
1805         function addWords ($words, $tag = '') {
1806                 if ($tag != $this->_tag)
1807                 $this->_flushGroup($tag);
1808
1809                 foreach ($words as $word) {
1810                         // new-line should only come as first char of word.
1811                         if ($word == '')
1812                         continue;
1813                         if ($word[0] == "\n") {
1814                                 $this->_flushLine($tag);
1815                                 $word = substr($word, 1);
1816                         }
1817                         assert(!strstr($word, "\n"));
1818                         $this->_group .= $word;
1819                 }
1820         }
1821
1822         function getLines() {
1823                 $this->_flushLine('~done');
1824                 return $this->_lines;
1825         }
1826 }
1827
1828 /**
1829  * @todo document
1830  * @private
1831  * @ingroup DifferenceEngine
1832  */
1833 class WordLevelDiff extends MappedDiff {
1834         const MAX_LINE_LENGTH = 10000;
1835
1836         function WordLevelDiff ($orig_lines, $closing_lines) {
1837                 wfProfileIn( __METHOD__ );
1838
1839                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1840                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1841
1842                 $this->MappedDiff($orig_words, $closing_words,
1843                 $orig_stripped, $closing_stripped);
1844                 wfProfileOut( __METHOD__ );
1845         }
1846
1847         function _split($lines) {
1848                 wfProfileIn( __METHOD__ );
1849
1850                 $words = array();
1851                 $stripped = array();
1852                 $first = true;
1853                 foreach ( $lines as $line ) {
1854                         # If the line is too long, just pretend the entire line is one big word
1855                         # This prevents resource exhaustion problems
1856                         if ( $first ) {
1857                                 $first = false;
1858                         } else {
1859                                 $words[] = "\n";
1860                                 $stripped[] = "\n";
1861                         }
1862                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1863                                 $words[] = $line;
1864                                 $stripped[] = $line;
1865                         } else {
1866                                 $m = array();
1867                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1868                                 $line, $m))
1869                                 {
1870                                         $words = array_merge( $words, $m[0] );
1871                                         $stripped = array_merge( $stripped, $m[1] );
1872                                 }
1873                         }
1874                 }
1875                 wfProfileOut( __METHOD__ );
1876                 return array($words, $stripped);
1877         }
1878
1879         function orig () {
1880                 wfProfileIn( __METHOD__ );
1881                 $orig = new _HWLDF_WordAccumulator;
1882
1883                 foreach ($this->edits as $edit) {
1884                         if ($edit->type == 'copy')
1885                         $orig->addWords($edit->orig);
1886                         elseif ($edit->orig)
1887                         $orig->addWords($edit->orig, 'del');
1888                 }
1889                 $lines = $orig->getLines();
1890                 wfProfileOut( __METHOD__ );
1891                 return $lines;
1892         }
1893
1894         function closing () {
1895                 wfProfileIn( __METHOD__ );
1896                 $closing = new _HWLDF_WordAccumulator;
1897
1898                 foreach ($this->edits as $edit) {
1899                         if ($edit->type == 'copy')
1900                         $closing->addWords($edit->closing);
1901                         elseif ($edit->closing)
1902                         $closing->addWords($edit->closing, 'ins');
1903                 }
1904                 $lines = $closing->getLines();
1905                 wfProfileOut( __METHOD__ );
1906                 return $lines;
1907         }
1908 }
1909
1910 /**
1911  * Wikipedia Table style diff formatter.
1912  * @todo document
1913  * @private
1914  * @ingroup DifferenceEngine
1915  */
1916 class TableDiffFormatter extends DiffFormatter {
1917         function TableDiffFormatter() {
1918                 $this->leading_context_lines = 2;
1919                 $this->trailing_context_lines = 2;
1920         }
1921
1922         public static function escapeWhiteSpace( $msg ) {
1923                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
1924                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
1925                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
1926                 return $msg;
1927         }
1928
1929         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1930                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1931                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1932                 return $r;
1933         }
1934
1935         function _start_block( $header ) {
1936                 echo $header;
1937         }
1938
1939         function _end_block() {
1940         }
1941
1942         function _lines( $lines, $prefix=' ', $color='white' ) {
1943         }
1944
1945         # HTML-escape parameter before calling this
1946         function addedLine( $line ) {
1947                 return $this->wrapLine( '+', 'diff-addedline', $line );
1948         }
1949
1950         # HTML-escape parameter before calling this
1951         function deletedLine( $line ) {
1952                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1953         }
1954
1955         # HTML-escape parameter before calling this
1956         function contextLine( $line ) {
1957                 return $this->wrapLine( ' ', 'diff-context', $line );
1958         }
1959
1960         private function wrapLine( $marker, $class, $line ) {
1961                 if( $line !== '' ) {
1962                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1963                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1964                 }
1965                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1966         }
1967
1968         function emptyLine() {
1969                 return '<td colspan="2">&nbsp;</td>';
1970         }
1971
1972         function _added( $lines ) {
1973                 foreach ($lines as $line) {
1974                         echo '<tr>' . $this->emptyLine() .
1975                         $this->addedLine( '<ins class="diffchange">' .
1976                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1977                 }
1978         }
1979
1980         function _deleted($lines) {
1981                 foreach ($lines as $line) {
1982                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1983                         htmlspecialchars ( $line ) . '</del>' ) .
1984                         $this->emptyLine() . "</tr>\n";
1985                 }
1986         }
1987
1988         function _context( $lines ) {
1989                 foreach ($lines as $line) {
1990                         echo '<tr>' .
1991                         $this->contextLine( htmlspecialchars ( $line ) ) .
1992                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1993                 }
1994         }
1995
1996         function _changed( $orig, $closing ) {
1997                 wfProfileIn( __METHOD__ );
1998
1999                 $diff = new WordLevelDiff( $orig, $closing );
2000                 $del = $diff->orig();
2001                 $add = $diff->closing();
2002
2003                 # Notice that WordLevelDiff returns HTML-escaped output.
2004                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2005
2006                 while ( $line = array_shift( $del ) ) {
2007                         $aline = array_shift( $add );
2008                         echo '<tr>' . $this->deletedLine( $line ) .
2009                         $this->addedLine( $aline ) . "</tr>\n";
2010                 }
2011                 foreach ($add as $line) {       # If any leftovers
2012                         echo '<tr>' . $this->emptyLine() .
2013                         $this->addedLine( $line ) . "</tr>\n";
2014                 }
2015                 wfProfileOut( __METHOD__ );
2016         }
2017 }