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() . ' ' . wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 116                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 117                         $wgOut->addWikiMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 118                         wfProfileOut( __METHOD__ );
 119                         return;
 120                 }
 121
 122                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 123
 124                 if ( $this->mNewRev->isCurrent() ) {
 125                         $wgOut->setArticleFlag( true );
 126                 }
 127
 128                 # mOldid is false if the difference engine is called with a "vague" query for
 129                 # a diff between a version V and its previous version V' AND the version V
 130                 # is the first version of that article. In that case, V' does not exist.
 131                 if ( $this->mOldid === false ) {
 132                         $this->showFirstRevision();
 133                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 134                         wfProfileOut( __METHOD__ );
 135                         return;
 136                 }
 137
 138                 $wgOut->suppressQuickbar();
 139
 140                 $oldTitle = $this->mOldPage->getPrefixedText();
 141                 $newTitle = $this->mNewPage->getPrefixedText();
 142                 if( $oldTitle == $newTitle ) {
 143                         $wgOut->setPageTitle( $newTitle );
 144                 } else {
 145                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 146                 }
 147                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 148                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 149
 150                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 151                         $wgOut->loginToUse();
 152                         $wgOut->output();
 153                         wfProfileOut( __METHOD__ );
 154                         exit;
 155                 }
 156
 157                 $sk = $wgUser->getSkin();
 158
 159                 // Check if page is editable
 160                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 161                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 162                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 163                 } else {
 164                         $rollback = '';
 165                 }
 166
 167                 // Prepare a change patrol link, if applicable
 168                 if( $wgUseRCPatrol && $wgUser->isAllowed( 'patrol' ) ) {
 169                         // If we've been given an explicit change identifier, use it; saves time
 170                         if( $this->mRcidMarkPatrolled ) {
 171                                 $rcid = $this->mRcidMarkPatrolled;
 172                         } else {
 173                                 // Look for an unpatrolled change corresponding to this diff
 174                                 $db = wfGetDB( DB_SLAVE );
 175                                 $change = RecentChange::newFromConds(
 176                                         array(
 177                                                 // Add redundant user,timestamp condition so we can use the existing index
 178                                                 'rc_user_text'  => $this->mNewRev->getRawUserText(),
 179                                                 'rc_timestamp' => $db->timestamp( $this->mNewRev->getTimestamp() ),
 180                                                 'rc_this_oldid' => $this->mNewid,
 181                                                 'rc_last_oldid' => $this->mOldid,
 182                                                 'rc_patrolled' => 0
 183                                         ),
 184                                         __METHOD__
 185                                 );
 186                                 if( $change instanceof RecentChange ) {
 187                                         $rcid = $change->mAttribs['rc_id'];
 188                                 } else {
 189                                         // None found
 190                                         $rcid = 0;
 191                                 }
 192                         }
 193                         // Build the link
 194                         if( $rcid ) {
 195                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj(
 196                                         $this->mTitle,
 197                                         wfMsgHtml( 'markaspatrolleddiff' ),
 198                                         "action=markpatrolled&rcid={$rcid}"
 199                                 ) . ']</span>';
 200                         } else {
 201                                 $patrol = '';
 202                         }
 203                 } else {
 204                         $patrol = '';
 205                 }
 206
 207                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 208                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 209                 if ( $this->mNewRev->isCurrent() ) {
 210                         $nextlink = '&nbsp;';
 211                 } else {
 212                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 213                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 214                 }
 215
 216                 $oldminor = '';
 217                 $newminor = '';
 218
 219                 if ($this->mOldRev->mMinorEdit == 1) {
 220                         $oldminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 221                 }
 222
 223                 if ($this->mNewRev->mMinorEdit == 1) {
 224                         $newminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 225                 }
 226
 227                 $rdel = ''; $ldel = '';
 228                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 229                         $revdel = SpecialPage::getTitleFor( 'Revisiondelete' );
 230                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 231                         // If revision was hidden from sysops
 232                                 $ldel = wfMsgHtml('rev-delundel');
 233                         } else {
 234                                 $ldel = $sk->makeKnownLinkObj( $revdel,
 235                                         wfMsgHtml('rev-delundel'),
 236                                         'target=' . urlencode( $this->mOldRev->mTitle->getPrefixedDbkey() ) .
 237                                         '&oldid=' . urlencode( $this->mOldRev->getId() ) );
 238                                 // Bolden oversighted content
 239                                 if( $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 240                                         $ldel = "<strong>$ldel</strong>";
 241                         }
 242                         $ldel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$ldel</small>)</tt> ";
 243                         // We don't currently handle well changing the top revision's settings
 244                         if( $this->mNewRev->isCurrent() ) {
 245                         // If revision was hidden from sysops
 246                                 $rdel = wfMsgHtml('rev-delundel');
 247                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 248                         // If revision was hidden from sysops
 249                                 $rdel = wfMsgHtml('rev-delundel');
 250                         } else {
 251                                 $rdel = $sk->makeKnownLinkObj( $revdel,
 252                                         wfMsgHtml('rev-delundel'),
 253                                         'target=' . urlencode( $this->mNewRev->mTitle->getPrefixedDbkey() ) .
 254                                         '&oldid=' . urlencode( $this->mNewRev->getId() ) );
 255                                 // Bolden oversighted content
 256                                 if( $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 257                                         $rdel = "<strong>$rdel</strong>";
 258                         }
 259                         $rdel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$rdel</small>)</tt> ";
 260                 }
 261
 262                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 263                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, true ) . "</div>" .
 264                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, true ) . $ldel . "</div>" .
 265                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 266                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 267                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, true ) . " $rollback</div>" .
 268                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, true ) . $rdel . "</div>" .
 269                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 270
 271                 $this->showDiff( $oldHeader, $newHeader );
 272
 273                 if ( !$diffOnly )
 274                         $this->renderNewRevision();
 275
 276                 wfProfileOut( __METHOD__ );
 277         }
 278
 279         /**
 280          * Show the new revision of the page.
 281          */
 282         function renderNewRevision() {
 283                 global $wgOut;
 284                 wfProfileIn( __METHOD__ );
 285
 286                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 287                 #add deleted rev tag if needed
 288                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 289                         $wgOut->addWikiMsg( 'rev-deleted-text-permission' );
 290                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 291                         $wgOut->addWikiMsg( 'rev-deleted-text-view' );
 292                 }
 293
 294                 if( !$this->mNewRev->isCurrent() ) {
 295                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 296                 }
 297
 298                 $this->loadNewText();
 299                 if( is_object( $this->mNewRev ) ) {
 300                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 301                 }
 302
 303                 if ($this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage()) {
 304                         // Stolen from Article::view --AG 2007-10-11
 305
 306                         // Give hooks a chance to customise the output
 307                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 308                                 // Wrap the whole lot in a <pre> and don't parse
 309                                 $m = array();
 310                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 311                                 $wgOut->addHtml( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 312                                 $wgOut->addHtml( htmlspecialchars( $this->mNewtext ) );
 313                                 $wgOut->addHtml( "\n</pre>\n" );
 314                         }
 315                 } else
 316                         $wgOut->addWikiTextTidy( $this->mNewtext );
 317
 318                 if( !$this->mNewRev->isCurrent() ) {
 319                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 320                 }
 321
 322                 wfProfileOut( __METHOD__ );
 323         }
 324
 325         /**
 326          * Show the first revision of an article. Uses normal diff headers in
 327          * contrast to normal "old revision" display style.
 328          */
 329         function showFirstRevision() {
 330                 global $wgOut, $wgUser;
 331                 wfProfileIn( __METHOD__ );
 332
 333                 # Get article text from the DB
 334                 #
 335                 if ( ! $this->loadNewText() ) {
 336                         $t = $this->mTitle->getPrefixedText() . ' ' . wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 337                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 338                         $wgOut->addWikiMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 339                         wfProfileOut( __METHOD__ );
 340                         return;
 341                 }
 342                 if ( $this->mNewRev->isCurrent() ) {
 343                         $wgOut->setArticleFlag( true );
 344                 }
 345
 346                 # Check if user is allowed to look at this page. If not, bail out.
 347                 #
 348                 if ( !( $this->mTitle->userCanRead() ) ) {
 349                         $wgOut->loginToUse();
 350                         $wgOut->output();
 351                         wfProfileOut( __METHOD__ );
 352                         exit;
 353                 }
 354
 355                 # Prepare the header box
 356                 #
 357                 $sk = $wgUser->getSkin();
 358
 359                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 360                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 361                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 362                         $sk->revComment( $this->mNewRev ) . "<br />" .
 363                         $nextlink . "</div>\n";
 364
 365                 $wgOut->addHTML( $header );
 366
 367                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 368                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 369
 370                 wfProfileOut( __METHOD__ );
 371         }
 372
 373         /**
 374          * Get the diff text, send it to $wgOut
 375          * Returns false if the diff could not be generated, otherwise returns true
 376          */
 377         function showDiff( $otitle, $ntitle ) {
 378                 global $wgOut;
 379                 $diff = $this->getDiff( $otitle, $ntitle );
 380                 if ( $diff === false ) {
 381                         $wgOut->addWikiMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" );
 382                         return false;
 383                 } else {
 384                         $this->showDiffStyle();
 385                         $wgOut->addHTML( $diff );
 386                         return true;
 387                 }
 388         }
 389
 390         /**
 391          * Add style sheets and supporting JS for diff display.
 392          */
 393         function showDiffStyle() {
 394                 global $wgStylePath, $wgStyleVersion, $wgOut;
 395                 $wgOut->addStyle( 'common/diff.css' );
 396
 397                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 398                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 399         }
 400
 401         /**
 402          * Get complete diff table, including header
 403          *
 404          * @param Title $otitle Old title
 405          * @param Title $ntitle New title
 406          * @return mixed
 407          */
 408         function getDiff( $otitle, $ntitle ) {
 409                 $body = $this->getDiffBody();
 410                 if ( $body === false ) {
 411                         return false;
 412                 } else {
 413                         $multi = $this->getMultiNotice();
 414                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 415                 }
 416         }
 417
 418         /**
 419          * Get the diff table body, without header
 420          *
 421          * @return mixed
 422          */
 423         function getDiffBody() {
 424                 global $wgMemc;
 425                 wfProfileIn( __METHOD__ );
 426                 // Check if the diff should be hidden from this user
 427                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 428                         return '';
 429                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 430                         return '';
 431                 }
 432                 // Cacheable?
 433                 $key = false;
 434                 if ( $this->mOldid && $this->mNewid ) {
 435                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 436                         // Try cache
 437                         if ( !$this->mRefreshCache ) {
 438                                 $difftext = $wgMemc->get( $key );
 439                                 if ( $difftext ) {
 440                                         wfIncrStats( 'diff_cache_hit' );
 441                                         $difftext = $this->localiseLineNumbers( $difftext );
 442                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 443                                         wfProfileOut( __METHOD__ );
 444                                         return $difftext;
 445                                 }
 446                         } // don't try to load but save the result
 447                 }
 448
 449                 // Loadtext is permission safe, this just clears out the diff
 450                 if ( !$this->loadText() ) {
 451                         wfProfileOut( __METHOD__ );
 452                         return false;
 453                 }
 454
 455                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 456
 457                 // Save to cache for 7 days
 458                 if ( $key !== false && $difftext !== false ) {
 459                         wfIncrStats( 'diff_cache_miss' );
 460                         $wgMemc->set( $key, $difftext, 7*86400 );
 461                 } else {
 462                         wfIncrStats( 'diff_uncacheable' );
 463                 }
 464                 // Replace line numbers with the text in the user's language
 465                 if ( $difftext !== false ) {
 466                         $difftext = $this->localiseLineNumbers( $difftext );
 467                 }
 468                 wfProfileOut( __METHOD__ );
 469                 return $difftext;
 470         }
 471
 472         /**
 473          * Generate a diff, no caching
 474          * $otext and $ntext must be already segmented
 475          */
 476         function generateDiffBody( $otext, $ntext ) {
 477                 global $wgExternalDiffEngine, $wgContLang;
 478
 479                 $otext = str_replace( "\r\n", "\n", $otext );
 480                 $ntext = str_replace( "\r\n", "\n", $ntext );
 481
 482                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 483                         # For historical reasons, external diff engine expects
 484                         # input text to be HTML-escaped already
 485                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 486                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 487                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 488                                 dl('php_wikidiff.so');
 489                         }
 490                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 491                                 $this->debug( 'wikidiff1' );
 492                 }
 493
 494                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 495                         # Better external diff engine, the 2 may some day be dropped
 496                         # This one does the escaping and segmenting itself
 497                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 498                                 wfProfileIn( __METHOD__ . "-dl" );
 499                                 @dl('php_wikidiff2.so');
 500                                 wfProfileOut( __METHOD__ . "-dl" );
 501                         }
 502                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 503                                 wfProfileIn( 'wikidiff2_do_diff' );
 504                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 505                                 $text .= $this->debug( 'wikidiff2' );
 506                                 wfProfileOut( 'wikidiff2_do_diff' );
 507                                 return $text;
 508                         }
 509                 }
 510                 if ( $wgExternalDiffEngine !== false ) {
 511                         # Diff via the shell
 512                         global $wgTmpDirectory;
 513                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 514                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 515
 516                         $tempFile1 = fopen( $tempName1, "w" );
 517                         if ( !$tempFile1 ) {
 518                                 wfProfileOut( __METHOD__ );
 519                                 return false;
 520                         }
 521                         $tempFile2 = fopen( $tempName2, "w" );
 522                         if ( !$tempFile2 ) {
 523                                 wfProfileOut( __METHOD__ );
 524                                 return false;
 525                         }
 526                         fwrite( $tempFile1, $otext );
 527                         fwrite( $tempFile2, $ntext );
 528                         fclose( $tempFile1 );
 529                         fclose( $tempFile2 );
 530                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 531                         wfProfileIn( __METHOD__ . "-shellexec" );
 532                         $difftext = wfShellExec( $cmd );
 533                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 534                         wfProfileOut( __METHOD__ . "-shellexec" );
 535                         unlink( $tempName1 );
 536                         unlink( $tempName2 );
 537                         return $difftext;
 538                 }
 539
 540                 # Native PHP diff
 541                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 542                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 543                 $diffs = new Diff( $ota, $nta );
 544                 $formatter = new TableDiffFormatter();
 545                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 546                         $this->debug();
 547         }
 548
 549         /**
 550          * Generate a debug comment indicating diff generating time,
 551          * server node, and generator backend.
 552          */
 553         protected function debug( $generator="internal" ) {
 554                 global $wgShowHostnames, $wgNodeName;
 555                 $data = array( $generator );
 556                 if( $wgShowHostnames ) {
 557                         $data[] = $wgNodeName;
 558                 }
 559                 $data[] = wfTimestamp( TS_DB );
 560                 return "<!-- diff generator: " .
 561                         implode( " ",
 562                                 array_map(
 563                                         "htmlspecialchars",
 564                                                 $data ) ) .
 565                         " -->\n";
 566         }
 567
 568         /**
 569          * Replace line numbers with the text in the user's language
 570          */
 571         function localiseLineNumbers( $text ) {
 572                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 573                         array( &$this, 'localiseLineNumbersCb' ), $text );
 574         }
 575
 576         function localiseLineNumbersCb( $matches ) {
 577                 global $wgLang;
 578                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 579         }
 580
 581
 582         /**
 583          * If there are revisions between the ones being compared, return a note saying so.
 584          */
 585         function getMultiNotice() {
 586                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 587                         return '';
 588
 589                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 590                         // Comparing two different pages? Count would be meaningless.
 591                         return '';
 592                 }
 593
 594                 $oldid = $this->mOldRev->getId();
 595                 $newid = $this->mNewRev->getId();
 596                 if ( $oldid > $newid ) {
 597                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 598                 }
 599
 600                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 601                 if ( !$n )
 602                         return '';
 603
 604                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 605         }
 606
 607
 608         /**
 609          * Add the header to a diff body
 610          */
 611         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 612                 global $wgOut;
 613
 614                 $header = "
 615                         <table class='diff'>
 616                         <col class='diff-marker' />
 617                         <col class='diff-content' />
 618                         <col class='diff-marker' />
 619                         <col class='diff-content' />
 620                         <tr valign='top'>
 621                                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 622                                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 623                         </tr>
 624                 ";
 625
 626                 if ( $multi != '' )
 627                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 628
 629                 return $header . $diff . "</table>";
 630         }
 631
 632         /**
 633          * Use specified text instead of loading from the database
 634          */
 635         function setText( $oldText, $newText ) {
 636                 $this->mOldtext = $oldText;
 637                 $this->mNewtext = $newText;
 638                 $this->mTextLoaded = 2;
 639         }
 640
 641         /**
 642          * Load revision metadata for the specified articles. If newid is 0, then compare
 643          * the old article in oldid to the current article; if oldid is 0, then
 644          * compare the current article to the immediately previous one (ignoring the
 645          * value of newid).
 646          *
 647          * If oldid is false, leave the corresponding revision object set
 648          * to false. This is impossible via ordinary user input, and is provided for
 649          * API convenience.
 650          */
 651         function loadRevisionData() {
 652                 global $wgLang;
 653                 if ( $this->mRevisionsLoaded ) {
 654                         return true;
 655                 } else {
 656                         // Whether it succeeds or fails, we don't want to try again
 657                         $this->mRevisionsLoaded = true;
 658                 }
 659
 660                 // Load the new revision object
 661                 $this->mNewRev = $this->mNewid
 662                         ? Revision::newFromId( $this->mNewid )
 663                         : Revision::newFromTitle( $this->mTitle );
 664                 if( !$this->mNewRev instanceof Revision )
 665                         return false;
 666
 667                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 668                 $this->mNewid = $this->mNewRev->getId();
 669
 670                 // Check if page is editable
 671                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 672
 673                 // Set assorted variables
 674                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 675                 $this->mNewPage = $this->mNewRev->getTitle();
 676                 if( $this->mNewRev->isCurrent() ) {
 677                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 678                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 679                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 680
 681                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)";
 682                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 683
 684                 } else {
 685                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 686                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 687                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 688
 689                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 690                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 691                 }
 692                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 693                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 694                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 695                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 696                 }
 697
 698                 // Load the old revision object
 699                 $this->mOldRev = false;
 700                 if( $this->mOldid ) {
 701                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 702                 } elseif ( $this->mOldid === 0 ) {
 703                         $rev = $this->mNewRev->getPrevious();
 704                         if( $rev ) {
 705                                 $this->mOldid = $rev->getId();
 706                                 $this->mOldRev = $rev;
 707                         } else {
 708                                 // No previous revision; mark to show as first-version only.
 709                                 $this->mOldid = false;
 710                                 $this->mOldRev = false;
 711                         }
 712                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 713
 714                 if( is_null( $this->mOldRev ) ) {
 715                         return false;
 716                 }
 717
 718                 if ( $this->mOldRev ) {
 719                         $this->mOldPage = $this->mOldRev->getTitle();
 720
 721                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 722                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 723                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 724                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 725
 726                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 727                                 . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 728                         // Add an "undo" link
 729                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 730                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 731                                 $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 732                         }
 733
 734                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 735                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 736                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 737                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 738                         }
 739                 }
 740
 741                 return true;
 742         }
 743
 744         /**
 745          * Load the text of the revisions, as well as revision data.
 746          */
 747         function loadText() {
 748                 if ( $this->mTextLoaded == 2 ) {
 749                         return true;
 750                 } else {
 751                         // Whether it succeeds or fails, we don't want to try again
 752                         $this->mTextLoaded = 2;
 753                 }
 754
 755                 if ( !$this->loadRevisionData() ) {
 756                         return false;
 757                 }
 758                 if ( $this->mOldRev ) {
 759                         $this->mOldtext = $this->mOldRev->revText();
 760                         if ( $this->mOldtext === false ) {
 761                                 return false;
 762                         }
 763                 }
 764                 if ( $this->mNewRev ) {
 765                         $this->mNewtext = $this->mNewRev->revText();
 766                         if ( $this->mNewtext === false ) {
 767                                 return false;
 768                         }
 769                 }
 770                 return true;
 771         }
 772
 773         /**
 774          * Load the text of the new revision, not the old one
 775          */
 776         function loadNewText() {
 777                 if ( $this->mTextLoaded >= 1 ) {
 778                         return true;
 779                 } else {
 780                         $this->mTextLoaded = 1;
 781                 }
 782                 if ( !$this->loadRevisionData() ) {
 783                         return false;
 784                 }
 785                 $this->mNewtext = $this->mNewRev->getText();
 786                 return true;
 787         }
 788
 789
 790 }
 791
 792 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 793 //
 794 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 795 // You may copy this code freely under the conditions of the GPL.
 796 //
 797
 798 define('USE_ASSERTS', function_exists('assert'));
 799
 800 /**
 801  * @todo document
 802  * @private
 803  * @ingroup DifferenceEngine
 804  */
 805 class _DiffOp {
 806         var $type;
 807         var $orig;
 808         var $closing;
 809
 810         function reverse() {
 811                 trigger_error('pure virtual', E_USER_ERROR);
 812         }
 813
 814         function norig() {
 815                 return $this->orig ? sizeof($this->orig) : 0;
 816         }
 817
 818         function nclosing() {
 819                 return $this->closing ? sizeof($this->closing) : 0;
 820         }
 821 }
 822
 823 /**
 824  * @todo document
 825  * @private
 826  * @ingroup DifferenceEngine
 827  */
 828 class _DiffOp_Copy extends _DiffOp {
 829         var $type = 'copy';
 830
 831         function _DiffOp_Copy ($orig, $closing = false) {
 832                 if (!is_array($closing))
 833                         $closing = $orig;
 834                 $this->orig = $orig;
 835                 $this->closing = $closing;
 836         }
 837
 838         function reverse() {
 839                 return new _DiffOp_Copy($this->closing, $this->orig);
 840         }
 841 }
 842
 843 /**
 844  * @todo document
 845  * @private
 846  * @ingroup DifferenceEngine
 847  */
 848 class _DiffOp_Delete extends _DiffOp {
 849         var $type = 'delete';
 850
 851         function _DiffOp_Delete ($lines) {
 852                 $this->orig = $lines;
 853                 $this->closing = false;
 854         }
 855
 856         function reverse() {
 857                 return new _DiffOp_Add($this->orig);
 858         }
 859 }
 860
 861 /**
 862  * @todo document
 863  * @private
 864  * @ingroup DifferenceEngine
 865  */
 866 class _DiffOp_Add extends _DiffOp {
 867         var $type = 'add';
 868
 869         function _DiffOp_Add ($lines) {
 870                 $this->closing = $lines;
 871                 $this->orig = false;
 872         }
 873
 874         function reverse() {
 875                 return new _DiffOp_Delete($this->closing);
 876         }
 877 }
 878
 879 /**
 880  * @todo document
 881  * @private
 882  * @ingroup DifferenceEngine
 883  */
 884 class _DiffOp_Change extends _DiffOp {
 885         var $type = 'change';
 886
 887         function _DiffOp_Change ($orig, $closing) {
 888                 $this->orig = $orig;
 889                 $this->closing = $closing;
 890         }
 891
 892         function reverse() {
 893                 return new _DiffOp_Change($this->closing, $this->orig);
 894         }
 895 }
 896
 897
 898 /**
 899  * Class used internally by Diff to actually compute the diffs.
 900  *
 901  * The algorithm used here is mostly lifted from the perl module
 902  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 903  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 904  *
 905  * More ideas are taken from:
 906  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 907  *
 908  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 909  * diffutils-2.7, which can be found at:
 910  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 911  *
 912  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 913  * are my own.
 914  *
 915  * Line length limits for robustness added by Tim Starling, 2005-08-31
 916  *
 917  * @author Geoffrey T. Dairiki, Tim Starling
 918  * @private
 919  * @ingroup DifferenceEngine
 920  */
 921 class _DiffEngine {
 922         const MAX_XREF_LENGTH =  10000;
 923
 924         function diff ($from_lines, $to_lines) {
 925                 wfProfileIn( __METHOD__ );
 926
 927                 $n_from = sizeof($from_lines);
 928                 $n_to = sizeof($to_lines);
 929
 930                 $this->xchanged = $this->ychanged = array();
 931                 $this->xv = $this->yv = array();
 932                 $this->xind = $this->yind = array();
 933                 unset($this->seq);
 934                 unset($this->in_seq);
 935                 unset($this->lcs);
 936
 937                 // Skip leading common lines.
 938                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 939                         if ($from_lines[$skip] !== $to_lines[$skip])
 940                                 break;
 941                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 942                 }
 943                 // Skip trailing common lines.
 944                 $xi = $n_from; $yi = $n_to;
 945                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 946                         if ($from_lines[$xi] !== $to_lines[$yi])
 947                                 break;
 948                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 949                 }
 950
 951                 // Ignore lines which do not exist in both files.
 952                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 953                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 954                 }
 955
 956                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 957                         $line = $to_lines[$yi];
 958                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 959                                 continue;
 960                         $yhash[$this->_line_hash($line)] = 1;
 961                         $this->yv[] = $line;
 962                         $this->yind[] = $yi;
 963                 }
 964                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 965                         $line = $from_lines[$xi];
 966                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 967                                 continue;
 968                         $this->xv[] = $line;
 969                         $this->xind[] = $xi;
 970                 }
 971
 972                 // Find the LCS.
 973                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 974
 975                 // Merge edits when possible
 976                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 977                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 978
 979                 // Compute the edit operations.
 980                 $edits = array();
 981                 $xi = $yi = 0;
 982                 while ($xi < $n_from || $yi < $n_to) {
 983                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 984                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 985
 986                         // Skip matching "snake".
 987                         $copy = array();
 988                         while ( $xi < $n_from && $yi < $n_to
 989                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 990                                 $copy[] = $from_lines[$xi++];
 991                                 ++$yi;
 992                         }
 993                         if ($copy)
 994                                 $edits[] = new _DiffOp_Copy($copy);
 995
 996                         // Find deletes & adds.
 997                         $delete = array();
 998                         while ($xi < $n_from && $this->xchanged[$xi])
 999                                 $delete[] = $from_lines[$xi++];
1000
1001                         $add = array();
1002                         while ($yi < $n_to && $this->ychanged[$yi])
1003                                 $add[] = $to_lines[$yi++];
1004
1005                         if ($delete && $add)
1006                                 $edits[] = new _DiffOp_Change($delete, $add);
1007                         elseif ($delete)
1008                                 $edits[] = new _DiffOp_Delete($delete);
1009                         elseif ($add)
1010                                 $edits[] = new _DiffOp_Add($add);
1011                 }
1012                 wfProfileOut( __METHOD__ );
1013                 return $edits;
1014         }
1015
1016         /**
1017          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1018          */
1019         function _line_hash( $line ) {
1020                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1021                         return md5( $line );
1022                 } else {
1023                         return $line;
1024                 }
1025         }
1026
1027
1028         /* Divide the Largest Common Subsequence (LCS) of the sequences
1029          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1030          * sized segments.
1031          *
1032          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1033          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1034          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1035          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1036          * that (X0, Y0) == (XOFF, YOFF) and
1037          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1038          *
1039          * This function assumes that the first lines of the specified portions
1040          * of the two files do not match, and likewise that the last lines do not
1041          * match.  The caller must trim matching lines from the beginning and end
1042          * of the portions it is going to specify.
1043          */
1044         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1045                 wfProfileIn( __METHOD__ );
1046                 $flip = false;
1047
1048                 if ($xlim - $xoff > $ylim - $yoff) {
1049                         // Things seems faster (I'm not sure I understand why)
1050                                 // when the shortest sequence in X.
1051                                 $flip = true;
1052                         list ($xoff, $xlim, $yoff, $ylim)
1053                         = array( $yoff, $ylim, $xoff, $xlim);
1054                 }
1055
1056                 if ($flip)
1057                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1058                                 $ymatches[$this->xv[$i]][] = $i;
1059                 else
1060                         for ($i = $ylim - 1; $i >= $yoff; $i--)
1061                                 $ymatches[$this->yv[$i]][] = $i;
1062
1063                 $this->lcs = 0;
1064                 $this->seq[0]= $yoff - 1;
1065                 $this->in_seq = array();
1066                 $ymids[0] = array();
1067
1068                 $numer = $xlim - $xoff + $nchunks - 1;
1069                 $x = $xoff;
1070                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1071                         wfProfileIn( __METHOD__ . "-chunk" );
1072                         if ($chunk > 0)
1073                                 for ($i = 0; $i <= $this->lcs; $i++)
1074                                         $ymids[$i][$chunk-1] = $this->seq[$i];
1075
1076                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1077                         for ( ; $x < $x1; $x++) {
1078                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1079                                         if (empty($ymatches[$line]))
1080                                                 continue;
1081                                 $matches = $ymatches[$line];
1082                                 reset($matches);
1083                                 while (list ($junk, $y) = each($matches))
1084                                         if (empty($this->in_seq[$y])) {
1085                                                 $k = $this->_lcs_pos($y);
1086                                                 USE_ASSERTS && assert($k > 0);
1087                                                 $ymids[$k] = $ymids[$k-1];
1088                                                 break;
1089                                         }
1090                                 while (list ( /* $junk */, $y) = each($matches)) {
1091                                         if ($y > $this->seq[$k-1]) {
1092                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1093                                                 // Optimization: this is a common case:
1094                                                 //      next match is just replacing previous match.
1095                                                 $this->in_seq[$this->seq[$k]] = false;
1096                                                 $this->seq[$k] = $y;
1097                                                 $this->in_seq[$y] = 1;
1098                                         } else if (empty($this->in_seq[$y])) {
1099                                                 $k = $this->_lcs_pos($y);
1100                                                 USE_ASSERTS && assert($k > 0);
1101                                                 $ymids[$k] = $ymids[$k-1];
1102                                         }
1103                                 }
1104                         }
1105                         wfProfileOut( __METHOD__ . "-chunk" );
1106                 }
1107
1108                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1109                 $ymid = $ymids[$this->lcs];
1110                 for ($n = 0; $n < $nchunks - 1; $n++) {
1111                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1112                         $y1 = $ymid[$n] + 1;
1113                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1114                 }
1115                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1116
1117                 wfProfileOut( __METHOD__ );
1118                 return array($this->lcs, $seps);
1119         }
1120
1121         function _lcs_pos ($ypos) {
1122                 wfProfileIn( __METHOD__ );
1123
1124                 $end = $this->lcs;
1125                 if ($end == 0 || $ypos > $this->seq[$end]) {
1126                         $this->seq[++$this->lcs] = $ypos;
1127                         $this->in_seq[$ypos] = 1;
1128                         wfProfileOut( __METHOD__ );
1129                         return $this->lcs;
1130                 }
1131
1132                 $beg = 1;
1133                 while ($beg < $end) {
1134                         $mid = (int)(($beg + $end) / 2);
1135                         if ( $ypos > $this->seq[$mid] )
1136                                 $beg = $mid + 1;
1137                         else
1138                                 $end = $mid;
1139                 }
1140
1141                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1142
1143                 $this->in_seq[$this->seq[$end]] = false;
1144                 $this->seq[$end] = $ypos;
1145                 $this->in_seq[$ypos] = 1;
1146                 wfProfileOut( __METHOD__ );
1147                 return $end;
1148         }
1149
1150         /* Find LCS of two sequences.
1151          *
1152          * The results are recorded in the vectors $this->{x,y}changed[], by
1153          * storing a 1 in the element for each line that is an insertion
1154          * or deletion (ie. is not in the LCS).
1155          *
1156          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1157          *
1158          * Note that XLIM, YLIM are exclusive bounds.
1159          * All line numbers are origin-0 and discarded lines are not counted.
1160          */
1161         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1162                 wfProfileIn( __METHOD__ );
1163
1164                 // Slide down the bottom initial diagonal.
1165                 while ($xoff < $xlim && $yoff < $ylim
1166                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1167                         ++$xoff;
1168                         ++$yoff;
1169                 }
1170
1171                 // Slide up the top initial diagonal.
1172                 while ($xlim > $xoff && $ylim > $yoff
1173                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1174                         --$xlim;
1175                         --$ylim;
1176                 }
1177
1178                 if ($xoff == $xlim || $yoff == $ylim)
1179                         $lcs = 0;
1180                 else {
1181                         // This is ad hoc but seems to work well.
1182                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1183                         //$nchunks = max(2,min(8,(int)$nchunks));
1184                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1185                         list ($lcs, $seps)
1186                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1187                 }
1188
1189                 if ($lcs == 0) {
1190                         // X and Y sequences have no common subsequence:
1191                         // mark all changed.
1192                         while ($yoff < $ylim)
1193                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1194                         while ($xoff < $xlim)
1195                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1196                 } else {
1197                         // Use the partitions to split this problem into subproblems.
1198                         reset($seps);
1199                         $pt1 = $seps[0];
1200                         while ($pt2 = next($seps)) {
1201                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1202                                 $pt1 = $pt2;
1203                         }
1204                 }
1205                 wfProfileOut( __METHOD__ );
1206         }
1207
1208         /* Adjust inserts/deletes of identical lines to join changes
1209          * as much as possible.
1210          *
1211          * We do something when a run of changed lines include a
1212          * line at one end and has an excluded, identical line at the other.
1213          * We are free to choose which identical line is included.
1214          * `compareseq' usually chooses the one at the beginning,
1215          * but usually it is cleaner to consider the following identical line
1216          * to be the "change".
1217          *
1218          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1219          */
1220         function _shift_boundaries ($lines, &$changed, $other_changed) {
1221                 wfProfileIn( __METHOD__ );
1222                 $i = 0;
1223                 $j = 0;
1224
1225                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1226                 $len = sizeof($lines);
1227                 $other_len = sizeof($other_changed);
1228
1229                 while (1) {
1230                         /*
1231                          * Scan forwards to find beginning of another run of changes.
1232                          * Also keep track of the corresponding point in the other file.
1233                          *
1234                          * Throughout this code, $i and $j are adjusted together so that
1235                          * the first $i elements of $changed and the first $j elements
1236                          * of $other_changed both contain the same number of zeros
1237                          * (unchanged lines).
1238                          * Furthermore, $j is always kept so that $j == $other_len or
1239                          * $other_changed[$j] == false.
1240                          */
1241                         while ($j < $other_len && $other_changed[$j])
1242                                 $j++;
1243
1244                         while ($i < $len && ! $changed[$i]) {
1245                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1246                                 $i++; $j++;
1247                                 while ($j < $other_len && $other_changed[$j])
1248                                         $j++;
1249                         }
1250
1251                         if ($i == $len)
1252                                 break;
1253
1254                         $start = $i;
1255
1256                         // Find the end of this run of changes.
1257                         while (++$i < $len && $changed[$i])
1258                                 continue;
1259
1260                         do {
1261                                 /*
1262                                  * Record the length of this run of changes, so that
1263                                  * we can later determine whether the run has grown.
1264                                  */
1265                                 $runlength = $i - $start;
1266
1267                                 /*
1268                                  * Move the changed region back, so long as the
1269                                  * previous unchanged line matches the last changed one.
1270                                  * This merges with previous changed regions.
1271                                  */
1272                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1273                                         $changed[--$start] = 1;
1274                                         $changed[--$i] = false;
1275                                         while ($start > 0 && $changed[$start - 1])
1276                                                 $start--;
1277                                         USE_ASSERTS && assert('$j > 0');
1278                                         while ($other_changed[--$j])
1279                                                 continue;
1280                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1281                                 }
1282
1283                                 /*
1284                                  * Set CORRESPONDING to the end of the changed run, at the last
1285                                  * point where it corresponds to a changed run in the other file.
1286                                  * CORRESPONDING == LEN means no such point has been found.
1287                                  */
1288                                 $corresponding = $j < $other_len ? $i : $len;
1289
1290                                 /*
1291                                  * Move the changed region forward, so long as the
1292                                  * first changed line matches the following unchanged one.
1293                                  * This merges with following changed regions.
1294                                  * Do this second, so that if there are no merges,
1295                                  * the changed region is moved forward as far as possible.
1296                                  */
1297                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1298                                         $changed[$start++] = false;
1299                                         $changed[$i++] = 1;
1300                                         while ($i < $len && $changed[$i])
1301                                                 $i++;
1302
1303                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1304                                         $j++;
1305                                         if ($j < $other_len && $other_changed[$j]) {
1306                                                 $corresponding = $i;
1307                                                 while ($j < $other_len && $other_changed[$j])
1308                                                         $j++;
1309                                         }
1310                                 }
1311                         } while ($runlength != $i - $start);
1312
1313                         /*
1314                          * If possible, move the fully-merged run of changes
1315                          * back to a corresponding run in the other file.
1316                          */
1317                         while ($corresponding < $i) {
1318                                 $changed[--$start] = 1;
1319                                 $changed[--$i] = 0;
1320                                 USE_ASSERTS && assert('$j > 0');
1321                                 while ($other_changed[--$j])
1322                                         continue;
1323                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1324                         }
1325                 }
1326                 wfProfileOut( __METHOD__ );
1327         }
1328 }
1329
1330 /**
1331  * Class representing a 'diff' between two sequences of strings.
1332  * @todo document
1333  * @private
1334  * @ingroup DifferenceEngine
1335  */
1336 class Diff
1337 {
1338         var $edits;
1339
1340         /**
1341          * Constructor.
1342          * Computes diff between sequences of strings.
1343          *
1344          * @param $from_lines array An array of strings.
1345          *                (Typically these are lines from a file.)
1346          * @param $to_lines array An array of strings.
1347          */
1348         function Diff($from_lines, $to_lines) {
1349                 $eng = new _DiffEngine;
1350                 $this->edits = $eng->diff($from_lines, $to_lines);
1351                 //$this->_check($from_lines, $to_lines);
1352         }
1353
1354         /**
1355          * Compute reversed Diff.
1356          *
1357          * SYNOPSIS:
1358          *
1359          *      $diff = new Diff($lines1, $lines2);
1360          *      $rev = $diff->reverse();
1361          * @return object A Diff object representing the inverse of the
1362          *                                original diff.
1363          */
1364         function reverse () {
1365                 $rev = $this;
1366                 $rev->edits = array();
1367                 foreach ($this->edits as $edit) {
1368                         $rev->edits[] = $edit->reverse();
1369                 }
1370                 return $rev;
1371         }
1372
1373         /**
1374          * Check for empty diff.
1375          *
1376          * @return bool True iff two sequences were identical.
1377          */
1378         function isEmpty () {
1379                 foreach ($this->edits as $edit) {
1380                         if ($edit->type != 'copy')
1381                                 return false;
1382                 }
1383                 return true;
1384         }
1385
1386         /**
1387          * Compute the length of the Longest Common Subsequence (LCS).
1388          *
1389          * This is mostly for diagnostic purposed.
1390          *
1391          * @return int The length of the LCS.
1392          */
1393         function lcs () {
1394                 $lcs = 0;
1395                 foreach ($this->edits as $edit) {
1396                         if ($edit->type == 'copy')
1397                                 $lcs += sizeof($edit->orig);
1398                 }
1399                 return $lcs;
1400         }
1401
1402         /**
1403          * Get the original set of lines.
1404          *
1405          * This reconstructs the $from_lines parameter passed to the
1406          * constructor.
1407          *
1408          * @return array The original sequence of strings.
1409          */
1410         function orig() {
1411                 $lines = array();
1412
1413                 foreach ($this->edits as $edit) {
1414                         if ($edit->orig)
1415                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1416                 }
1417                 return $lines;
1418         }
1419
1420         /**
1421          * Get the closing set of lines.
1422          *
1423          * This reconstructs the $to_lines parameter passed to the
1424          * constructor.
1425          *
1426          * @return array The sequence of strings.
1427          */
1428         function closing() {
1429                 $lines = array();
1430
1431                 foreach ($this->edits as $edit) {
1432                         if ($edit->closing)
1433                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1434                 }
1435                 return $lines;
1436         }
1437
1438         /**
1439          * Check a Diff for validity.
1440          *
1441          * This is here only for debugging purposes.
1442          */
1443         function _check ($from_lines, $to_lines) {
1444                 wfProfileIn( __METHOD__ );
1445                 if (serialize($from_lines) != serialize($this->orig()))
1446                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1447                 if (serialize($to_lines) != serialize($this->closing()))
1448                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1449
1450                 $rev = $this->reverse();
1451                 if (serialize($to_lines) != serialize($rev->orig()))
1452                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1453                 if (serialize($from_lines) != serialize($rev->closing()))
1454                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1455
1456
1457                 $prevtype = 'none';
1458                 foreach ($this->edits as $edit) {
1459                         if ( $prevtype == $edit->type )
1460                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1461                         $prevtype = $edit->type;
1462                 }
1463
1464                 $lcs = $this->lcs();
1465                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1466                 wfProfileOut( __METHOD__ );
1467         }
1468 }
1469
1470 /**
1471  * @todo document, bad name.
1472  * @private
1473  * @ingroup DifferenceEngine
1474  */
1475 class MappedDiff extends Diff
1476 {
1477         /**
1478          * Constructor.
1479          *
1480          * Computes diff between sequences of strings.
1481          *
1482          * This can be used to compute things like
1483          * case-insensitve diffs, or diffs which ignore
1484          * changes in white-space.
1485          *
1486          * @param $from_lines array An array of strings.
1487          *      (Typically these are lines from a file.)
1488          *
1489          * @param $to_lines array An array of strings.
1490          *
1491          * @param $mapped_from_lines array This array should
1492          *      have the same size number of elements as $from_lines.
1493          *      The elements in $mapped_from_lines and
1494          *      $mapped_to_lines are what is actually compared
1495          *      when computing the diff.
1496          *
1497          * @param $mapped_to_lines array This array should
1498          *      have the same number of elements as $to_lines.
1499          */
1500         function MappedDiff($from_lines, $to_lines,
1501                 $mapped_from_lines, $mapped_to_lines) {
1502                 wfProfileIn( __METHOD__ );
1503
1504                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1505                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1506
1507                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1508
1509                 $xi = $yi = 0;
1510                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1511                         $orig = &$this->edits[$i]->orig;
1512                         if (is_array($orig)) {
1513                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1514                                 $xi += sizeof($orig);
1515                         }
1516
1517                         $closing = &$this->edits[$i]->closing;
1518                         if (is_array($closing)) {
1519                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1520                                 $yi += sizeof($closing);
1521                         }
1522                 }
1523                 wfProfileOut( __METHOD__ );
1524         }
1525 }
1526
1527 /**
1528  * A class to format Diffs
1529  *
1530  * This class formats the diff in classic diff format.
1531  * It is intended that this class be customized via inheritance,
1532  * to obtain fancier outputs.
1533  * @todo document
1534  * @private
1535  * @ingroup DifferenceEngine
1536  */
1537 class DiffFormatter {
1538         /**
1539          * Number of leading context "lines" to preserve.
1540          *
1541          * This should be left at zero for this class, but subclasses
1542          * may want to set this to other values.
1543          */
1544         var $leading_context_lines = 0;
1545
1546         /**
1547          * Number of trailing context "lines" to preserve.
1548          *
1549          * This should be left at zero for this class, but subclasses
1550          * may want to set this to other values.
1551          */
1552         var $trailing_context_lines = 0;
1553
1554         /**
1555          * Format a diff.
1556          *
1557          * @param $diff object A Diff object.
1558          * @return string The formatted output.
1559          */
1560         function format($diff) {
1561                 wfProfileIn( __METHOD__ );
1562
1563                 $xi = $yi = 1;
1564                 $block = false;
1565                 $context = array();
1566
1567                 $nlead = $this->leading_context_lines;
1568                 $ntrail = $this->trailing_context_lines;
1569
1570                 $this->_start_diff();
1571
1572                 foreach ($diff->edits as $edit) {
1573                         if ($edit->type == 'copy') {
1574                                 if (is_array($block)) {
1575                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1576                                                 $block[] = $edit;
1577                                         }
1578                                         else{
1579                                                 if ($ntrail) {
1580                                                         $context = array_slice($edit->orig, 0, $ntrail);
1581                                                         $block[] = new _DiffOp_Copy($context);
1582                                                 }
1583                                                 $this->_block($x0, $ntrail + $xi - $x0,
1584                                                                           $y0, $ntrail + $yi - $y0,
1585                                                                           $block);
1586                                                 $block = false;
1587                                         }
1588                                 }
1589                                 $context = $edit->orig;
1590                         }
1591                         else {
1592                                 if (! is_array($block)) {
1593                                         $context = array_slice($context, sizeof($context) - $nlead);
1594                                         $x0 = $xi - sizeof($context);
1595                                         $y0 = $yi - sizeof($context);
1596                                         $block = array();
1597                                         if ($context)
1598                                                 $block[] = new _DiffOp_Copy($context);
1599                                 }
1600                                 $block[] = $edit;
1601                         }
1602
1603                         if ($edit->orig)
1604                                 $xi += sizeof($edit->orig);
1605                         if ($edit->closing)
1606                                 $yi += sizeof($edit->closing);
1607                 }
1608
1609                 if (is_array($block))
1610                         $this->_block($x0, $xi - $x0,
1611                                                   $y0, $yi - $y0,
1612                                                   $block);
1613
1614                 $end = $this->_end_diff();
1615                 wfProfileOut( __METHOD__ );
1616                 return $end;
1617         }
1618
1619         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1620                 wfProfileIn( __METHOD__ );
1621                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1622                 foreach ($edits as $edit) {
1623                         if ($edit->type == 'copy')
1624                                 $this->_context($edit->orig);
1625                         elseif ($edit->type == 'add')
1626                                 $this->_added($edit->closing);
1627                         elseif ($edit->type == 'delete')
1628                                 $this->_deleted($edit->orig);
1629                         elseif ($edit->type == 'change')
1630                                 $this->_changed($edit->orig, $edit->closing);
1631                         else
1632                                 trigger_error('Unknown edit type', E_USER_ERROR);
1633                 }
1634                 $this->_end_block();
1635                 wfProfileOut( __METHOD__ );
1636         }
1637
1638         function _start_diff() {
1639                 ob_start();
1640         }
1641
1642         function _end_diff() {
1643                 $val = ob_get_contents();
1644                 ob_end_clean();
1645                 return $val;
1646         }
1647
1648         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1649                 if ($xlen > 1)
1650                         $xbeg .= "," . ($xbeg + $xlen - 1);
1651                 if ($ylen > 1)
1652                         $ybeg .= "," . ($ybeg + $ylen - 1);
1653
1654                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1655         }
1656
1657         function _start_block($header) {
1658                 echo $header . "\n";
1659         }
1660
1661         function _end_block() {
1662         }
1663
1664         function _lines($lines, $prefix = ' ') {
1665                 foreach ($lines as $line)
1666                         echo "$prefix $line\n";
1667         }
1668
1669         function _context($lines) {
1670                 $this->_lines($lines);
1671         }
1672
1673         function _added($lines) {
1674                 $this->_lines($lines, '>');
1675         }
1676         function _deleted($lines) {
1677                 $this->_lines($lines, '<');
1678         }
1679
1680         function _changed($orig, $closing) {
1681                 $this->_deleted($orig);
1682                 echo "---\n";
1683                 $this->_added($closing);
1684         }
1685 }
1686
1687 /**
1688  * A formatter that outputs unified diffs
1689  * @ingroup DifferenceEngine
1690  */
1691
1692 class UnifiedDiffFormatter extends DiffFormatter {
1693         var $leading_context_lines = 2;
1694         var $trailing_context_lines = 2;
1695
1696         function _added($lines) {
1697                 $this->_lines($lines, '+');
1698         }
1699         function _deleted($lines) {
1700                 $this->_lines($lines, '-');
1701         }
1702         function _changed($orig, $closing) {
1703                 $this->_deleted($orig);
1704                 $this->_added($closing);
1705         }
1706         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1707                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1708         }
1709 }
1710
1711 /**
1712  * A pseudo-formatter that just passes along the Diff::$edits array
1713  * @ingroup DifferenceEngine
1714  */
1715 class ArrayDiffFormatter extends DiffFormatter {
1716         function format($diff) {
1717                 $oldline = 1;
1718                 $newline = 1;
1719                 $retval = array();
1720                 foreach($diff->edits as $edit)
1721                         switch($edit->type) {
1722                                 case 'add':
1723                                         foreach($edit->closing as $l) {
1724                                                 $retval[] = array(
1725                                                         'action' => 'add',
1726                                                         'new'=> $l,
1727                                                         'newline' => $newline++
1728                                                 );
1729                                         }
1730                                         break;
1731                                 case 'delete':
1732                                         foreach($edit->orig as $l) {
1733                                                 $retval[] = array(
1734                                                         'action' => 'delete',
1735                                                         'old' => $l,
1736                                                         'oldline' => $oldline++,
1737                                                 );
1738                                         }
1739                                         break;
1740                                 case 'change':
1741                                         foreach($edit->orig as $i => $l) {
1742                                                 $retval[] = array(
1743                                                         'action' => 'change',
1744                                                         'old' => $l,
1745                                                         'new' => @$edit->closing[$i],
1746                                                         'oldline' => $oldline++,
1747                                                         'newline' => $newline++,
1748                                                 );
1749                                         }
1750                                         break;
1751                                 case 'copy':
1752                                         $oldline += count($edit->orig);
1753                                         $newline += count($edit->orig);
1754                         }
1755                 return $retval;
1756         }
1757 }
1758
1759 /**
1760  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1761  *
1762  */
1763
1764 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1765
1766 /**
1767  * @todo document
1768  * @private
1769  * @ingroup DifferenceEngine
1770  */
1771 class _HWLDF_WordAccumulator {
1772         function _HWLDF_WordAccumulator () {
1773                 $this->_lines = array();
1774                 $this->_line = '';
1775                 $this->_group = '';
1776                 $this->_tag = '';
1777         }
1778
1779         function _flushGroup ($new_tag) {
1780                 if ($this->_group !== '') {
1781                         if ($this->_tag == 'ins')
1782                                 $this->_line .= '<ins class="diffchange diffchange-inline">' .
1783                                         htmlspecialchars ( $this->_group ) . '</ins>';
1784                         elseif ($this->_tag == 'del')
1785                                 $this->_line .= '<del class="diffchange diffchange-inline">' .
1786                                         htmlspecialchars ( $this->_group ) . '</del>';
1787                         else
1788                                 $this->_line .= htmlspecialchars ( $this->_group );
1789                 }
1790                 $this->_group = '';
1791                 $this->_tag = $new_tag;
1792         }
1793
1794         function _flushLine ($new_tag) {
1795                 $this->_flushGroup($new_tag);
1796                 if ($this->_line != '')
1797                         array_push ( $this->_lines, $this->_line );
1798                 else
1799                         # make empty lines visible by inserting an NBSP
1800                         array_push ( $this->_lines, NBSP );
1801                 $this->_line = '';
1802         }
1803
1804         function addWords ($words, $tag = '') {
1805                 if ($tag != $this->_tag)
1806                         $this->_flushGroup($tag);
1807
1808                 foreach ($words as $word) {
1809                         // new-line should only come as first char of word.
1810                         if ($word == '')
1811                                 continue;
1812                         if ($word[0] == "\n") {
1813                                 $this->_flushLine($tag);
1814                                 $word = substr($word, 1);
1815                         }
1816                         assert(!strstr($word, "\n"));
1817                         $this->_group .= $word;
1818                 }
1819         }
1820
1821         function getLines() {
1822                 $this->_flushLine('~done');
1823                 return $this->_lines;
1824         }
1825 }
1826
1827 /**
1828  * @todo document
1829  * @private
1830  * @ingroup DifferenceEngine
1831  */
1832 class WordLevelDiff extends MappedDiff {
1833         const MAX_LINE_LENGTH = 10000;
1834
1835         function WordLevelDiff ($orig_lines, $closing_lines) {
1836                 wfProfileIn( __METHOD__ );
1837
1838                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1839                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1840
1841                 $this->MappedDiff($orig_words, $closing_words,
1842                         $orig_stripped, $closing_stripped);
1843                 wfProfileOut( __METHOD__ );
1844         }
1845
1846         function _split($lines) {
1847                 wfProfileIn( __METHOD__ );
1848
1849                 $words = array();
1850                 $stripped = array();
1851                 $first = true;
1852                 foreach ( $lines as $line ) {
1853                         # If the line is too long, just pretend the entire line is one big word
1854                         # This prevents resource exhaustion problems
1855                         if ( $first ) {
1856                                 $first = false;
1857                         } else {
1858                                 $words[] = "\n";
1859                                 $stripped[] = "\n";
1860                         }
1861                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1862                                 $words[] = $line;
1863                                 $stripped[] = $line;
1864                         } else {
1865                                 $m = array();
1866                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1867                                         $line, $m))
1868                                 {
1869                                         $words = array_merge( $words, $m[0] );
1870                                         $stripped = array_merge( $stripped, $m[1] );
1871                                 }
1872                         }
1873                 }
1874                 wfProfileOut( __METHOD__ );
1875                 return array($words, $stripped);
1876         }
1877
1878         function orig () {
1879                 wfProfileIn( __METHOD__ );
1880                 $orig = new _HWLDF_WordAccumulator;
1881
1882                 foreach ($this->edits as $edit) {
1883                         if ($edit->type == 'copy')
1884                                 $orig->addWords($edit->orig);
1885                         elseif ($edit->orig)
1886                                 $orig->addWords($edit->orig, 'del');
1887                 }
1888                 $lines = $orig->getLines();
1889                 wfProfileOut( __METHOD__ );
1890                 return $lines;
1891         }
1892
1893         function closing () {
1894                 wfProfileIn( __METHOD__ );
1895                 $closing = new _HWLDF_WordAccumulator;
1896
1897                 foreach ($this->edits as $edit) {
1898                         if ($edit->type == 'copy')
1899                                 $closing->addWords($edit->closing);
1900                         elseif ($edit->closing)
1901                                 $closing->addWords($edit->closing, 'ins');
1902                 }
1903                 $lines = $closing->getLines();
1904                 wfProfileOut( __METHOD__ );
1905                 return $lines;
1906         }
1907 }
1908
1909 /**
1910  * Wikipedia Table style diff formatter.
1911  * @todo document
1912  * @private
1913  * @ingroup DifferenceEngine
1914  */
1915 class TableDiffFormatter extends DiffFormatter {
1916         function TableDiffFormatter() {
1917                 $this->leading_context_lines = 2;
1918                 $this->trailing_context_lines = 2;
1919         }
1920
1921         public static function escapeWhiteSpace( $msg ) {
1922                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
1923                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
1924                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
1925                 return $msg;
1926         }
1927
1928         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1929                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
1930                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
1931                 return $r;
1932         }
1933
1934         function _start_block( $header ) {
1935                 echo $header;
1936         }
1937
1938         function _end_block() {
1939         }
1940
1941         function _lines( $lines, $prefix=' ', $color='white' ) {
1942         }
1943
1944         # HTML-escape parameter before calling this
1945         function addedLine( $line ) {
1946                 return $this->wrapLine( '+', 'diff-addedline', $line );
1947         }
1948
1949         # HTML-escape parameter before calling this
1950         function deletedLine( $line ) {
1951                 return $this->wrapLine( '-', 'diff-deletedline', $line );
1952         }
1953
1954         # HTML-escape parameter before calling this
1955         function contextLine( $line ) {
1956                 return $this->wrapLine( ' ', 'diff-context', $line );
1957         }
1958
1959         private function wrapLine( $marker, $class, $line ) {
1960                 if( $line !== '' ) {
1961                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
1962                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
1963                 }
1964                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
1965         }
1966
1967         function emptyLine() {
1968                 return '<td colspan="2">&nbsp;</td>';
1969         }
1970
1971         function _added( $lines ) {
1972                 foreach ($lines as $line) {
1973                         echo '<tr>' . $this->emptyLine() .
1974                                 $this->addedLine( '<ins class="diffchange">' .
1975                                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
1976                 }
1977         }
1978
1979         function _deleted($lines) {
1980                 foreach ($lines as $line) {
1981                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
1982                                 htmlspecialchars ( $line ) . '</del>' ) .
1983                           $this->emptyLine() . "</tr>\n";
1984                 }
1985         }
1986
1987         function _context( $lines ) {
1988                 foreach ($lines as $line) {
1989                         echo '<tr>' .
1990                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1991                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1992                 }
1993         }
1994
1995         function _changed( $orig, $closing ) {
1996                 wfProfileIn( __METHOD__ );
1997
1998                 $diff = new WordLevelDiff( $orig, $closing );
1999                 $del = $diff->orig();
2000                 $add = $diff->closing();
2001
2002                 # Notice that WordLevelDiff returns HTML-escaped output.
2003                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2004
2005                 while ( $line = array_shift( $del ) ) {
2006                         $aline = array_shift( $add );
2007                         echo '<tr>' . $this->deletedLine( $line ) .
2008                                 $this->addedLine( $aline ) . "</tr>\n";
2009                 }
2010                 foreach ($add as $line) {       # If any leftovers
2011                         echo '<tr>' . $this->emptyLine() .
2012                                 $this->addedLine( $line ) . "</tr>\n";
2013                 }
2014                 wfProfileOut( __METHOD__ );
2015         }
2016 }