includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * @defgroup DifferenceEngine DifferenceEngine
   4  *
   5  * @file
   6  * See diff.doc
   7  * @todo indicate where diff.doc can be found.
   8  */
   9
  10 /**
  11  * Constant to indicate diff cache compatibility.
  12  * Bump this when changing the diff formatting in a way that
  13  * fixes important bugs or such to force cached diff views to
  14  * clear.
  15  */
  16 define( 'MW_DIFF_VERSION', '1.11a' );
  17
  18 /**
  19  * @todo document
  20  * @ingroup DifferenceEngine
  21  */
  22 class DifferenceEngine {
  23         /**#@+
  24          * @private
  25          */
  26         var $mOldid, $mNewid, $mTitle;
  27         var $mOldtitle, $mNewtitle, $mPagetitle;
  28         var $mOldtext, $mNewtext;
  29         var $mOldPage, $mNewPage;
  30         var $mRcidMarkPatrolled;
  31         var $mOldRev, $mNewRev;
  32         var $mRevisionsLoaded = false; // Have the revisions been loaded
  33         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  34         /**#@-*/
  35
  36         /**
  37          * Constructor
  38          * @param $titleObj Title object that the diff is associated with
  39          * @param $old Integer: old ID we want to show and diff with.
  40          * @param $new String: either 'prev' or 'next'.
  41          * @param $rcid Integer: ??? FIXME (default 0)
  42          * @param $refreshCache boolean If set, refreshes the diff cache
  43          */
  44         function __construct( $titleObj = null, $old = 0, $new = 0, $rcid = 0, $refreshCache = false ) {
  45                 $this->mTitle = $titleObj;
  46                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  47
  48                 if ( 'prev' === $new ) {
  49                         # Show diff between revision $old and the previous one.
  50                         # Get previous one from DB.
  51                         #
  52                         $this->mNewid = intval($old);
  53
  54                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  55
  56                 } elseif ( 'next' === $new ) {
  57                         # Show diff between revision $old and the previous one.
  58                         # Get previous one from DB.
  59                         #
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  62                         if ( false === $this->mNewid ) {
  63                                 # if no result, NewId points to the newest old revision. The only newer
  64                                 # revision is cur, which is "0".
  65                                 $this->mNewid = 0;
  66                         }
  67
  68                 } else {
  69                         $this->mOldid = intval($old);
  70                         $this->mNewid = intval($new);
  71                 }
  72                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  73                 $this->mRefreshCache = $refreshCache;
  74         }
  75
  76         function getTitle() {
  77                 return $this->mTitle;
  78         }
  79
  80         function showDiffPage( $diffOnly = false ) {
  81                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  82                 wfProfileIn( __METHOD__ );
  83
  84                 # If external diffs are enabled both globally and for the user,
  85                 # we'll use the application/x-external-editor interface to call
  86                 # an external diff tool like kompare, kdiff3, etc.
  87                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  88                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  89                         $wgOut->disable();
  90                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  91                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  92                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  93                         $special=$wgLang->getNsText(NS_SPECIAL);
  94                         $control=<<<CONTROL
  95 [Process]
  96 Type=Diff text
  97 Engine=MediaWiki
  98 Script={$wgServer}{$wgScript}
  99 Special namespace={$special}
 100
 101 [File]
 102 Extension=wiki
 103 URL=$url1
 104
 105 [File 2]
 106 Extension=wiki
 107 URL=$url2
 108 CONTROL;
 109                         echo($control);
 110                         return;
 111                 }
 112
 113                 $wgOut->setArticleFlag( false );
 114                 if ( ! $this->loadRevisionData() ) {
 115                         $t = $this->mTitle->getPrefixedText();
 116                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 117                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 118                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 119                         wfProfileOut( __METHOD__ );
 120                         return;
 121                 }
 122
 123                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 124
 125                 if ( $this->mNewRev->isCurrent() ) {
 126                         $wgOut->setArticleFlag( true );
 127                 }
 128
 129                 # mOldid is false if the difference engine is called with a "vague" query for
 130                 # a diff between a version V and its previous version V' AND the version V
 131                 # is the first version of that article. In that case, V' does not exist.
 132                 if ( $this->mOldid === false ) {
 133                         $this->showFirstRevision();
 134                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 135                         wfProfileOut( __METHOD__ );
 136                         return;
 137                 }
 138
 139                 $wgOut->suppressQuickbar();
 140
 141                 $oldTitle = $this->mOldPage->getPrefixedText();
 142                 $newTitle = $this->mNewPage->getPrefixedText();
 143                 if( $oldTitle == $newTitle ) {
 144                         $wgOut->setPageTitle( $newTitle );
 145                 } else {
 146                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 147                 }
 148                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 149                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 150
 151                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 152                         $wgOut->loginToUse();
 153                         $wgOut->output();
 154                         wfProfileOut( __METHOD__ );
 155                         exit;
 156                 }
 157
 158                 $sk = $wgUser->getSkin();
 159
 160                 // Check if page is editable
 161                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 162                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 163                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 164                 } else {
 165                         $rollback = '';
 166                 }
 167
 168                 // Prepare a change patrol link, if applicable
 169                 if( $wgUseRCPatrol && $wgUser->isAllowed( 'patrol' ) ) {
 170                         // If we've been given an explicit change identifier, use it; saves time
 171                         if( $this->mRcidMarkPatrolled ) {
 172                                 $rcid = $this->mRcidMarkPatrolled;
 173                         } else {
 174                                 // Look for an unpatrolled change corresponding to this diff
 175                                 $db = wfGetDB( DB_SLAVE );
 176                                 $change = RecentChange::newFromConds(
 177                                         array(
 178                                                 // Add redundant user,timestamp condition so we can use the existing index
 179                                                 'rc_user_text'  => $this->mNewRev->getRawUserText(),
 180                                                 'rc_timestamp' => $db->timestamp( $this->mNewRev->getTimestamp() ),
 181                                                 'rc_this_oldid' => $this->mNewid,
 182                                                 'rc_last_oldid' => $this->mOldid,
 183                                                 'rc_patrolled' => 0
 184                                         ),
 185                                         __METHOD__
 186                                 );
 187                                 if( $change instanceof RecentChange ) {
 188                                         $rcid = $change->mAttribs['rc_id'];
 189                                 } else {
 190                                         // None found
 191                                         $rcid = 0;
 192                                 }
 193                         }
 194                         // Build the link
 195                         if( $rcid ) {
 196                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj(
 197                                         $this->mTitle,
 198                                         wfMsgHtml( 'markaspatrolleddiff' ),
 199                                         "action=markpatrolled&rcid={$rcid}"
 200                                 ) . ']</span>';
 201                         } else {
 202                                 $patrol = '';
 203                         }
 204                 } else {
 205                         $patrol = '';
 206                 }
 207
 208                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 209                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 210                 if ( $this->mNewRev->isCurrent() ) {
 211                         $nextlink = '&nbsp;';
 212                 } else {
 213                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 214                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 215                 }
 216
 217                 $oldminor = '';
 218                 $newminor = '';
 219
 220                 if ($this->mOldRev->mMinorEdit == 1) {
 221                         $oldminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 222                 }
 223
 224                 if ($this->mNewRev->mMinorEdit == 1) {
 225                         $newminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 226                 }
 227
 228                 $rdel = ''; $ldel = '';
 229                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 230                         $revdel = SpecialPage::getTitleFor( 'Revisiondelete' );
 231                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 232                         // If revision was hidden from sysops
 233                                 $ldel = wfMsgHtml('rev-delundel');
 234                         } else {
 235                                 $ldel = $sk->makeKnownLinkObj( $revdel,
 236                                         wfMsgHtml('rev-delundel'),
 237                                         'target=' . urlencode( $this->mOldRev->mTitle->getPrefixedDbkey() ) .
 238                                         '&oldid=' . urlencode( $this->mOldRev->getId() ) );
 239                                 // Bolden oversighted content
 240                                 if( $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 241                                         $ldel = "<strong>$ldel</strong>";
 242                         }
 243                         $ldel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$ldel</small>)</tt> ";
 244                         // We don't currently handle well changing the top revision's settings
 245                         if( $this->mNewRev->isCurrent() ) {
 246                         // If revision was hidden from sysops
 247                                 $rdel = wfMsgHtml('rev-delundel');
 248                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 249                         // If revision was hidden from sysops
 250                                 $rdel = wfMsgHtml('rev-delundel');
 251                         } else {
 252                                 $rdel = $sk->makeKnownLinkObj( $revdel,
 253                                         wfMsgHtml('rev-delundel'),
 254                                         'target=' . urlencode( $this->mNewRev->mTitle->getPrefixedDbkey() ) .
 255                                         '&oldid=' . urlencode( $this->mNewRev->getId() ) );
 256                                 // Bolden oversighted content
 257                                 if( $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 258                                         $rdel = "<strong>$rdel</strong>";
 259                         }
 260                         $rdel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$rdel</small>)</tt> ";
 261                 }
 262
 263                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 264                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, true ) . "</div>" .
 265                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, true ) . $ldel . "</div>" .
 266                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 267                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 268                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, true ) . " $rollback</div>" .
 269                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, true ) . $rdel . "</div>" .
 270                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 271
 272                 $this->showDiff( $oldHeader, $newHeader );
 273
 274                 if ( !$diffOnly )
 275                         $this->renderNewRevision();
 276
 277                 wfProfileOut( __METHOD__ );
 278         }
 279
 280         /**
 281          * Show the new revision of the page.
 282          */
 283         function renderNewRevision() {
 284                 global $wgOut;
 285                 wfProfileIn( __METHOD__ );
 286
 287                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 288                 #add deleted rev tag if needed
 289                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 290                         $wgOut->addWikiMsg( 'rev-deleted-text-permission' );
 291                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 292                         $wgOut->addWikiMsg( 'rev-deleted-text-view' );
 293                 }
 294
 295                 if( !$this->mNewRev->isCurrent() ) {
 296                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 297                 }
 298
 299                 $this->loadNewText();
 300                 if( is_object( $this->mNewRev ) ) {
 301                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 302                 }
 303
 304                 if ($this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage()) {
 305                         // Stolen from Article::view --AG 2007-10-11
 306
 307                         // Give hooks a chance to customise the output
 308                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 309                                 // Wrap the whole lot in a <pre> and don't parse
 310                                 $m = array();
 311                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 312                                 $wgOut->addHtml( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 313                                 $wgOut->addHtml( htmlspecialchars( $this->mNewtext ) );
 314                                 $wgOut->addHtml( "\n</pre>\n" );
 315                         }
 316                 } else
 317                         $wgOut->addWikiTextTidy( $this->mNewtext );
 318
 319                 if( !$this->mNewRev->isCurrent() ) {
 320                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 321                 }
 322
 323                 wfProfileOut( __METHOD__ );
 324         }
 325
 326         /**
 327          * Show the first revision of an article. Uses normal diff headers in
 328          * contrast to normal "old revision" display style.
 329          */
 330         function showFirstRevision() {
 331                 global $wgOut, $wgUser;
 332                 wfProfileIn( __METHOD__ );
 333
 334                 # Get article text from the DB
 335                 #
 336                 if ( ! $this->loadNewText() ) {
 337                         $t = $this->mTitle->getPrefixedText();
 338                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 339                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 340                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 341                         wfProfileOut( __METHOD__ );
 342                         return;
 343                 }
 344                 if ( $this->mNewRev->isCurrent() ) {
 345                         $wgOut->setArticleFlag( true );
 346                 }
 347
 348                 # Check if user is allowed to look at this page. If not, bail out.
 349                 #
 350                 if ( !( $this->mTitle->userCanRead() ) ) {
 351                         $wgOut->loginToUse();
 352                         $wgOut->output();
 353                         wfProfileOut( __METHOD__ );
 354                         exit;
 355                 }
 356
 357                 # Prepare the header box
 358                 #
 359                 $sk = $wgUser->getSkin();
 360
 361                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 362                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 363                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 364                         $sk->revComment( $this->mNewRev ) . "<br />" .
 365                         $nextlink . "</div>\n";
 366
 367                 $wgOut->addHTML( $header );
 368
 369                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 370                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 371
 372                 wfProfileOut( __METHOD__ );
 373         }
 374
 375         /**
 376          * Get the diff text, send it to $wgOut
 377          * Returns false if the diff could not be generated, otherwise returns true
 378          */
 379         function showDiff( $otitle, $ntitle ) {
 380                 global $wgOut;
 381                 $diff = $this->getDiff( $otitle, $ntitle );
 382                 if ( $diff === false ) {
 383                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 384                         return false;
 385                 } else {
 386                         $this->showDiffStyle();
 387                         $wgOut->addHTML( $diff );
 388                         return true;
 389                 }
 390         }
 391
 392         /**
 393          * Add style sheets and supporting JS for diff display.
 394          */
 395         function showDiffStyle() {
 396                 global $wgStylePath, $wgStyleVersion, $wgOut;
 397                 $wgOut->addStyle( 'common/diff.css' );
 398
 399                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 400                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 401         }
 402
 403         /**
 404          * Get complete diff table, including header
 405          *
 406          * @param Title $otitle Old title
 407          * @param Title $ntitle New title
 408          * @return mixed
 409          */
 410         function getDiff( $otitle, $ntitle ) {
 411                 $body = $this->getDiffBody();
 412                 if ( $body === false ) {
 413                         return false;
 414                 } else {
 415                         $multi = $this->getMultiNotice();
 416                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 417                 }
 418         }
 419
 420         /**
 421          * Get the diff table body, without header
 422          *
 423          * @return mixed
 424          */
 425         function getDiffBody() {
 426                 global $wgMemc;
 427                 wfProfileIn( __METHOD__ );
 428                 // Check if the diff should be hidden from this user
 429                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 430                         return '';
 431                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 432                         return '';
 433                 }
 434                 // Cacheable?
 435                 $key = false;
 436                 if ( $this->mOldid && $this->mNewid ) {
 437                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 438                         // Try cache
 439                         if ( !$this->mRefreshCache ) {
 440                                 $difftext = $wgMemc->get( $key );
 441                                 if ( $difftext ) {
 442                                         wfIncrStats( 'diff_cache_hit' );
 443                                         $difftext = $this->localiseLineNumbers( $difftext );
 444                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 445                                         wfProfileOut( __METHOD__ );
 446                                         return $difftext;
 447                                 }
 448                         } // don't try to load but save the result
 449                 }
 450
 451                 // Loadtext is permission safe, this just clears out the diff
 452                 if ( !$this->loadText() ) {
 453                         wfProfileOut( __METHOD__ );
 454                         return false;
 455                 }
 456
 457                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 458
 459                 // Save to cache for 7 days
 460                 if ( $key !== false && $difftext !== false ) {
 461                         wfIncrStats( 'diff_cache_miss' );
 462                         $wgMemc->set( $key, $difftext, 7*86400 );
 463                 } else {
 464                         wfIncrStats( 'diff_uncacheable' );
 465                 }
 466                 // Replace line numbers with the text in the user's language
 467                 if ( $difftext !== false ) {
 468                         $difftext = $this->localiseLineNumbers( $difftext );
 469                 }
 470                 wfProfileOut( __METHOD__ );
 471                 return $difftext;
 472         }
 473
 474         /**
 475          * Generate a diff, no caching
 476          * $otext and $ntext must be already segmented
 477          */
 478         function generateDiffBody( $otext, $ntext ) {
 479                 global $wgExternalDiffEngine, $wgContLang;
 480
 481                 $otext = str_replace( "\r\n", "\n", $otext );
 482                 $ntext = str_replace( "\r\n", "\n", $ntext );
 483
 484                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 485                         # For historical reasons, external diff engine expects
 486                         # input text to be HTML-escaped already
 487                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 488                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 489                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 490                                 dl('php_wikidiff.so');
 491                         }
 492                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 493                                 $this->debug( 'wikidiff1' );
 494                 }
 495
 496                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 497                         # Better external diff engine, the 2 may some day be dropped
 498                         # This one does the escaping and segmenting itself
 499                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 500                                 wfProfileIn( __METHOD__ . "-dl" );
 501                                 @dl('php_wikidiff2.so');
 502                                 wfProfileOut( __METHOD__ . "-dl" );
 503                         }
 504                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 505                                 wfProfileIn( 'wikidiff2_do_diff' );
 506                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 507                                 $text .= $this->debug( 'wikidiff2' );
 508                                 wfProfileOut( 'wikidiff2_do_diff' );
 509                                 return $text;
 510                         }
 511                 }
 512                 if ( $wgExternalDiffEngine !== false ) {
 513                         # Diff via the shell
 514                         global $wgTmpDirectory;
 515                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 516                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 517
 518                         $tempFile1 = fopen( $tempName1, "w" );
 519                         if ( !$tempFile1 ) {
 520                                 wfProfileOut( __METHOD__ );
 521                                 return false;
 522                         }
 523                         $tempFile2 = fopen( $tempName2, "w" );
 524                         if ( !$tempFile2 ) {
 525                                 wfProfileOut( __METHOD__ );
 526                                 return false;
 527                         }
 528                         fwrite( $tempFile1, $otext );
 529                         fwrite( $tempFile2, $ntext );
 530                         fclose( $tempFile1 );
 531                         fclose( $tempFile2 );
 532                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 533                         wfProfileIn( __METHOD__ . "-shellexec" );
 534                         $difftext = wfShellExec( $cmd );
 535                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 536                         wfProfileOut( __METHOD__ . "-shellexec" );
 537                         unlink( $tempName1 );
 538                         unlink( $tempName2 );
 539                         return $difftext;
 540                 }
 541
 542                 # Native PHP diff
 543                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 544                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 545                 $diffs = new Diff( $ota, $nta );
 546                 $formatter = new TableDiffFormatter();
 547                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 548                         $this->debug();
 549         }
 550
 551         /**
 552          * Generate a debug comment indicating diff generating time,
 553          * server node, and generator backend.
 554          */
 555         protected function debug( $generator="internal" ) {
 556                 global $wgShowHostnames, $wgNodeName;
 557                 $data = array( $generator );
 558                 if( $wgShowHostnames ) {
 559                         $data[] = $wgNodeName;
 560                 }
 561                 $data[] = wfTimestamp( TS_DB );
 562                 return "<!-- diff generator: " .
 563                         implode( " ",
 564                                 array_map(
 565                                         "htmlspecialchars",
 566                                                 $data ) ) .
 567                         " -->\n";
 568         }
 569
 570         /**
 571          * Replace line numbers with the text in the user's language
 572          */
 573         function localiseLineNumbers( $text ) {
 574                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 575                         array( &$this, 'localiseLineNumbersCb' ), $text );
 576         }
 577
 578         function localiseLineNumbersCb( $matches ) {
 579                 global $wgLang;
 580                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 581         }
 582
 583
 584         /**
 585          * If there are revisions between the ones being compared, return a note saying so.
 586          */
 587         function getMultiNotice() {
 588                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 589                         return '';
 590
 591                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 592                         // Comparing two different pages? Count would be meaningless.
 593                         return '';
 594                 }
 595
 596                 $oldid = $this->mOldRev->getId();
 597                 $newid = $this->mNewRev->getId();
 598                 if ( $oldid > $newid ) {
 599                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 600                 }
 601
 602                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 603                 if ( !$n )
 604                         return '';
 605
 606                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 607         }
 608
 609
 610         /**
 611          * Add the header to a diff body
 612          */
 613         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 614                 global $wgOut;
 615
 616                 $header = "
 617                         <table class='diff'>
 618                         <col class='diff-marker' />
 619                         <col class='diff-content' />
 620                         <col class='diff-marker' />
 621                         <col class='diff-content' />
 622                         <tr valign='top'>
 623                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 624                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 625                         </tr>
 626                 ";
 627
 628                 if ( $multi != '' )
 629                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 630
 631                 return $header . $diff . "</table>";
 632         }
 633
 634         /**
 635          * Use specified text instead of loading from the database
 636          */
 637         function setText( $oldText, $newText ) {
 638                 $this->mOldtext = $oldText;
 639                 $this->mNewtext = $newText;
 640                 $this->mTextLoaded = 2;
 641         }
 642
 643         /**
 644          * Load revision metadata for the specified articles. If newid is 0, then compare
 645          * the old article in oldid to the current article; if oldid is 0, then
 646          * compare the current article to the immediately previous one (ignoring the
 647          * value of newid).
 648          *
 649          * If oldid is false, leave the corresponding revision object set
 650          * to false. This is impossible via ordinary user input, and is provided for
 651          * API convenience.
 652          */
 653         function loadRevisionData() {
 654                 global $wgLang;
 655                 if ( $this->mRevisionsLoaded ) {
 656                         return true;
 657                 } else {
 658                         // Whether it succeeds or fails, we don't want to try again
 659                         $this->mRevisionsLoaded = true;
 660                 }
 661
 662                 // Load the new revision object
 663                 $this->mNewRev = $this->mNewid
 664                         ? Revision::newFromId( $this->mNewid )
 665                         : Revision::newFromTitle( $this->mTitle );
 666                 if( !$this->mNewRev instanceof Revision )
 667                         return false;
 668
 669                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 670                 $this->mNewid = $this->mNewRev->getId();
 671
 672                 // Check if page is editable
 673                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 674
 675                 // Set assorted variables
 676                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 677                 $this->mNewPage = $this->mNewRev->getTitle();
 678                 if( $this->mNewRev->isCurrent() ) {
 679                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 680                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 681                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 682
 683                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)";
 684                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 685
 686                 } else {
 687                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 688                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 689                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 690
 691                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 692                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 693                 }
 694                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 695                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 696                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 697                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 698                 }
 699
 700                 // Load the old revision object
 701                 $this->mOldRev = false;
 702                 if( $this->mOldid ) {
 703                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 704                 } elseif ( $this->mOldid === 0 ) {
 705                         $rev = $this->mNewRev->getPrevious();
 706                         if( $rev ) {
 707                                 $this->mOldid = $rev->getId();
 708                                 $this->mOldRev = $rev;
 709                         } else {
 710                                 // No previous revision; mark to show as first-version only.
 711                                 $this->mOldid = false;
 712                                 $this->mOldRev = false;
 713                         }
 714                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 715
 716                 if( is_null( $this->mOldRev ) ) {
 717                         return false;
 718                 }
 719
 720                 if ( $this->mOldRev ) {
 721                         $this->mOldPage = $this->mOldRev->getTitle();
 722
 723                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 724                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 725                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 726                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 727
 728                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 729                                 . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 730                         // Add an "undo" link
 731                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 732                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 733                                 $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 734                         }
 735
 736                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 737                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 738                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 739                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 740                         }
 741                 }
 742
 743                 return true;
 744         }
 745
 746         /**
 747          * Load the text of the revisions, as well as revision data.
 748          */
 749         function loadText() {
 750                 if ( $this->mTextLoaded == 2 ) {
 751                         return true;
 752                 } else {
 753                         // Whether it succeeds or fails, we don't want to try again
 754                         $this->mTextLoaded = 2;
 755                 }
 756
 757                 if ( !$this->loadRevisionData() ) {
 758                         return false;
 759                 }
 760                 if ( $this->mOldRev ) {
 761                         $this->mOldtext = $this->mOldRev->revText();
 762                         if ( $this->mOldtext === false ) {
 763                                 return false;
 764                         }
 765                 }
 766                 if ( $this->mNewRev ) {
 767                         $this->mNewtext = $this->mNewRev->revText();
 768                         if ( $this->mNewtext === false ) {
 769                                 return false;
 770                         }
 771                 }
 772                 return true;
 773         }
 774
 775         /**
 776          * Load the text of the new revision, not the old one
 777          */
 778         function loadNewText() {
 779                 if ( $this->mTextLoaded >= 1 ) {
 780                         return true;
 781                 } else {
 782                         $this->mTextLoaded = 1;
 783                 }
 784                 if ( !$this->loadRevisionData() ) {
 785                         return false;
 786                 }
 787                 $this->mNewtext = $this->mNewRev->getText();
 788                 return true;
 789         }
 790
 791
 792 }
 793
 794 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 795 //
 796 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 797 // You may copy this code freely under the conditions of the GPL.
 798 //
 799
 800 define('USE_ASSERTS', function_exists('assert'));
 801
 802 /**
 803  * @todo document
 804  * @private
 805  * @ingroup DifferenceEngine
 806  */
 807 class _DiffOp {
 808         var $type;
 809         var $orig;
 810         var $closing;
 811
 812         function reverse() {
 813                 trigger_error('pure virtual', E_USER_ERROR);
 814         }
 815
 816         function norig() {
 817                 return $this->orig ? sizeof($this->orig) : 0;
 818         }
 819
 820         function nclosing() {
 821                 return $this->closing ? sizeof($this->closing) : 0;
 822         }
 823 }
 824
 825 /**
 826  * @todo document
 827  * @private
 828  * @ingroup DifferenceEngine
 829  */
 830 class _DiffOp_Copy extends _DiffOp {
 831         var $type = 'copy';
 832
 833         function _DiffOp_Copy ($orig, $closing = false) {
 834                 if (!is_array($closing))
 835                         $closing = $orig;
 836                 $this->orig = $orig;
 837                 $this->closing = $closing;
 838         }
 839
 840         function reverse() {
 841                 return new _DiffOp_Copy($this->closing, $this->orig);
 842         }
 843 }
 844
 845 /**
 846  * @todo document
 847  * @private
 848  * @ingroup DifferenceEngine
 849  */
 850 class _DiffOp_Delete extends _DiffOp {
 851         var $type = 'delete';
 852
 853         function _DiffOp_Delete ($lines) {
 854                 $this->orig = $lines;
 855                 $this->closing = false;
 856         }
 857
 858         function reverse() {
 859                 return new _DiffOp_Add($this->orig);
 860         }
 861 }
 862
 863 /**
 864  * @todo document
 865  * @private
 866  * @ingroup DifferenceEngine
 867  */
 868 class _DiffOp_Add extends _DiffOp {
 869         var $type = 'add';
 870
 871         function _DiffOp_Add ($lines) {
 872                 $this->closing = $lines;
 873                 $this->orig = false;
 874         }
 875
 876         function reverse() {
 877                 return new _DiffOp_Delete($this->closing);
 878         }
 879 }
 880
 881 /**
 882  * @todo document
 883  * @private
 884  * @ingroup DifferenceEngine
 885  */
 886 class _DiffOp_Change extends _DiffOp {
 887         var $type = 'change';
 888
 889         function _DiffOp_Change ($orig, $closing) {
 890                 $this->orig = $orig;
 891                 $this->closing = $closing;
 892         }
 893
 894         function reverse() {
 895                 return new _DiffOp_Change($this->closing, $this->orig);
 896         }
 897 }
 898
 899
 900 /**
 901  * Class used internally by Diff to actually compute the diffs.
 902  *
 903  * The algorithm used here is mostly lifted from the perl module
 904  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 905  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 906  *
 907  * More ideas are taken from:
 908  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 909  *
 910  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 911  * diffutils-2.7, which can be found at:
 912  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 913  *
 914  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 915  * are my own.
 916  *
 917  * Line length limits for robustness added by Tim Starling, 2005-08-31
 918  *
 919  * @author Geoffrey T. Dairiki, Tim Starling
 920  * @private
 921  * @ingroup DifferenceEngine
 922  */
 923 class _DiffEngine {
 924         const MAX_XREF_LENGTH =  10000;
 925
 926         function diff ($from_lines, $to_lines) {
 927                 wfProfileIn( __METHOD__ );
 928
 929                 $n_from = sizeof($from_lines);
 930                 $n_to = sizeof($to_lines);
 931
 932                 $this->xchanged = $this->ychanged = array();
 933                 $this->xv = $this->yv = array();
 934                 $this->xind = $this->yind = array();
 935                 unset($this->seq);
 936                 unset($this->in_seq);
 937                 unset($this->lcs);
 938
 939                 // Skip leading common lines.
 940                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 941                         if ($from_lines[$skip] !== $to_lines[$skip])
 942                                 break;
 943                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 944                 }
 945                 // Skip trailing common lines.
 946                 $xi = $n_from; $yi = $n_to;
 947                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 948                         if ($from_lines[$xi] !== $to_lines[$yi])
 949                                 break;
 950                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 951                 }
 952
 953                 // Ignore lines which do not exist in both files.
 954                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 955                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 956                 }
 957
 958                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 959                         $line = $to_lines[$yi];
 960                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 961                                 continue;
 962                         $yhash[$this->_line_hash($line)] = 1;
 963                         $this->yv[] = $line;
 964                         $this->yind[] = $yi;
 965                 }
 966                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 967                         $line = $from_lines[$xi];
 968                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 969                                 continue;
 970                         $this->xv[] = $line;
 971                         $this->xind[] = $xi;
 972                 }
 973
 974                 // Find the LCS.
 975                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 976
 977                 // Merge edits when possible
 978                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 979                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 980
 981                 // Compute the edit operations.
 982                 $edits = array();
 983                 $xi = $yi = 0;
 984                 while ($xi < $n_from || $yi < $n_to) {
 985                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 986                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 987
 988                         // Skip matching "snake".
 989                         $copy = array();
 990                         while ( $xi < $n_from && $yi < $n_to
 991                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 992                                 $copy[] = $from_lines[$xi++];
 993                                 ++$yi;
 994                         }
 995                         if ($copy)
 996                                 $edits[] = new _DiffOp_Copy($copy);
 997
 998                         // Find deletes & adds.
 999                         $delete = array();
1000                         while ($xi < $n_from && $this->xchanged[$xi])
1001                                 $delete[] = $from_lines[$xi++];
1002
1003                         $add = array();
1004                         while ($yi < $n_to && $this->ychanged[$yi])
1005                                 $add[] = $to_lines[$yi++];
1006
1007                         if ($delete && $add)
1008                                 $edits[] = new _DiffOp_Change($delete, $add);
1009                         elseif ($delete)
1010                                 $edits[] = new _DiffOp_Delete($delete);
1011                         elseif ($add)
1012                                 $edits[] = new _DiffOp_Add($add);
1013                 }
1014                 wfProfileOut( __METHOD__ );
1015                 return $edits;
1016         }
1017
1018         /**
1019          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1020          */
1021         function _line_hash( $line ) {
1022                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1023                         return md5( $line );
1024                 } else {
1025                         return $line;
1026                 }
1027         }
1028
1029
1030         /* Divide the Largest Common Subsequence (LCS) of the sequences
1031          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1032          * sized segments.
1033          *
1034          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1035          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1036          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1037          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1038          * that (X0, Y0) == (XOFF, YOFF) and
1039          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1040          *
1041          * This function assumes that the first lines of the specified portions
1042          * of the two files do not match, and likewise that the last lines do not
1043          * match.  The caller must trim matching lines from the beginning and end
1044          * of the portions it is going to specify.
1045          */
1046         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1047                 wfProfileIn( __METHOD__ );
1048                 $flip = false;
1049
1050                 if ($xlim - $xoff > $ylim - $yoff) {
1051                         // Things seems faster (I'm not sure I understand why)
1052                                 // when the shortest sequence in X.
1053                                 $flip = true;
1054                         list ($xoff, $xlim, $yoff, $ylim)
1055                         = array( $yoff, $ylim, $xoff, $xlim);
1056                 }
1057
1058                 if ($flip)
1059                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1060                                 $ymatches[$this->xv[$i]][] = $i;
1061                 else
1062                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1063                                 $ymatches[$this->yv[$i]][] = $i;
1064
1065                 $this->lcs = 0;
1066                 $this->seq[0]= $yoff - 1;
1067                 $this->in_seq = array();
1068                 $ymids[0] = array();
1069
1070                 $numer = $xlim - $xoff + $nchunks - 1;
1071                 $x = $xoff;
1072                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1073                         wfProfileIn( __METHOD__ . "-chunk" );
1074                         if ($chunk > 0)
1075                                 for ($i = 0; $i <= $this->lcs; $i++)
1076                                         $ymids[$i][$chunk-1] = $this->seq[$i];
1077
1078                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1079                         for ( ; $x < $x1; $x++) {
1080                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1081                                         if (empty($ymatches[$line]))
1082                                                 continue;
1083                                 $matches = $ymatches[$line];
1084                                 reset($matches);
1085                                 while (list ($junk, $y) = each($matches))
1086                                         if (empty($this->in_seq[$y])) {
1087                                                 $k = $this->_lcs_pos($y);
1088                                                 USE_ASSERTS && assert($k > 0);
1089                                                 $ymids[$k] = $ymids[$k-1];
1090                                                 break;
1091                                         }
1092                                 while (list ( /* $junk */, $y) = each($matches)) {
1093                                         if ($y > $this->seq[$k-1]) {
1094                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1095                                                 // Optimization: this is a common case:
1096                                                 //      next match is just replacing previous match.
1097                                                 $this->in_seq[$this->seq[$k]] = false;
1098                                                 $this->seq[$k] = $y;
1099                                                 $this->in_seq[$y] = 1;
1100                                         } else if (empty($this->in_seq[$y])) {
1101                                                 $k = $this->_lcs_pos($y);
1102                                                 USE_ASSERTS && assert($k > 0);
1103                                                 $ymids[$k] = $ymids[$k-1];
1104                                         }
1105                                 }
1106                         }
1107                         wfProfileOut( __METHOD__ . "-chunk" );
1108                 }
1109
1110                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1111                 $ymid = $ymids[$this->lcs];
1112                 for ($n = 0; $n < $nchunks - 1; $n++) {
1113                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1114                         $y1 = $ymid[$n] + 1;
1115                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1116                 }
1117                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1118
1119                 wfProfileOut( __METHOD__ );
1120                 return array($this->lcs, $seps);
1121         }
1122
1123         function _lcs_pos ($ypos) {
1124                 wfProfileIn( __METHOD__ );
1125
1126                 $end = $this->lcs;
1127                 if ($end == 0 || $ypos > $this->seq[$end]) {
1128                         $this->seq[++$this->lcs] = $ypos;
1129                         $this->in_seq[$ypos] = 1;
1130                         wfProfileOut( __METHOD__ );
1131                         return $this->lcs;
1132                 }
1133
1134                 $beg = 1;
1135                 while ($beg < $end) {
1136                         $mid = (int)(($beg + $end) / 2);
1137                         if ( $ypos > $this->seq[$mid] )
1138                                 $beg = $mid + 1;
1139                         else
1140                                 $end = $mid;
1141                 }
1142
1143                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1144
1145                 $this->in_seq[$this->seq[$end]] = false;
1146                 $this->seq[$end] = $ypos;
1147                 $this->in_seq[$ypos] = 1;
1148                 wfProfileOut( __METHOD__ );
1149                 return $end;
1150         }
1151
1152         /* Find LCS of two sequences.
1153          *
1154          * The results are recorded in the vectors $this->{x,y}changed[], by
1155          * storing a 1 in the element for each line that is an insertion
1156          * or deletion (ie. is not in the LCS).
1157          *
1158          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1159          *
1160          * Note that XLIM, YLIM are exclusive bounds.
1161          * All line numbers are origin-0 and discarded lines are not counted.
1162          */
1163         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1164                 wfProfileIn( __METHOD__ );
1165
1166                 // Slide down the bottom initial diagonal.
1167                 while ($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 = $pt2;
1205                         }
1206                 }
1207                 wfProfileOut( __METHOD__ );
1208         }
1209
1210         /* Adjust inserts/deletes of identical lines to join changes
1211          * as much as possible.
1212          *
1213          * We do something when a run of changed lines include a
1214          * line at one end and has an excluded, identical line at the other.
1215          * We are free to choose which identical line is included.
1216          * `compareseq' usually chooses the one at the beginning,
1217          * but usually it is cleaner to consider the following identical line
1218          * to be the "change".
1219          *
1220          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1221          */
1222         function _shift_boundaries ($lines, &$changed, $other_changed) {
1223                 wfProfileIn( __METHOD__ );
1224                 $i = 0;
1225                 $j = 0;
1226
1227                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1228                 $len = sizeof($lines);
1229                 $other_len = sizeof($other_changed);
1230
1231                 while (1) {
1232                         /*
1233                          * Scan forwards to find beginning of another run of changes.
1234                          * Also keep track of the corresponding point in the other file.
1235                          *
1236                          * Throughout this code, $i and $j are adjusted together so that
1237                          * the first $i elements of $changed and the first $j elements
1238                          * of $other_changed both contain the same number of zeros
1239                          * (unchanged lines).
1240                          * Furthermore, $j is always kept so that $j == $other_len or
1241                          * $other_changed[$j] == false.
1242                          */
1243                         while ($j < $other_len && $other_changed[$j])
1244                                 $j++;
1245
1246                         while ($i < $len && ! $changed[$i]) {
1247                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1248                                 $i++; $j++;
1249                                 while ($j < $other_len && $other_changed[$j])
1250                                         $j++;
1251                         }
1252
1253                         if ($i == $len)
1254                                 break;
1255
1256                         $start = $i;
1257
1258                         // Find the end of this run of changes.
1259                         while (++$i < $len && $changed[$i])
1260                                 continue;
1261
1262                         do {
1263                                 /*
1264                                  * Record the length of this run of changes, so that
1265                                  * we can later determine whether the run has grown.
1266                                  */
1267                                 $runlength = $i - $start;
1268
1269                                 /*
1270                                  * Move the changed region back, so long as the
1271                                  * previous unchanged line matches the last changed one.
1272                                  * This merges with previous changed regions.
1273                                  */
1274                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1275                                         $changed[--$start] = 1;
1276                                         $changed[--$i] = false;
1277                                         while ($start > 0 && $changed[$start - 1])
1278                                                 $start--;
1279                                         USE_ASSERTS && assert('$j > 0');
1280                                         while ($other_changed[--$j])
1281                                                 continue;
1282                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1283                                 }
1284
1285                                 /*
1286                                  * Set CORRESPONDING to the end of the changed run, at the last
1287                                  * point where it corresponds to a changed run in the other file.
1288                                  * CORRESPONDING == LEN means no such point has been found.
1289                                  */
1290                                 $corresponding = $j < $other_len ? $i : $len;
1291
1292                                 /*
1293                                  * Move the changed region forward, so long as the
1294                                  * first changed line matches the following unchanged one.
1295                                  * This merges with following changed regions.
1296                                  * Do this second, so that if there are no merges,
1297                                  * the changed region is moved forward as far as possible.
1298                                  */
1299                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1300                                         $changed[$start++] = false;
1301                                         $changed[$i++] = 1;
1302                                         while ($i < $len && $changed[$i])
1303                                                 $i++;
1304
1305                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1306                                         $j++;
1307                                         if ($j < $other_len && $other_changed[$j]) {
1308                                                 $corresponding = $i;
1309                                                 while ($j < $other_len && $other_changed[$j])
1310                                                         $j++;
1311                                         }
1312                                 }
1313                         } while ($runlength != $i - $start);
1314
1315                         /*
1316                          * If possible, move the fully-merged run of changes
1317                          * back to a corresponding run in the other file.
1318                          */
1319                         while ($corresponding < $i) {
1320                                 $changed[--$start] = 1;
1321                                 $changed[--$i] = 0;
1322                                 USE_ASSERTS && assert('$j > 0');
1323                                 while ($other_changed[--$j])
1324                                         continue;
1325                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1326                         }
1327                 }
1328                 wfProfileOut( __METHOD__ );
1329         }
1330 }
1331
1332 /**
1333  * Class representing a 'diff' between two sequences of strings.
1334  * @todo document
1335  * @private
1336  * @ingroup DifferenceEngine
1337  */
1338 class Diff
1339 {
1340         var $edits;
1341
1342         /**
1343          * Constructor.
1344          * Computes diff between sequences of strings.
1345          *
1346          * @param $from_lines array An array of strings.
1347          *                (Typically these are lines from a file.)
1348          * @param $to_lines array An array of strings.
1349          */
1350         function Diff($from_lines, $to_lines) {
1351                 $eng = new _DiffEngine;
1352                 $this->edits = $eng->diff($from_lines, $to_lines);
1353                 //$this->_check($from_lines, $to_lines);
1354         }
1355
1356         /**
1357          * Compute reversed Diff.
1358          *
1359          * SYNOPSIS:
1360          *
1361          *      $diff = new Diff($lines1, $lines2);
1362          *      $rev = $diff->reverse();
1363          * @return object A Diff object representing the inverse of the
1364          *                                original diff.
1365          */
1366         function reverse () {
1367                 $rev = $this;
1368                 $rev->edits = array();
1369                 foreach ($this->edits as $edit) {
1370                         $rev->edits[] = $edit->reverse();
1371                 }
1372                 return $rev;
1373         }
1374
1375         /**
1376          * Check for empty diff.
1377          *
1378          * @return bool True iff two sequences were identical.
1379          */
1380         function isEmpty () {
1381                 foreach ($this->edits as $edit) {
1382                         if ($edit->type != 'copy')
1383                                 return false;
1384                 }
1385                 return true;
1386         }
1387
1388         /**
1389          * Compute the length of the Longest Common Subsequence (LCS).
1390          *
1391          * This is mostly for diagnostic purposed.
1392          *
1393          * @return int The length of the LCS.
1394          */
1395         function lcs () {
1396                 $lcs = 0;
1397                 foreach ($this->edits as $edit) {
1398                         if ($edit->type == 'copy')
1399                                 $lcs += sizeof($edit->orig);
1400                 }
1401                 return $lcs;
1402         }
1403
1404         /**
1405          * Get the original set of lines.
1406          *
1407          * This reconstructs the $from_lines parameter passed to the
1408          * constructor.
1409          *
1410          * @return array The original sequence of strings.
1411          */
1412         function orig() {
1413                 $lines = array();
1414
1415                 foreach ($this->edits as $edit) {
1416                         if ($edit->orig)
1417                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1418                 }
1419                 return $lines;
1420         }
1421
1422         /**
1423          * Get the closing set of lines.
1424          *
1425          * This reconstructs the $to_lines parameter passed to the
1426          * constructor.
1427          *
1428          * @return array The sequence of strings.
1429          */
1430         function closing() {
1431                 $lines = array();
1432
1433                 foreach ($this->edits as $edit) {
1434                         if ($edit->closing)
1435                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1436                 }
1437                 return $lines;
1438         }
1439
1440         /**
1441          * Check a Diff for validity.
1442          *
1443          * This is here only for debugging purposes.
1444          */
1445         function _check ($from_lines, $to_lines) {
1446                 wfProfileIn( __METHOD__ );
1447                 if (serialize($from_lines) != serialize($this->orig()))
1448                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1449                 if (serialize($to_lines) != serialize($this->closing()))
1450                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1451
1452                 $rev = $this->reverse();
1453                 if (serialize($to_lines) != serialize($rev->orig()))
1454                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1455                 if (serialize($from_lines) != serialize($rev->closing()))
1456                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1457
1458
1459                 $prevtype = 'none';
1460                 foreach ($this->edits as $edit) {
1461                         if ( $prevtype == $edit->type )
1462                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1463                         $prevtype = $edit->type;
1464                 }
1465
1466                 $lcs = $this->lcs();
1467                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1468                 wfProfileOut( __METHOD__ );
1469         }
1470 }
1471
1472 /**
1473  * @todo document, bad name.
1474  * @private
1475  * @ingroup DifferenceEngine
1476  */
1477 class MappedDiff extends Diff
1478 {
1479         /**
1480          * Constructor.
1481          *
1482          * Computes diff between sequences of strings.
1483          *
1484          * This can be used to compute things like
1485          * case-insensitve diffs, or diffs which ignore
1486          * changes in white-space.
1487          *
1488          * @param $from_lines array An array of strings.
1489          *      (Typically these are lines from a file.)
1490          *
1491          * @param $to_lines array An array of strings.
1492          *
1493          * @param $mapped_from_lines array This array should
1494          *      have the same size number of elements as $from_lines.
1495          *      The elements in $mapped_from_lines and
1496          *      $mapped_to_lines are what is actually compared
1497          *      when computing the diff.
1498          *
1499          * @param $mapped_to_lines array This array should
1500          *      have the same number of elements as $to_lines.
1501          */
1502         function MappedDiff($from_lines, $to_lines,
1503                 $mapped_from_lines, $mapped_to_lines) {
1504                 wfProfileIn( __METHOD__ );
1505
1506                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1507                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1508
1509                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1510
1511                 $xi = $yi = 0;
1512                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1513                         $orig = &$this->edits[$i]->orig;
1514                         if (is_array($orig)) {
1515                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1516                                 $xi += sizeof($orig);
1517                         }
1518
1519                         $closing = &$this->edits[$i]->closing;
1520                         if (is_array($closing)) {
1521                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1522                                 $yi += sizeof($closing);
1523                         }
1524                 }
1525                 wfProfileOut( __METHOD__ );
1526         }
1527 }
1528
1529 /**
1530  * A class to format Diffs
1531  *
1532  * This class formats the diff in classic diff format.
1533  * It is intended that this class be customized via inheritance,
1534  * to obtain fancier outputs.
1535  * @todo document
1536  * @private
1537  * @ingroup DifferenceEngine
1538  */
1539 class DiffFormatter {
1540         /**
1541          * Number of leading context "lines" to preserve.
1542          *
1543          * This should be left at zero for this class, but subclasses
1544          * may want to set this to other values.
1545          */
1546         var $leading_context_lines = 0;
1547
1548         /**
1549          * Number of trailing context "lines" to preserve.
1550          *
1551          * This should be left at zero for this class, but subclasses
1552          * may want to set this to other values.
1553          */
1554         var $trailing_context_lines = 0;
1555
1556         /**
1557          * Format a diff.
1558          *
1559          * @param $diff object A Diff object.
1560          * @return string The formatted output.
1561          */
1562         function format($diff) {
1563                 wfProfileIn( __METHOD__ );
1564
1565                 $xi = $yi = 1;
1566                 $block = false;
1567                 $context = array();
1568
1569                 $nlead = $this->leading_context_lines;
1570                 $ntrail = $this->trailing_context_lines;
1571
1572                 $this->_start_diff();
1573
1574                 foreach ($diff->edits as $edit) {
1575                         if ($edit->type == 'copy') {
1576                                 if (is_array($block)) {
1577                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1578                                                 $block[] = $edit;
1579                                         }
1580                                         else{
1581                                                 if ($ntrail) {
1582                                                         $context = array_slice($edit->orig, 0, $ntrail);
1583                                                         $block[] = new _DiffOp_Copy($context);
1584                                                 }
1585                                                 $this->_block($x0, $ntrail + $xi - $x0,
1586                                                                           $y0, $ntrail + $yi - $y0,
1587                                                                           $block);
1588                                                 $block = false;
1589                                         }
1590                                 }
1591                                 $context = $edit->orig;
1592                         }
1593                         else {
1594                                 if (! is_array($block)) {
1595                                         $context = array_slice($context, sizeof($context) - $nlead);
1596                                         $x0 = $xi - sizeof($context);
1597                                         $y0 = $yi - sizeof($context);
1598                                         $block = array();
1599                                         if ($context)
1600                                                 $block[] = new _DiffOp_Copy($context);
1601                                 }
1602                                 $block[] = $edit;
1603                         }
1604
1605                         if ($edit->orig)
1606                                 $xi += sizeof($edit->orig);
1607                         if ($edit->closing)
1608                                 $yi += sizeof($edit->closing);
1609                 }
1610
1611                 if (is_array($block))
1612                         $this->_block($x0, $xi - $x0,
1613                                                   $y0, $yi - $y0,
1614                                                   $block);
1615
1616                 $end = $this->_end_diff();
1617                 wfProfileOut( __METHOD__ );
1618                 return $end;
1619         }
1620
1621         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1622                 wfProfileIn( __METHOD__ );
1623                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1624                 foreach ($edits as $edit) {
1625                         if ($edit->type == 'copy')
1626                                 $this->_context($edit->orig);
1627                         elseif ($edit->type == 'add')
1628                                 $this->_added($edit->closing);
1629                         elseif ($edit->type == 'delete')
1630                                 $this->_deleted($edit->orig);
1631                         elseif ($edit->type == 'change')
1632                                 $this->_changed($edit->orig, $edit->closing);
1633                         else
1634                                 trigger_error('Unknown edit type', E_USER_ERROR);
1635                 }
1636                 $this->_end_block();
1637                 wfProfileOut( __METHOD__ );
1638         }
1639
1640         function _start_diff() {
1641                 ob_start();
1642         }
1643
1644         function _end_diff() {
1645                 $val = ob_get_contents();
1646                 ob_end_clean();
1647                 return $val;
1648         }
1649
1650         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1651                 if ($xlen > 1)
1652                         $xbeg .= "," . ($xbeg + $xlen - 1);
1653                 if ($ylen > 1)
1654                         $ybeg .= "," . ($ybeg + $ylen - 1);
1655
1656                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1657         }
1658
1659         function _start_block($header) {
1660                 echo $header . "\n";
1661         }
1662
1663         function _end_block() {
1664         }
1665
1666         function _lines($lines, $prefix = ' ') {
1667                 foreach ($lines as $line)
1668                         echo "$prefix $line\n";
1669         }
1670
1671         function _context($lines) {
1672                 $this->_lines($lines);
1673         }
1674
1675         function _added($lines) {
1676                 $this->_lines($lines, '>');
1677         }
1678         function _deleted($lines) {
1679                 $this->_lines($lines, '<');
1680         }
1681
1682         function _changed($orig, $closing) {
1683                 $this->_deleted($orig);
1684                 echo "---\n";
1685                 $this->_added($closing);
1686         }
1687 }
1688
1689 /**
1690  * A formatter that outputs unified diffs
1691  * @ingroup DifferenceEngine
1692  */
1693
1694 class UnifiedDiffFormatter extends DiffFormatter {
1695         var $leading_context_lines = 2;
1696         var $trailing_context_lines = 2;
1697
1698         function _added($lines) {
1699                 $this->_lines($lines, '+');
1700         }
1701         function _deleted($lines) {
1702                 $this->_lines($lines, '-');
1703         }
1704         function _changed($orig, $closing) {
1705                 $this->_deleted($orig);
1706                 $this->_added($closing);
1707         }
1708         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1709                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1710         }
1711 }
1712
1713 /**
1714  * A pseudo-formatter that just passes along the Diff::$edits array
1715  * @ingroup DifferenceEngine
1716  */
1717 class ArrayDiffFormatter extends DiffFormatter {
1718         function format($diff) {
1719                 $oldline = 1;
1720                 $newline = 1;
1721                 $retval = array();
1722                 foreach($diff->edits as $edit)
1723                         switch($edit->type) {
1724                                 case 'add':
1725                                         foreach($edit->closing as $l) {
1726                                                 $retval[] = array(
1727                                                         'action' => 'add',
1728                                                         'new'=> $l,
1729                                                         'newline' => $newline++
1730                                                 );
1731                                         }
1732                                         break;
1733                                 case 'delete':
1734                                         foreach($edit->orig as $l) {
1735                                                 $retval[] = array(
1736                                                         'action' => 'delete',
1737                                                         'old' => $l,
1738                                                         'oldline' => $oldline++,
1739                                                 );
1740                                         }
1741                                         break;
1742                                 case 'change':
1743                                         foreach($edit->orig as $i => $l) {
1744                                                 $retval[] = array(
1745                                                         'action' => 'change',
1746                                                         'old' => $l,
1747                                                         'new' => @$edit->closing[$i],
1748                                                         'oldline' => $oldline++,
1749                                                         'newline' => $newline++,
1750                                                 );
1751                                         }
1752                                         break;
1753                                 case 'copy':
1754                                         $oldline += count($edit->orig);
1755                                         $newline += count($edit->orig);
1756                         }
1757                 return $retval;
1758         }
1759 }
1760
1761 /**
1762  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1763  *
1764  */
1765
1766 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1767
1768 /**
1769  * @todo document
1770  * @private
1771  * @ingroup DifferenceEngine
1772  */
1773 class _HWLDF_WordAccumulator {
1774         function _HWLDF_WordAccumulator () {
1775                 $this->_lines = array();
1776                 $this->_line = '';
1777                 $this->_group = '';
1778                 $this->_tag = '';
1779         }
1780
1781         function _flushGroup ($new_tag) {
1782                 if ($this->_group !== '') {
1783                         if ($this->_tag == 'ins')
1784                                 $this->_line .= '<ins class="diffchange diffchange-inline">' .
1785                                         htmlspecialchars ( $this->_group ) . '</ins>';
1786                         elseif ($this->_tag == 'del')
1787                                 $this->_line .= '<del class="diffchange diffchange-inline">' .
1788                                         htmlspecialchars ( $this->_group ) . '</del>';
1789                         else
1790                                 $this->_line .= htmlspecialchars ( $this->_group );
1791                 }
1792                 $this->_group = '';
1793                 $this->_tag = $new_tag;
1794         }
1795
1796         function _flushLine ($new_tag) {
1797                 $this->_flushGroup($new_tag);
1798                 if ($this->_line != '')
1799                         array_push ( $this->_lines, $this->_line );
1800                 else
1801                         # make empty lines visible by inserting an NBSP
1802                         array_push ( $this->_lines, NBSP );
1803                 $this->_line = '';
1804         }
1805
1806         function addWords ($words, $tag = '') {
1807                 if ($tag != $this->_tag)
1808                         $this->_flushGroup($tag);
1809
1810                 foreach ($words as $word) {
1811                         // new-line should only come as first char of word.
1812                         if ($word == '')
1813                                 continue;
1814                         if ($word[0] == "\n") {
1815                                 $this->_flushLine($tag);
1816                                 $word = substr($word, 1);
1817                         }
1818                         assert(!strstr($word, "\n"));
1819                         $this->_group .= $word;
1820                 }
1821         }
1822
1823         function getLines() {
1824                 $this->_flushLine('~done');
1825                 return $this->_lines;
1826         }
1827 }
1828
1829 /**
1830  * @todo document
1831  * @private
1832  * @ingroup DifferenceEngine
1833  */
1834 class WordLevelDiff extends MappedDiff {
1835         const MAX_LINE_LENGTH = 10000;
1836
1837         function WordLevelDiff ($orig_lines, $closing_lines) {
1838                 wfProfileIn( __METHOD__ );
1839
1840                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1841                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1842
1843                 $this->MappedDiff($orig_words, $closing_words,
1844                         $orig_stripped, $closing_stripped);
1845                 wfProfileOut( __METHOD__ );
1846         }
1847
1848         function _split($lines) {
1849                 wfProfileIn( __METHOD__ );
1850
1851                 $words = array();
1852                 $stripped = array();
1853                 $first = true;
1854                 foreach ( $lines as $line ) {
1855                         # If the line is too long, just pretend the entire line is one big word
1856                         # This prevents resource exhaustion problems
1857                         if ( $first ) {
1858                                 $first = false;
1859                         } else {
1860                                 $words[] = "\n";
1861                                 $stripped[] = "\n";
1862                         }
1863                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1864                                 $words[] = $line;
1865                                 $stripped[] = $line;
1866                         } else {
1867                                 $m = array();
1868                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1869                                         $line, $m))
1870                                 {
1871                                         $words = array_merge( $words, $m[0] );
1872                                         $stripped = array_merge( $stripped, $m[1] );
1873                                 }
1874                         }
1875                 }
1876                 wfProfileOut( __METHOD__ );
1877                 return array($words, $stripped);
1878         }
1879
1880         function orig () {
1881                 wfProfileIn( __METHOD__ );
1882                 $orig = new _HWLDF_WordAccumulator;
1883
1884                 foreach ($this->edits as $edit) {
1885                         if ($edit->type == 'copy')
1886                                 $orig->addWords($edit->orig);
1887                         elseif ($edit->orig)
1888                                 $orig->addWords($edit->orig, 'del');
1889                 }
1890                 $lines = $orig->getLines();
1891                 wfProfileOut( __METHOD__ );
1892                 return $lines;
1893         }
1894
1895         function closing () {
1896                 wfProfileIn( __METHOD__ );
1897                 $closing = new _HWLDF_WordAccumulator;
1898
1899                 foreach ($this->edits as $edit) {
1900                         if ($edit->type == 'copy')
1901                                 $closing->addWords($edit->closing);
1902                         elseif ($edit->closing)
1903                                 $closing->addWords($edit->closing, 'ins');
1904                 }
1905                 $lines = $closing->getLines();
1906                 wfProfileOut( __METHOD__ );
1907                 return $lines;
1908         }
1909 }
1910
1911 /**
1912  * Wikipedia Table style diff formatter.
1913  * @todo document
1914  * @private
1915  * @ingroup DifferenceEngine
1916  */
1917 class TableDiffFormatter extends DiffFormatter {
1918         function TableDiffFormatter() {
1919                 $this->leading_context_lines = 2;
1920                 $this->trailing_context_lines = 2;
1921         }
1922
1923         public static function escapeWhiteSpace( $msg ) {
1924                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
1925                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
1926                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
1927                 return $msg;
1928         }
1929
1930         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1931                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1932                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1933                 return $r;
1934         }
1935
1936         function _start_block( $header ) {
1937                 echo $header;
1938         }
1939
1940         function _end_block() {
1941         }
1942
1943         function _lines( $lines, $prefix=' ', $color='white' ) {
1944         }
1945
1946         # HTML-escape parameter before calling this
1947         function addedLine( $line ) {
1948                 return $this->wrapLine( '+', 'diff-addedline', $line );
1949         }
1950
1951         # HTML-escape parameter before calling this
1952         function deletedLine( $line ) {
1953                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1954         }
1955
1956         # HTML-escape parameter before calling this
1957         function contextLine( $line ) {
1958                 return $this->wrapLine( ' ', 'diff-context', $line );
1959         }
1960
1961         private function wrapLine( $marker, $class, $line ) {
1962                 if( $line !== '' ) {
1963                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1964                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1965                 }
1966                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1967         }
1968
1969         function emptyLine() {
1970                 return '<td colspan="2">&nbsp;</td>';
1971         }
1972
1973         function _added( $lines ) {
1974                 foreach ($lines as $line) {
1975                         echo '<tr>' . $this->emptyLine() .
1976                                 $this->addedLine( '<ins class="diffchange">' .
1977                                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1978                 }
1979         }
1980
1981         function _deleted($lines) {
1982                 foreach ($lines as $line) {
1983                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1984                                 htmlspecialchars ( $line ) . '</del>' ) .
1985                           $this->emptyLine() . "</tr>\n";
1986                 }
1987         }
1988
1989         function _context( $lines ) {
1990                 foreach ($lines as $line) {
1991                         echo '<tr>' .
1992                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1993                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1994                 }
1995         }
1996
1997         function _changed( $orig, $closing ) {
1998                 wfProfileIn( __METHOD__ );
1999
2000                 $diff = new WordLevelDiff( $orig, $closing );
2001                 $del = $diff->orig();
2002                 $add = $diff->closing();
2003
2004                 # Notice that WordLevelDiff returns HTML-escaped output.
2005                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2006
2007                 while ( $line = array_shift( $del ) ) {
2008                         $aline = array_shift( $add );
2009                         echo '<tr>' . $this->deletedLine( $line ) .
2010                                 $this->addedLine( $aline ) . "</tr>\n";
2011                 }
2012                 foreach ($add as $line) {       # If any leftovers
2013                         echo '<tr>' . $this->emptyLine() .
2014                                 $this->addedLine( $line ) . "</tr>\n";
2015                 }
2016                 wfProfileOut( __METHOD__ );
2017         }
2018 }