includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * @defgroup DifferenceEngine DifferenceEngine
   4  */
   5
   6 /**
   7  * Constant to indicate diff cache compatibility.
   8  * Bump this when changing the diff formatting in a way that
   9  * fixes important bugs or such to force cached diff views to
  10  * clear.
  11  */
  12 define( 'MW_DIFF_VERSION', '1.11a' );
  13
  14 /**
  15  * @todo document
  16  * @ingroup DifferenceEngine
  17  */
  18 class DifferenceEngine {
  19         /**#@+
  20          * @private
  21          */
  22         var $mOldid, $mNewid, $mTitle;
  23         var $mOldtitle, $mNewtitle, $mPagetitle;
  24         var $mOldtext, $mNewtext;
  25         var $mOldPage, $mNewPage;
  26         var $mRcidMarkPatrolled;
  27         var $mOldRev, $mNewRev;
  28         var $mRevisionsLoaded = false; // Have the revisions been loaded
  29         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  30         /**#@-*/
  31
  32         /**
  33          * Constructor
  34          * @param $titleObj Title object that the diff is associated with
  35          * @param $old Integer: old ID we want to show and diff with.
  36          * @param $new String: either 'prev' or 'next'.
  37          * @param $rcid Integer: ??? FIXME (default 0)
  38          * @param $refreshCache boolean If set, refreshes the diff cache
  39          */
  40         function __construct( $titleObj = null, $old = 0, $new = 0, $rcid = 0, $refreshCache = false ) {
  41                 $this->mTitle = $titleObj;
  42                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  43
  44                 if ( 'prev' === $new ) {
  45                         # Show diff between revision $old and the previous one.
  46                         # Get previous one from DB.
  47                         #
  48                         $this->mNewid = intval($old);
  49
  50                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  51
  52                 } elseif ( 'next' === $new ) {
  53                         # Show diff between revision $old and the previous one.
  54                         # Get previous one from DB.
  55                         #
  56                         $this->mOldid = intval($old);
  57                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  58                         if ( false === $this->mNewid ) {
  59                                 # if no result, NewId points to the newest old revision. The only newer
  60                                 # revision is cur, which is "0".
  61                                 $this->mNewid = 0;
  62                         }
  63
  64                 } else {
  65                         $this->mOldid = intval($old);
  66                         $this->mNewid = intval($new);
  67                 }
  68                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  69                 $this->mRefreshCache = $refreshCache;
  70         }
  71
  72         function getTitle() {
  73                 return $this->mTitle;
  74         }
  75
  76         function showDiffPage( $diffOnly = false ) {
  77                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol, $wgEnableHtmlDiff;
  78                 wfProfileIn( __METHOD__ );
  79
  80
  81                 # If external diffs are enabled both globally and for the user,
  82                 # we'll use the application/x-external-editor interface to call
  83                 # an external diff tool like kompare, kdiff3, etc.
  84                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  85                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  86                         $wgOut->disable();
  87                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  88                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  89                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  90                         $special=$wgLang->getNsText(NS_SPECIAL);
  91                         $control=<<<CONTROL
  92                         [Process]
  93                         Type=Diff text
  94                         Engine=MediaWiki
  95                         Script={$wgServer}{$wgScript}
  96                         Special namespace={$special}
  97
  98                         [File]
  99                         Extension=wiki
 100                         URL=$url1
 101
 102                         [File 2]
 103                         Extension=wiki
 104                         URL=$url2
 105 CONTROL;
 106                         echo($control);
 107                         return;
 108                 }
 109
 110                 $wgOut->setArticleFlag( false );
 111                 if ( ! $this->loadRevisionData() ) {
 112                         $t = $this->mTitle->getPrefixedText();
 113                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 114                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 115                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 116                         wfProfileOut( __METHOD__ );
 117                         return;
 118                 }
 119
 120                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 121
 122                 if ( $this->mNewRev->isCurrent() ) {
 123                         $wgOut->setArticleFlag( true );
 124                 }
 125
 126                 # mOldid is false if the difference engine is called with a "vague" query for
 127                 # a diff between a version V and its previous version V' AND the version V
 128                 # is the first version of that article. In that case, V' does not exist.
 129                 if ( $this->mOldid === false ) {
 130                         $this->showFirstRevision();
 131                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 132                         wfProfileOut( __METHOD__ );
 133                         return;
 134                 }
 135
 136                 $wgOut->suppressQuickbar();
 137
 138                 $oldTitle = $this->mOldPage->getPrefixedText();
 139                 $newTitle = $this->mNewPage->getPrefixedText();
 140                 if( $oldTitle == $newTitle ) {
 141                         $wgOut->setPageTitle( $newTitle );
 142                 } else {
 143                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 144                 }
 145                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 146                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 147
 148                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 149                         $wgOut->loginToUse();
 150                         $wgOut->output();
 151                         wfProfileOut( __METHOD__ );
 152                         exit;
 153                 }
 154
 155                 $sk = $wgUser->getSkin();
 156
 157                 // Check if page is editable
 158                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 159                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed( 'rollback' ) ) {
 160                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 161                 } else {
 162                         $rollback = '';
 163                 }
 164
 165                 // Prepare a change patrol link, if applicable
 166                 if( $wgUseRCPatrol && $wgUser->isAllowed( 'patrol' ) ) {
 167                         // If we've been given an explicit change identifier, use it; saves time
 168                         if( $this->mRcidMarkPatrolled ) {
 169                                 $rcid = $this->mRcidMarkPatrolled;
 170                         } else {
 171                                 // Look for an unpatrolled change corresponding to this diff
 172                                 $db = wfGetDB( DB_SLAVE );
 173                                 $change = RecentChange::newFromConds(
 174                                 array(
 175                                 // Add redundant user,timestamp condition so we can use the existing index
 176                                                 'rc_user_text'  => $this->mNewRev->getRawUserText(),
 177                                                 'rc_timestamp' => $db->timestamp( $this->mNewRev->getTimestamp() ),
 178                                                 'rc_this_oldid' => $this->mNewid,
 179                                                 'rc_last_oldid' => $this->mOldid,
 180                                                 'rc_patrolled' => 0
 181                                 ),
 182                                 __METHOD__
 183                                 );
 184                                 if( $change instanceof RecentChange ) {
 185                                         $rcid = $change->mAttribs['rc_id'];
 186                                 } else {
 187                                         // None found
 188                                         $rcid = 0;
 189                                 }
 190                         }
 191                         // Build the link
 192                         if( $rcid ) {
 193                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj(
 194                                 $this->mTitle,
 195                                 wfMsgHtml( 'markaspatrolleddiff' ),
 196                                         "action=markpatrolled&rcid={$rcid}"
 197                                 ) . ']</span>';
 198                         } else {
 199                                 $patrol = '';
 200                         }
 201                 } else {
 202                         $patrol = '';
 203                 }
 204
 205                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 206                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 207                 if ( $this->mNewRev->isCurrent() ) {
 208                         $nextlink = '&nbsp;';
 209                 } else {
 210                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 211                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 212                 }
 213
 214                 $oldminor = '';
 215                 $newminor = '';
 216
 217                 if ($this->mOldRev->mMinorEdit == 1) {
 218                         $oldminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 219                 }
 220
 221                 if ($this->mNewRev->mMinorEdit == 1) {
 222                         $newminor = Xml::span( wfMsg( 'minoreditletter'), 'minor' ) . ' ';
 223                 }
 224
 225                 $rdel = ''; $ldel = '';
 226                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 227                         $revdel = SpecialPage::getTitleFor( 'Revisiondelete' );
 228                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 229                                 // If revision was hidden from sysops
 230                                 $ldel = wfMsgHtml('rev-delundel');
 231                         } else {
 232                                 $ldel = $sk->makeKnownLinkObj( $revdel,
 233                                 wfMsgHtml('rev-delundel'),
 234                                         'target=' . urlencode( $this->mOldRev->mTitle->getPrefixedDbkey() ) .
 235                                         '&oldid=' . urlencode( $this->mOldRev->getId() ) );
 236                                 // Bolden oversighted content
 237                                 if( $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 238                                 $ldel = "<strong>$ldel</strong>";
 239                         }
 240                         $ldel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$ldel</small>)</tt> ";
 241                         // We don't currently handle well changing the top revision's settings
 242                         if( $this->mNewRev->isCurrent() ) {
 243                                 // If revision was hidden from sysops
 244                                 $rdel = wfMsgHtml('rev-delundel');
 245                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 246                                 // If revision was hidden from sysops
 247                                 $rdel = wfMsgHtml('rev-delundel');
 248                         } else {
 249                                 $rdel = $sk->makeKnownLinkObj( $revdel,
 250                                 wfMsgHtml('rev-delundel'),
 251                                         'target=' . urlencode( $this->mNewRev->mTitle->getPrefixedDbkey() ) .
 252                                         '&oldid=' . urlencode( $this->mNewRev->getId() ) );
 253                                 // Bolden oversighted content
 254                                 if( $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 255                                 $rdel = "<strong>$rdel</strong>";
 256                         }
 257                         $rdel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$rdel</small>)</tt> ";
 258                 }
 259
 260                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 261                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, true ) . "</div>" .
 262                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, true ) . $ldel . "</div>" .
 263                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 264                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 265                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, true ) . " $rollback</div>" .
 266                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, true ) . $rdel . "</div>" .
 267                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 268
 269                 if( $wgEnableHtmlDiff ) {
 270                         $multi = $this->getMultiNotice();
 271                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 272                         $this->renderHtmlDiff();
 273                 } else {
 274
 275                         $this->showDiff( $oldHeader, $newHeader );
 276
 277                         if( !$diffOnly ) {
 278                                 $this->renderNewRevision();
 279                         }
 280                 }
 281                 wfProfileOut( __METHOD__ );
 282         }
 283
 284         /**
 285          * Show the new revision of the page.
 286          */
 287         function renderNewRevision() {
 288                 global $wgOut;
 289                 wfProfileIn( __METHOD__ );
 290
 291                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 292                 #add deleted rev tag if needed
 293                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 294                         $wgOut->addWikiMsg( 'rev-deleted-text-permission' );
 295                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 296                         $wgOut->addWikiMsg( 'rev-deleted-text-view' );
 297                 }
 298
 299                 if( !$this->mNewRev->isCurrent() ) {
 300                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 301                 }
 302
 303                 $this->loadNewText();
 304                 if( is_object( $this->mNewRev ) ) {
 305                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 306                 }
 307
 308                 if ($this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage()) {
 309                         // Stolen from Article::view --AG 2007-10-11
 310
 311                         // Give hooks a chance to customise the output
 312                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 313                                 // Wrap the whole lot in a <pre> and don't parse
 314                                 $m = array();
 315                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 316                                 $wgOut->addHtml( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 317                                 $wgOut->addHtml( htmlspecialchars( $this->mNewtext ) );
 318                                 $wgOut->addHtml( "\n</pre>\n" );
 319                         }
 320                 } else
 321                 $wgOut->addWikiTextTidy( $this->mNewtext );
 322
 323                 if( !$this->mNewRev->isCurrent() ) {
 324                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 325                 }
 326
 327                 wfProfileOut( __METHOD__ );
 328         }
 329
 330
 331         function renderHtmlDiff() {
 332                 global $wgOut, $wgTitle, $wgParser;
 333                 wfProfileIn( __METHOD__ );
 334
 335                 $this->showDiffStyle();
 336
 337                 $wgOut->addHTML( "<hr /><h2>HTML diff</h2>\n" );
 338                 #add deleted rev tag if needed
 339                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 340                         $wgOut->addWikiMsg( 'rev-deleted-text-permission' );
 341                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 342                         $wgOut->addWikiMsg( 'rev-deleted-text-view' );
 343                 }
 344
 345                 if( !$this->mNewRev->isCurrent() ) {
 346                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 347                 }
 348
 349                 $this->loadText();
 350
 351                 // Old revision
 352                 if( is_object( $this->mOldRev ) ) {
 353                         $wgOut->setRevisionId( $this->mOldRev->getId() );
 354                 }
 355
 356                 $popts = $wgOut->parserOptions();
 357                 $oldTidy = $popts->setTidy( true );
 358
 359                 $parserOutput = $wgParser->parse( $this->mOldtext, $wgTitle, $popts, true, true, $wgOut->getRevisionId() );
 360                 $popts->setTidy( $oldTidy );
 361
 362                 //only for new?
 363                 //$wgOut->addParserOutputNoText( $parserOutput );
 364                 $oldHtml = $parserOutput->getText();
 365                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$oldHtml ) );
 366
 367                 // New revision
 368                 if( is_object( $this->mNewRev ) ) {
 369                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 370                 }
 371
 372                 $popts = $wgOut->parserOptions();
 373                 $oldTidy = $popts->setTidy( true );
 374
 375                 $parserOutput = $wgParser->parse( $this->mNewtext, $wgTitle, $popts, true, true, $wgOut->getRevisionId() );
 376                 $popts->setTidy( $oldTidy );
 377
 378                 $wgOut->addParserOutputNoText( $parserOutput );
 379                 $newHtml = $parserOutput->getText();
 380                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$newHtml ) );
 381
 382                 unset($parserOutput, $popts);
 383
 384                 $differ = new HTMLDiffer(new DelegatingContentHandler($wgOut));
 385                 $differ->htmlDiff($oldHtml, $newHtml);
 386
 387                 wfProfileOut( __METHOD__ );
 388         }
 389
 390         /**
 391          * Show the first revision of an article. Uses normal diff headers in
 392          * contrast to normal "old revision" display style.
 393          */
 394         function showFirstRevision() {
 395                 global $wgOut, $wgUser;
 396                 wfProfileIn( __METHOD__ );
 397
 398                 # Get article text from the DB
 399                 #
 400                 if ( ! $this->loadNewText() ) {
 401                         $t = $this->mTitle->getPrefixedText();
 402                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 403                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 404                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 405                         wfProfileOut( __METHOD__ );
 406                         return;
 407                 }
 408                 if ( $this->mNewRev->isCurrent() ) {
 409                         $wgOut->setArticleFlag( true );
 410                 }
 411
 412                 # Check if user is allowed to look at this page. If not, bail out.
 413                 #
 414                 if ( !( $this->mTitle->userCanRead() ) ) {
 415                         $wgOut->loginToUse();
 416                         $wgOut->output();
 417                         wfProfileOut( __METHOD__ );
 418                         exit;
 419                 }
 420
 421                 # Prepare the header box
 422                 #
 423                 $sk = $wgUser->getSkin();
 424
 425                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 426                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 427                 $sk->revUserTools( $this->mNewRev ) . "<br />" .
 428                 $sk->revComment( $this->mNewRev ) . "<br />" .
 429                 $nextlink . "</div>\n";
 430
 431                 $wgOut->addHTML( $header );
 432
 433                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 434                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 435
 436                 wfProfileOut( __METHOD__ );
 437         }
 438
 439         /**
 440          * Get the diff text, send it to $wgOut
 441          * Returns false if the diff could not be generated, otherwise returns true
 442          */
 443         function showDiff( $otitle, $ntitle ) {
 444                 global $wgOut;
 445                 $diff = $this->getDiff( $otitle, $ntitle );
 446                 if ( $diff === false ) {
 447                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 448                         return false;
 449                 } else {
 450                         $this->showDiffStyle();
 451                         $wgOut->addHTML( $diff );
 452                         return true;
 453                 }
 454         }
 455
 456         /**
 457          * Add style sheets and supporting JS for diff display.
 458          */
 459         function showDiffStyle() {
 460                 global $wgStylePath, $wgStyleVersion, $wgOut;
 461                 $wgOut->addStyle( 'common/diff.css' );
 462
 463                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 464                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 465         }
 466
 467         /**
 468          * Get complete diff table, including header
 469          *
 470          * @param Title $otitle Old title
 471          * @param Title $ntitle New title
 472          * @return mixed
 473          */
 474         function getDiff( $otitle, $ntitle ) {
 475                 $body = $this->getDiffBody();
 476                 if ( $body === false ) {
 477                         return false;
 478                 } else {
 479                         $multi = $this->getMultiNotice();
 480                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 481                 }
 482         }
 483
 484         /**
 485          * Get the diff table body, without header
 486          *
 487          * @return mixed
 488          */
 489         function getDiffBody() {
 490                 global $wgMemc;
 491                 wfProfileIn( __METHOD__ );
 492                 // Check if the diff should be hidden from this user
 493                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 494                         return '';
 495                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 496                         return '';
 497                 }
 498                 // Cacheable?
 499                 $key = false;
 500                 if ( $this->mOldid && $this->mNewid ) {
 501                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 502                         // Try cache
 503                         if ( !$this->mRefreshCache ) {
 504                                 $difftext = $wgMemc->get( $key );
 505                                 if ( $difftext ) {
 506                                         wfIncrStats( 'diff_cache_hit' );
 507                                         $difftext = $this->localiseLineNumbers( $difftext );
 508                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 509                                         wfProfileOut( __METHOD__ );
 510                                         return $difftext;
 511                                 }
 512                         } // don't try to load but save the result
 513                 }
 514
 515                 // Loadtext is permission safe, this just clears out the diff
 516                 if ( !$this->loadText() ) {
 517                         wfProfileOut( __METHOD__ );
 518                         return false;
 519                 }
 520
 521                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 522
 523                 // Save to cache for 7 days
 524                 if ( $key !== false && $difftext !== false ) {
 525                         wfIncrStats( 'diff_cache_miss' );
 526                         $wgMemc->set( $key, $difftext, 7*86400 );
 527                 } else {
 528                         wfIncrStats( 'diff_uncacheable' );
 529                 }
 530                 // Replace line numbers with the text in the user's language
 531                 if ( $difftext !== false ) {
 532                         $difftext = $this->localiseLineNumbers( $difftext );
 533                 }
 534                 wfProfileOut( __METHOD__ );
 535                 return $difftext;
 536         }
 537
 538         /**
 539          * Generate a diff, no caching
 540          * $otext and $ntext must be already segmented
 541          */
 542         function generateDiffBody( $otext, $ntext ) {
 543                 global $wgExternalDiffEngine, $wgContLang;
 544
 545                 $otext = str_replace( "\r\n", "\n", $otext );
 546                 $ntext = str_replace( "\r\n", "\n", $ntext );
 547
 548                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 549                         # For historical reasons, external diff engine expects
 550                         # input text to be HTML-escaped already
 551                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 552                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 553                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 554                                 dl('php_wikidiff.so');
 555                         }
 556                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 557                         $this->debug( 'wikidiff1' );
 558                 }
 559
 560                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 561                         # Better external diff engine, the 2 may some day be dropped
 562                         # This one does the escaping and segmenting itself
 563                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 564                                 wfProfileIn( __METHOD__ . "-dl" );
 565                                 @dl('php_wikidiff2.so');
 566                                 wfProfileOut( __METHOD__ . "-dl" );
 567                         }
 568                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 569                                 wfProfileIn( 'wikidiff2_do_diff' );
 570                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 571                                 $text .= $this->debug( 'wikidiff2' );
 572                                 wfProfileOut( 'wikidiff2_do_diff' );
 573                                 return $text;
 574                         }
 575                 }
 576                 if ( $wgExternalDiffEngine != 'wikidiff3' && $wgExternalDiffEngine !== false ) {
 577                         # Diff via the shell
 578                         global $wgTmpDirectory;
 579                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 580                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 581
 582                         $tempFile1 = fopen( $tempName1, "w" );
 583                         if ( !$tempFile1 ) {
 584                                 wfProfileOut( __METHOD__ );
 585                                 return false;
 586                         }
 587                         $tempFile2 = fopen( $tempName2, "w" );
 588                         if ( !$tempFile2 ) {
 589                                 wfProfileOut( __METHOD__ );
 590                                 return false;
 591                         }
 592                         fwrite( $tempFile1, $otext );
 593                         fwrite( $tempFile2, $ntext );
 594                         fclose( $tempFile1 );
 595                         fclose( $tempFile2 );
 596                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 597                         wfProfileIn( __METHOD__ . "-shellexec" );
 598                         $difftext = wfShellExec( $cmd );
 599                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 600                         wfProfileOut( __METHOD__ . "-shellexec" );
 601                         unlink( $tempName1 );
 602                         unlink( $tempName2 );
 603                         return $difftext;
 604                 }
 605
 606                 # Native PHP diff
 607                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 608                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 609                 $diffs = new Diff( $ota, $nta );
 610                 $formatter = new TableDiffFormatter();
 611                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 612                 $this->debug();
 613         }
 614
 615         /**
 616          * Generate a debug comment indicating diff generating time,
 617          * server node, and generator backend.
 618          */
 619         protected function debug( $generator="internal" ) {
 620                 global $wgShowHostnames, $wgNodeName;
 621                 $data = array( $generator );
 622                 if( $wgShowHostnames ) {
 623                         $data[] = $wgNodeName;
 624                 }
 625                 $data[] = wfTimestamp( TS_DB );
 626                 return "<!-- diff generator: " .
 627                 implode( " ",
 628                 array_map(
 629                                         "htmlspecialchars",
 630                 $data ) ) .
 631                         " -->\n";
 632         }
 633
 634         /**
 635          * Replace line numbers with the text in the user's language
 636          */
 637         function localiseLineNumbers( $text ) {
 638                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 639                 array( &$this, 'localiseLineNumbersCb' ), $text );
 640         }
 641
 642         function localiseLineNumbersCb( $matches ) {
 643                 global $wgLang;
 644                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 645         }
 646
 647
 648         /**
 649          * If there are revisions between the ones being compared, return a note saying so.
 650          */
 651         function getMultiNotice() {
 652                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 653                 return '';
 654
 655                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 656                         // Comparing two different pages? Count would be meaningless.
 657                         return '';
 658                 }
 659
 660                 $oldid = $this->mOldRev->getId();
 661                 $newid = $this->mNewRev->getId();
 662                 if ( $oldid > $newid ) {
 663                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 664                 }
 665
 666                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 667                 if ( !$n )
 668                 return '';
 669
 670                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 671         }
 672
 673
 674         /**
 675          * Add the header to a diff body
 676          */
 677         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 678                 $header = "
 679                 <table class='diff'>
 680                 <col class='diff-marker' />
 681                 <col class='diff-content' />
 682                 <col class='diff-marker' />
 683                 <col class='diff-content' />
 684                 <tr valign='top'>
 685                 <td colspan='2' class='diff-otitle'>{$otitle}</td>
 686                 <td colspan='2' class='diff-ntitle'>{$ntitle}</td>
 687                 </tr>
 688                 ";
 689
 690                 if ( $multi != '' )
 691                 $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 692
 693                 return $header . $diff . "</table>";
 694         }
 695
 696         /**
 697          * Use specified text instead of loading from the database
 698          */
 699         function setText( $oldText, $newText ) {
 700                 $this->mOldtext = $oldText;
 701                 $this->mNewtext = $newText;
 702                 $this->mTextLoaded = 2;
 703         }
 704
 705         /**
 706          * Load revision metadata for the specified articles. If newid is 0, then compare
 707          * the old article in oldid to the current article; if oldid is 0, then
 708          * compare the current article to the immediately previous one (ignoring the
 709          * value of newid).
 710          *
 711          * If oldid is false, leave the corresponding revision object set
 712          * to false. This is impossible via ordinary user input, and is provided for
 713          * API convenience.
 714          */
 715         function loadRevisionData() {
 716                 global $wgLang;
 717                 if ( $this->mRevisionsLoaded ) {
 718                         return true;
 719                 } else {
 720                         // Whether it succeeds or fails, we don't want to try again
 721                         $this->mRevisionsLoaded = true;
 722                 }
 723
 724                 // Load the new revision object
 725                 $this->mNewRev = $this->mNewid
 726                 ? Revision::newFromId( $this->mNewid )
 727                 : Revision::newFromTitle( $this->mTitle );
 728                 if( !$this->mNewRev instanceof Revision )
 729                 return false;
 730
 731                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 732                 $this->mNewid = $this->mNewRev->getId();
 733
 734                 // Check if page is editable
 735                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 736
 737                 // Set assorted variables
 738                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 739                 $this->mNewPage = $this->mNewRev->getTitle();
 740                 if( $this->mNewRev->isCurrent() ) {
 741                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 742                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 743                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 744
 745                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)";
 746                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 747
 748                 } else {
 749                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 750                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 751                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 752
 753                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 754                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 755                 }
 756                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 757                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 758                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 759                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 760                 }
 761
 762                 // Load the old revision object
 763                 $this->mOldRev = false;
 764                 if( $this->mOldid ) {
 765                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 766                 } elseif ( $this->mOldid === 0 ) {
 767                         $rev = $this->mNewRev->getPrevious();
 768                         if( $rev ) {
 769                                 $this->mOldid = $rev->getId();
 770                                 $this->mOldRev = $rev;
 771                         } else {
 772                                 // No previous revision; mark to show as first-version only.
 773                                 $this->mOldid = false;
 774                                 $this->mOldRev = false;
 775                         }
 776                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 777
 778                 if( is_null( $this->mOldRev ) ) {
 779                         return false;
 780                 }
 781
 782                 if ( $this->mOldRev ) {
 783                         $this->mOldPage = $this->mOldRev->getTitle();
 784
 785                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 786                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 787                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 788                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 789
 790                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 791                         . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 792                         // Add an "undo" link
 793                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 794                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 795                                 $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 796                         }
 797
 798                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 799                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 800                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 801                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 802                         }
 803                 }
 804
 805                 return true;
 806         }
 807
 808         /**
 809          * Load the text of the revisions, as well as revision data.
 810          */
 811         function loadText() {
 812                 if ( $this->mTextLoaded == 2 ) {
 813                         return true;
 814                 } else {
 815                         // Whether it succeeds or fails, we don't want to try again
 816                         $this->mTextLoaded = 2;
 817                 }
 818
 819                 if ( !$this->loadRevisionData() ) {
 820                         return false;
 821                 }
 822                 if ( $this->mOldRev ) {
 823                         $this->mOldtext = $this->mOldRev->revText();
 824                         if ( $this->mOldtext === false ) {
 825                                 return false;
 826                         }
 827                 }
 828                 if ( $this->mNewRev ) {
 829                         $this->mNewtext = $this->mNewRev->revText();
 830                         if ( $this->mNewtext === false ) {
 831                                 return false;
 832                         }
 833                 }
 834                 return true;
 835         }
 836
 837         /**
 838          * Load the text of the new revision, not the old one
 839          */
 840         function loadNewText() {
 841                 if ( $this->mTextLoaded >= 1 ) {
 842                         return true;
 843                 } else {
 844                         $this->mTextLoaded = 1;
 845                 }
 846                 if ( !$this->loadRevisionData() ) {
 847                         return false;
 848                 }
 849                 $this->mNewtext = $this->mNewRev->getText();
 850                 return true;
 851         }
 852
 853
 854 }
 855
 856 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 857 //
 858 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 859 // You may copy this code freely under the conditions of the GPL.
 860 //
 861
 862 define('USE_ASSERTS', function_exists('assert'));
 863
 864 /**
 865  * @todo document
 866  * @private
 867  * @ingroup DifferenceEngine
 868  */
 869 class _DiffOp {
 870         var $type;
 871         var $orig;
 872         var $closing;
 873
 874         function reverse() {
 875                 trigger_error('pure virtual', E_USER_ERROR);
 876         }
 877
 878         function norig() {
 879                 return $this->orig ? sizeof($this->orig) : 0;
 880         }
 881
 882         function nclosing() {
 883                 return $this->closing ? sizeof($this->closing) : 0;
 884         }
 885 }
 886
 887 /**
 888  * @todo document
 889  * @private
 890  * @ingroup DifferenceEngine
 891  */
 892 class _DiffOp_Copy extends _DiffOp {
 893         var $type = 'copy';
 894
 895         function _DiffOp_Copy ($orig, $closing = false) {
 896                 if (!is_array($closing))
 897                 $closing = $orig;
 898                 $this->orig = $orig;
 899                 $this->closing = $closing;
 900         }
 901
 902         function reverse() {
 903                 return new _DiffOp_Copy($this->closing, $this->orig);
 904         }
 905 }
 906
 907 /**
 908  * @todo document
 909  * @private
 910  * @ingroup DifferenceEngine
 911  */
 912 class _DiffOp_Delete extends _DiffOp {
 913         var $type = 'delete';
 914
 915         function _DiffOp_Delete ($lines) {
 916                 $this->orig = $lines;
 917                 $this->closing = false;
 918         }
 919
 920         function reverse() {
 921                 return new _DiffOp_Add($this->orig);
 922         }
 923 }
 924
 925 /**
 926  * @todo document
 927  * @private
 928  * @ingroup DifferenceEngine
 929  */
 930 class _DiffOp_Add extends _DiffOp {
 931         var $type = 'add';
 932
 933         function _DiffOp_Add ($lines) {
 934                 $this->closing = $lines;
 935                 $this->orig = false;
 936         }
 937
 938         function reverse() {
 939                 return new _DiffOp_Delete($this->closing);
 940         }
 941 }
 942
 943 /**
 944  * @todo document
 945  * @private
 946  * @ingroup DifferenceEngine
 947  */
 948 class _DiffOp_Change extends _DiffOp {
 949         var $type = 'change';
 950
 951         function _DiffOp_Change ($orig, $closing) {
 952                 $this->orig = $orig;
 953                 $this->closing = $closing;
 954         }
 955
 956         function reverse() {
 957                 return new _DiffOp_Change($this->closing, $this->orig);
 958         }
 959 }
 960
 961 /**
 962  * Class used internally by Diff to actually compute the diffs.
 963  *
 964  * The algorithm used here is mostly lifted from the perl module
 965  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 966  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 967  *
 968  * More ideas are taken from:
 969  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 970  *
 971  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 972  * diffutils-2.7, which can be found at:
 973  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 974  *
 975  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 976  * are my own.
 977  *
 978  * Line length limits for robustness added by Tim Starling, 2005-08-31
 979  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
 980  *
 981  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
 982  * @private
 983  * @ingroup DifferenceEngine
 984  */
 985 class _DiffEngine {
 986
 987         const MAX_XREF_LENGTH =  10000;
 988
 989         function diff ($from_lines, $to_lines){
 990                 wfProfileIn( __METHOD__ );
 991
 992                 // Diff and store locally
 993                 $this->diff_local($from_lines, $to_lines);
 994
 995                 // Merge edits when possible
 996                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 997                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 998
 999                 // Compute the edit operations.
1000                 $n_from = sizeof($from_lines);
1001                 $n_to = sizeof($to_lines);
1002
1003                 $edits = array();
1004                 $xi = $yi = 0;
1005                 while ($xi < $n_from || $yi < $n_to) {
1006                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
1007                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
1008
1009                         // Skip matching "snake".
1010                         $copy = array();
1011                         while ( $xi < $n_from && $yi < $n_to
1012                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
1013                                 $copy[] = $from_lines[$xi++];
1014                                 ++$yi;
1015                         }
1016                         if ($copy)
1017                         $edits[] = new _DiffOp_Copy($copy);
1018
1019                         // Find deletes & adds.
1020                         $delete = array();
1021                         while ($xi < $n_from && $this->xchanged[$xi])
1022                         $delete[] = $from_lines[$xi++];
1023
1024                         $add = array();
1025                         while ($yi < $n_to && $this->ychanged[$yi])
1026                         $add[] = $to_lines[$yi++];
1027
1028                         if ($delete && $add)
1029                         $edits[] = new _DiffOp_Change($delete, $add);
1030                         elseif ($delete)
1031                         $edits[] = new _DiffOp_Delete($delete);
1032                         elseif ($add)
1033                         $edits[] = new _DiffOp_Add($add);
1034                 }
1035                 wfProfileOut( __METHOD__ );
1036                 return $edits;
1037         }
1038
1039         function diff_local ($from_lines, $to_lines) {
1040                 global $wgExternalDiffEngine;
1041                 wfProfileIn( __METHOD__);
1042
1043                 if($wgExternalDiffEngine == 'wikidiff3'){
1044                         // wikidiff3
1045                         $wikidiff3 = new WikiDiff3();
1046                         $wikidiff3->diff($from_lines, $to_lines);
1047                         $this->xchanged = $wikidiff3->removed;
1048                         $this->ychanged = $wikidiff3->added;
1049                         unset($wikidiff3);
1050                 }else{
1051                         // old diff
1052                         $n_from = sizeof($from_lines);
1053                         $n_to = sizeof($to_lines);
1054                         $this->xchanged = $this->ychanged = array();
1055                         $this->xv = $this->yv = array();
1056                         $this->xind = $this->yind = array();
1057                         unset($this->seq);
1058                         unset($this->in_seq);
1059                         unset($this->lcs);
1060
1061                         // Skip leading common lines.
1062                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
1063                                 if ($from_lines[$skip] !== $to_lines[$skip])
1064                                 break;
1065                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
1066                         }
1067                         // Skip trailing common lines.
1068                         $xi = $n_from; $yi = $n_to;
1069                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
1070                                 if ($from_lines[$xi] !== $to_lines[$yi])
1071                                 break;
1072                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
1073                         }
1074
1075                         // Ignore lines which do not exist in both files.
1076                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1077                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
1078                         }
1079
1080                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
1081                                 $line = $to_lines[$yi];
1082                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
1083                                 continue;
1084                                 $yhash[$this->_line_hash($line)] = 1;
1085                                 $this->yv[] = $line;
1086                                 $this->yind[] = $yi;
1087                         }
1088                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1089                                 $line = $from_lines[$xi];
1090                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
1091                                 continue;
1092                                 $this->xv[] = $line;
1093                                 $this->xind[] = $xi;
1094                         }
1095
1096                         // Find the LCS.
1097                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
1098                 }
1099                 wfProfileOut( __METHOD__ );
1100         }
1101
1102         /**
1103          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1104          */
1105         function _line_hash( $line ) {
1106                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1107                         return md5( $line );
1108                 } else {
1109                         return $line;
1110                 }
1111         }
1112
1113         /* Divide the Largest Common Subsequence (LCS) of the sequences
1114          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1115          * sized segments.
1116          *
1117          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1118          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1119          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1120          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1121          * that (X0, Y0) == (XOFF, YOFF) and
1122          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1123          *
1124          * This function assumes that the first lines of the specified portions
1125          * of the two files do not match, and likewise that the last lines do not
1126          * match.  The caller must trim matching lines from the beginning and end
1127          * of the portions it is going to specify.
1128          */
1129         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1130                 $flip = false;
1131
1132                 if ($xlim - $xoff > $ylim - $yoff) {
1133                         // Things seems faster (I'm not sure I understand why)
1134                         // when the shortest sequence in X.
1135                         $flip = true;
1136                         list ($xoff, $xlim, $yoff, $ylim)
1137                         = array( $yoff, $ylim, $xoff, $xlim);
1138                 }
1139
1140                 if ($flip)
1141                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1142                 $ymatches[$this->xv[$i]][] = $i;
1143                 else
1144                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1145                 $ymatches[$this->yv[$i]][] = $i;
1146
1147                 $this->lcs = 0;
1148                 $this->seq[0]= $yoff - 1;
1149                 $this->in_seq = array();
1150                 $ymids[0] = array();
1151
1152                 $numer = $xlim - $xoff + $nchunks - 1;
1153                 $x = $xoff;
1154                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1155                         if ($chunk > 0)
1156                         for ($i = 0; $i <= $this->lcs; $i++)
1157                         $ymids[$i][$chunk-1] = $this->seq[$i];
1158
1159                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1160                         for ( ; $x < $x1; $x++) {
1161                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1162                                 if (empty($ymatches[$line]))
1163                                 continue;
1164                                 $matches = $ymatches[$line];
1165                                 reset($matches);
1166                                 while (list ($junk, $y) = each($matches))
1167                                 if (empty($this->in_seq[$y])) {
1168                                         $k = $this->_lcs_pos($y);
1169                                         USE_ASSERTS && assert($k > 0);
1170                                         $ymids[$k] = $ymids[$k-1];
1171                                         break;
1172                                 }
1173                                 while (list ( /* $junk */, $y) = each($matches)) {
1174                                         if ($y > $this->seq[$k-1]) {
1175                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1176                                                 // Optimization: this is a common case:
1177                                                 //      next match is just replacing previous match.
1178                                                 $this->in_seq[$this->seq[$k]] = false;
1179                                                 $this->seq[$k] = $y;
1180                                                 $this->in_seq[$y] = 1;
1181                                         } else if (empty($this->in_seq[$y])) {
1182                                                 $k = $this->_lcs_pos($y);
1183                                                 USE_ASSERTS && assert($k > 0);
1184                                                 $ymids[$k] = $ymids[$k-1];
1185                                         }
1186                                 }
1187                         }
1188                 }
1189
1190                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1191                 $ymid = $ymids[$this->lcs];
1192                 for ($n = 0; $n < $nchunks - 1; $n++) {
1193                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1194                         $y1 = $ymid[$n] + 1;
1195                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1196                 }
1197                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1198
1199                 return array($this->lcs, $seps);
1200         }
1201
1202         function _lcs_pos ($ypos) {
1203                 $end = $this->lcs;
1204                 if ($end == 0 || $ypos > $this->seq[$end]) {
1205                         $this->seq[++$this->lcs] = $ypos;
1206                         $this->in_seq[$ypos] = 1;
1207                         return $this->lcs;
1208                 }
1209
1210                 $beg = 1;
1211                 while ($beg < $end) {
1212                         $mid = (int)(($beg + $end) / 2);
1213                         if ( $ypos > $this->seq[$mid] )
1214                         $beg = $mid + 1;
1215                         else
1216                         $end = $mid;
1217                 }
1218
1219                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1220
1221                 $this->in_seq[$this->seq[$end]] = false;
1222                 $this->seq[$end] = $ypos;
1223                 $this->in_seq[$ypos] = 1;
1224                 return $end;
1225         }
1226
1227         /* Find LCS of two sequences.
1228          *
1229          * The results are recorded in the vectors $this->{x,y}changed[], by
1230          * storing a 1 in the element for each line that is an insertion
1231          * or deletion (ie. is not in the LCS).
1232          *
1233          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1234          *
1235          * Note that XLIM, YLIM are exclusive bounds.
1236          * All line numbers are origin-0 and discarded lines are not counted.
1237          */
1238         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1239                 // Slide down the bottom initial diagonal.
1240                 while ($xoff < $xlim && $yoff < $ylim
1241                 && $this->xv[$xoff] == $this->yv[$yoff]) {
1242                         ++$xoff;
1243                         ++$yoff;
1244                 }
1245
1246                 // Slide up the top initial diagonal.
1247                 while ($xlim > $xoff && $ylim > $yoff
1248                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1249                         --$xlim;
1250                         --$ylim;
1251                 }
1252
1253                 if ($xoff == $xlim || $yoff == $ylim)
1254                 $lcs = 0;
1255                 else {
1256                         // This is ad hoc but seems to work well.
1257                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1258                         //$nchunks = max(2,min(8,(int)$nchunks));
1259                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1260                         list ($lcs, $seps)
1261                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1262                 }
1263
1264                 if ($lcs == 0) {
1265                         // X and Y sequences have no common subsequence:
1266                         // mark all changed.
1267                         while ($yoff < $ylim)
1268                         $this->ychanged[$this->yind[$yoff++]] = 1;
1269                         while ($xoff < $xlim)
1270                         $this->xchanged[$this->xind[$xoff++]] = 1;
1271                 } else {
1272                         // Use the partitions to split this problem into subproblems.
1273                         reset($seps);
1274                         $pt1 = $seps[0];
1275                         while ($pt2 = next($seps)) {
1276                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1277                                 $pt1 = $pt2;
1278                         }
1279                 }
1280         }
1281
1282         /* Adjust inserts/deletes of identical lines to join changes
1283          * as much as possible.
1284          *
1285          * We do something when a run of changed lines include a
1286          * line at one end and has an excluded, identical line at the other.
1287          * We are free to choose which identical line is included.
1288          * `compareseq' usually chooses the one at the beginning,
1289          * but usually it is cleaner to consider the following identical line
1290          * to be the "change".
1291          *
1292          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1293          */
1294         function _shift_boundaries ($lines, &$changed, $other_changed) {
1295                 wfProfileIn( __METHOD__ );
1296                 $i = 0;
1297                 $j = 0;
1298
1299                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1300                 $len = sizeof($lines);
1301                 $other_len = sizeof($other_changed);
1302
1303                 while (1) {
1304                         /*
1305                          * Scan forwards to find beginning of another run of changes.
1306                          * Also keep track of the corresponding point in the other file.
1307                          *
1308                          * Throughout this code, $i and $j are adjusted together so that
1309                          * the first $i elements of $changed and the first $j elements
1310                          * of $other_changed both contain the same number of zeros
1311                          * (unchanged lines).
1312                          * Furthermore, $j is always kept so that $j == $other_len or
1313                          * $other_changed[$j] == false.
1314                          */
1315                         while ($j < $other_len && $other_changed[$j])
1316                         $j++;
1317
1318                         while ($i < $len && ! $changed[$i]) {
1319                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1320                                 $i++; $j++;
1321                                 while ($j < $other_len && $other_changed[$j])
1322                                 $j++;
1323                         }
1324
1325                         if ($i == $len)
1326                         break;
1327
1328                         $start = $i;
1329
1330                         // Find the end of this run of changes.
1331                         while (++$i < $len && $changed[$i])
1332                         continue;
1333
1334                         do {
1335                                 /*
1336                                  * Record the length of this run of changes, so that
1337                                  * we can later determine whether the run has grown.
1338                                  */
1339                                 $runlength = $i - $start;
1340
1341                                 /*
1342                                  * Move the changed region back, so long as the
1343                                  * previous unchanged line matches the last changed one.
1344                                  * This merges with previous changed regions.
1345                                  */
1346                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1347                                         $changed[--$start] = 1;
1348                                         $changed[--$i] = false;
1349                                         while ($start > 0 && $changed[$start - 1])
1350                                         $start--;
1351                                         USE_ASSERTS && assert('$j > 0');
1352                                         while ($other_changed[--$j])
1353                                         continue;
1354                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1355                                 }
1356
1357                                 /*
1358                                  * Set CORRESPONDING to the end of the changed run, at the last
1359                                  * point where it corresponds to a changed run in the other file.
1360                                  * CORRESPONDING == LEN means no such point has been found.
1361                                  */
1362                                 $corresponding = $j < $other_len ? $i : $len;
1363
1364                                 /*
1365                                  * Move the changed region forward, so long as the
1366                                  * first changed line matches the following unchanged one.
1367                                  * This merges with following changed regions.
1368                                  * Do this second, so that if there are no merges,
1369                                  * the changed region is moved forward as far as possible.
1370                                  */
1371                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1372                                         $changed[$start++] = false;
1373                                         $changed[$i++] = 1;
1374                                         while ($i < $len && $changed[$i])
1375                                         $i++;
1376
1377                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1378                                         $j++;
1379                                         if ($j < $other_len && $other_changed[$j]) {
1380                                                 $corresponding = $i;
1381                                                 while ($j < $other_len && $other_changed[$j])
1382                                                 $j++;
1383                                         }
1384                                 }
1385                         } while ($runlength != $i - $start);
1386
1387                         /*
1388                          * If possible, move the fully-merged run of changes
1389                          * back to a corresponding run in the other file.
1390                          */
1391                         while ($corresponding < $i) {
1392                                 $changed[--$start] = 1;
1393                                 $changed[--$i] = 0;
1394                                 USE_ASSERTS && assert('$j > 0');
1395                                 while ($other_changed[--$j])
1396                                 continue;
1397                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1398                         }
1399                 }
1400                 wfProfileOut( __METHOD__ );
1401         }
1402 }
1403
1404 /**
1405  * Class representing a 'diff' between two sequences of strings.
1406  * @todo document
1407  * @private
1408  * @ingroup DifferenceEngine
1409  */
1410 class Diff
1411 {
1412         var $edits;
1413
1414         /**
1415          * Constructor.
1416          * Computes diff between sequences of strings.
1417          *
1418          * @param $from_lines array An array of strings.
1419          *                (Typically these are lines from a file.)
1420          * @param $to_lines array An array of strings.
1421          */
1422         function Diff($from_lines, $to_lines) {
1423                 $eng = new _DiffEngine;
1424                 $this->edits = $eng->diff($from_lines, $to_lines);
1425                 //$this->_check($from_lines, $to_lines);
1426         }
1427
1428         /**
1429          * Compute reversed Diff.
1430          *
1431          * SYNOPSIS:
1432          *
1433          *      $diff = new Diff($lines1, $lines2);
1434          *      $rev = $diff->reverse();
1435          * @return object A Diff object representing the inverse of the
1436          *                                original diff.
1437          */
1438         function reverse () {
1439                 $rev = $this;
1440                 $rev->edits = array();
1441                 foreach ($this->edits as $edit) {
1442                         $rev->edits[] = $edit->reverse();
1443                 }
1444                 return $rev;
1445         }
1446
1447         /**
1448          * Check for empty diff.
1449          *
1450          * @return bool True iff two sequences were identical.
1451          */
1452         function isEmpty () {
1453                 foreach ($this->edits as $edit) {
1454                         if ($edit->type != 'copy')
1455                         return false;
1456                 }
1457                 return true;
1458         }
1459
1460         /**
1461          * Compute the length of the Longest Common Subsequence (LCS).
1462          *
1463          * This is mostly for diagnostic purposed.
1464          *
1465          * @return int The length of the LCS.
1466          */
1467         function lcs () {
1468                 $lcs = 0;
1469                 foreach ($this->edits as $edit) {
1470                         if ($edit->type == 'copy')
1471                         $lcs += sizeof($edit->orig);
1472                 }
1473                 return $lcs;
1474         }
1475
1476         /**
1477          * Get the original set of lines.
1478          *
1479          * This reconstructs the $from_lines parameter passed to the
1480          * constructor.
1481          *
1482          * @return array The original sequence of strings.
1483          */
1484         function orig() {
1485                 $lines = array();
1486
1487                 foreach ($this->edits as $edit) {
1488                         if ($edit->orig)
1489                         array_splice($lines, sizeof($lines), 0, $edit->orig);
1490                 }
1491                 return $lines;
1492         }
1493
1494         /**
1495          * Get the closing set of lines.
1496          *
1497          * This reconstructs the $to_lines parameter passed to the
1498          * constructor.
1499          *
1500          * @return array The sequence of strings.
1501          */
1502         function closing() {
1503                 $lines = array();
1504
1505                 foreach ($this->edits as $edit) {
1506                         if ($edit->closing)
1507                         array_splice($lines, sizeof($lines), 0, $edit->closing);
1508                 }
1509                 return $lines;
1510         }
1511
1512         /**
1513          * Check a Diff for validity.
1514          *
1515          * This is here only for debugging purposes.
1516          */
1517         function _check ($from_lines, $to_lines) {
1518                 wfProfileIn( __METHOD__ );
1519                 if (serialize($from_lines) != serialize($this->orig()))
1520                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1521                 if (serialize($to_lines) != serialize($this->closing()))
1522                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1523
1524                 $rev = $this->reverse();
1525                 if (serialize($to_lines) != serialize($rev->orig()))
1526                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
1527                 if (serialize($from_lines) != serialize($rev->closing()))
1528                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1529
1530
1531                 $prevtype = 'none';
1532                 foreach ($this->edits as $edit) {
1533                         if ( $prevtype == $edit->type )
1534                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1535                         $prevtype = $edit->type;
1536                 }
1537
1538                 $lcs = $this->lcs();
1539                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1540                 wfProfileOut( __METHOD__ );
1541         }
1542 }
1543
1544 /**
1545  * @todo document, bad name.
1546  * @private
1547  * @ingroup DifferenceEngine
1548  */
1549 class MappedDiff extends Diff
1550 {
1551         /**
1552          * Constructor.
1553          *
1554          * Computes diff between sequences of strings.
1555          *
1556          * This can be used to compute things like
1557          * case-insensitve diffs, or diffs which ignore
1558          * changes in white-space.
1559          *
1560          * @param $from_lines array An array of strings.
1561          *      (Typically these are lines from a file.)
1562          *
1563          * @param $to_lines array An array of strings.
1564          *
1565          * @param $mapped_from_lines array This array should
1566          *      have the same size number of elements as $from_lines.
1567          *      The elements in $mapped_from_lines and
1568          *      $mapped_to_lines are what is actually compared
1569          *      when computing the diff.
1570          *
1571          * @param $mapped_to_lines array This array should
1572          *      have the same number of elements as $to_lines.
1573          */
1574         function MappedDiff($from_lines, $to_lines,
1575         $mapped_from_lines, $mapped_to_lines) {
1576                 wfProfileIn( __METHOD__ );
1577
1578                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1579                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1580
1581                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1582
1583                 $xi = $yi = 0;
1584                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1585                         $orig = &$this->edits[$i]->orig;
1586                         if (is_array($orig)) {
1587                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1588                                 $xi += sizeof($orig);
1589                         }
1590
1591                         $closing = &$this->edits[$i]->closing;
1592                         if (is_array($closing)) {
1593                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1594                                 $yi += sizeof($closing);
1595                         }
1596                 }
1597                 wfProfileOut( __METHOD__ );
1598         }
1599 }
1600
1601 /**
1602  * A class to format Diffs
1603  *
1604  * This class formats the diff in classic diff format.
1605  * It is intended that this class be customized via inheritance,
1606  * to obtain fancier outputs.
1607  * @todo document
1608  * @private
1609  * @ingroup DifferenceEngine
1610  */
1611 class DiffFormatter {
1612         /**
1613          * Number of leading context "lines" to preserve.
1614          *
1615          * This should be left at zero for this class, but subclasses
1616          * may want to set this to other values.
1617          */
1618         var $leading_context_lines = 0;
1619
1620         /**
1621          * Number of trailing context "lines" to preserve.
1622          *
1623          * This should be left at zero for this class, but subclasses
1624          * may want to set this to other values.
1625          */
1626         var $trailing_context_lines = 0;
1627
1628         /**
1629          * Format a diff.
1630          *
1631          * @param $diff object A Diff object.
1632          * @return string The formatted output.
1633          */
1634         function format($diff) {
1635                 wfProfileIn( __METHOD__ );
1636
1637                 $xi = $yi = 1;
1638                 $block = false;
1639                 $context = array();
1640
1641                 $nlead = $this->leading_context_lines;
1642                 $ntrail = $this->trailing_context_lines;
1643
1644                 $this->_start_diff();
1645
1646                 foreach ($diff->edits as $edit) {
1647                         if ($edit->type == 'copy') {
1648                                 if (is_array($block)) {
1649                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1650                                                 $block[] = $edit;
1651                                         }
1652                                         else{
1653                                                 if ($ntrail) {
1654                                                         $context = array_slice($edit->orig, 0, $ntrail);
1655                                                         $block[] = new _DiffOp_Copy($context);
1656                                                 }
1657                                                 $this->_block($x0, $ntrail + $xi - $x0,
1658                                                 $y0, $ntrail + $yi - $y0,
1659                                                 $block);
1660                                                 $block = false;
1661                                         }
1662                                 }
1663                                 $context = $edit->orig;
1664                         }
1665                         else {
1666                                 if (! is_array($block)) {
1667                                         $context = array_slice($context, sizeof($context) - $nlead);
1668                                         $x0 = $xi - sizeof($context);
1669                                         $y0 = $yi - sizeof($context);
1670                                         $block = array();
1671                                         if ($context)
1672                                         $block[] = new _DiffOp_Copy($context);
1673                                 }
1674                                 $block[] = $edit;
1675                         }
1676
1677                         if ($edit->orig)
1678                         $xi += sizeof($edit->orig);
1679                         if ($edit->closing)
1680                         $yi += sizeof($edit->closing);
1681                 }
1682
1683                 if (is_array($block))
1684                 $this->_block($x0, $xi - $x0,
1685                 $y0, $yi - $y0,
1686                 $block);
1687
1688                 $end = $this->_end_diff();
1689                 wfProfileOut( __METHOD__ );
1690                 return $end;
1691         }
1692
1693         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1694                 wfProfileIn( __METHOD__ );
1695                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1696                 foreach ($edits as $edit) {
1697                         if ($edit->type == 'copy')
1698                         $this->_context($edit->orig);
1699                         elseif ($edit->type == 'add')
1700                         $this->_added($edit->closing);
1701                         elseif ($edit->type == 'delete')
1702                         $this->_deleted($edit->orig);
1703                         elseif ($edit->type == 'change')
1704                         $this->_changed($edit->orig, $edit->closing);
1705                         else
1706                         trigger_error('Unknown edit type', E_USER_ERROR);
1707                 }
1708                 $this->_end_block();
1709                 wfProfileOut( __METHOD__ );
1710         }
1711
1712         function _start_diff() {
1713                 ob_start();
1714         }
1715
1716         function _end_diff() {
1717                 $val = ob_get_contents();
1718                 ob_end_clean();
1719                 return $val;
1720         }
1721
1722         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1723                 if ($xlen > 1)
1724                 $xbeg .= "," . ($xbeg + $xlen - 1);
1725                 if ($ylen > 1)
1726                 $ybeg .= "," . ($ybeg + $ylen - 1);
1727
1728                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1729         }
1730
1731         function _start_block($header) {
1732                 echo $header . "\n";
1733         }
1734
1735         function _end_block() {
1736         }
1737
1738         function _lines($lines, $prefix = ' ') {
1739                 foreach ($lines as $line)
1740                 echo "$prefix $line\n";
1741         }
1742
1743         function _context($lines) {
1744                 $this->_lines($lines);
1745         }
1746
1747         function _added($lines) {
1748                 $this->_lines($lines, '>');
1749         }
1750         function _deleted($lines) {
1751                 $this->_lines($lines, '<');
1752         }
1753
1754         function _changed($orig, $closing) {
1755                 $this->_deleted($orig);
1756                 echo "---\n";
1757                 $this->_added($closing);
1758         }
1759 }
1760
1761 /**
1762  * A formatter that outputs unified diffs
1763  * @ingroup DifferenceEngine
1764  */
1765
1766 class UnifiedDiffFormatter extends DiffFormatter {
1767         var $leading_context_lines = 2;
1768         var $trailing_context_lines = 2;
1769
1770         function _added($lines) {
1771                 $this->_lines($lines, '+');
1772         }
1773         function _deleted($lines) {
1774                 $this->_lines($lines, '-');
1775         }
1776         function _changed($orig, $closing) {
1777                 $this->_deleted($orig);
1778                 $this->_added($closing);
1779         }
1780         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1781                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1782         }
1783 }
1784
1785 /**
1786  * A pseudo-formatter that just passes along the Diff::$edits array
1787  * @ingroup DifferenceEngine
1788  */
1789 class ArrayDiffFormatter extends DiffFormatter {
1790         function format($diff) {
1791                 $oldline = 1;
1792                 $newline = 1;
1793                 $retval = array();
1794                 foreach($diff->edits as $edit)
1795                 switch($edit->type) {
1796                         case 'add':
1797                                 foreach($edit->closing as $l) {
1798                                         $retval[] = array(
1799                                                         'action' => 'add',
1800                                                         'new'=> $l,
1801                                                         'newline' => $newline++
1802                                         );
1803                                 }
1804                                 break;
1805                         case 'delete':
1806                                 foreach($edit->orig as $l) {
1807                                         $retval[] = array(
1808                                                         'action' => 'delete',
1809                                                         'old' => $l,
1810                                                         'oldline' => $oldline++,
1811                                         );
1812                                 }
1813                                 break;
1814                         case 'change':
1815                                 foreach($edit->orig as $i => $l) {
1816                                         $retval[] = array(
1817                                                         'action' => 'change',
1818                                                         'old' => $l,
1819                                                         'new' => @$edit->closing[$i],
1820                                                         'oldline' => $oldline++,
1821                                                         'newline' => $newline++,
1822                                         );
1823                                 }
1824                                 break;
1825                         case 'copy':
1826                                 $oldline += count($edit->orig);
1827                                 $newline += count($edit->orig);
1828                 }
1829                 return $retval;
1830         }
1831 }
1832
1833 /**
1834  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1835  *
1836  */
1837
1838 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1839
1840 /**
1841  * @todo document
1842  * @private
1843  * @ingroup DifferenceEngine
1844  */
1845 class _HWLDF_WordAccumulator {
1846         function _HWLDF_WordAccumulator () {
1847                 $this->_lines = array();
1848                 $this->_line = '';
1849                 $this->_group = '';
1850                 $this->_tag = '';
1851         }
1852
1853         function _flushGroup ($new_tag) {
1854                 if ($this->_group !== '') {
1855                         if ($this->_tag == 'ins')
1856                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1857                         htmlspecialchars ( $this->_group ) . '</ins>';
1858                         elseif ($this->_tag == 'del')
1859                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1860                         htmlspecialchars ( $this->_group ) . '</del>';
1861                         else
1862                         $this->_line .= htmlspecialchars ( $this->_group );
1863                 }
1864                 $this->_group = '';
1865                 $this->_tag = $new_tag;
1866         }
1867
1868         function _flushLine ($new_tag) {
1869                 $this->_flushGroup($new_tag);
1870                 if ($this->_line != '')
1871                 array_push ( $this->_lines, $this->_line );
1872                 else
1873                 # make empty lines visible by inserting an NBSP
1874                 array_push ( $this->_lines, NBSP );
1875                 $this->_line = '';
1876         }
1877
1878         function addWords ($words, $tag = '') {
1879                 if ($tag != $this->_tag)
1880                 $this->_flushGroup($tag);
1881
1882                 foreach ($words as $word) {
1883                         // new-line should only come as first char of word.
1884                         if ($word == '')
1885                         continue;
1886                         if ($word[0] == "\n") {
1887                                 $this->_flushLine($tag);
1888                                 $word = substr($word, 1);
1889                         }
1890                         assert(!strstr($word, "\n"));
1891                         $this->_group .= $word;
1892                 }
1893         }
1894
1895         function getLines() {
1896                 $this->_flushLine('~done');
1897                 return $this->_lines;
1898         }
1899 }
1900
1901 /**
1902  * @todo document
1903  * @private
1904  * @ingroup DifferenceEngine
1905  */
1906 class WordLevelDiff extends MappedDiff {
1907         const MAX_LINE_LENGTH = 10000;
1908
1909         function WordLevelDiff ($orig_lines, $closing_lines) {
1910                 wfProfileIn( __METHOD__ );
1911
1912                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1913                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1914
1915                 $this->MappedDiff($orig_words, $closing_words,
1916                 $orig_stripped, $closing_stripped);
1917                 wfProfileOut( __METHOD__ );
1918         }
1919
1920         function _split($lines) {
1921                 wfProfileIn( __METHOD__ );
1922
1923                 $words = array();
1924                 $stripped = array();
1925                 $first = true;
1926                 foreach ( $lines as $line ) {
1927                         # If the line is too long, just pretend the entire line is one big word
1928                         # This prevents resource exhaustion problems
1929                         if ( $first ) {
1930                                 $first = false;
1931                         } else {
1932                                 $words[] = "\n";
1933                                 $stripped[] = "\n";
1934                         }
1935                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1936                                 $words[] = $line;
1937                                 $stripped[] = $line;
1938                         } else {
1939                                 $m = array();
1940                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1941                                 $line, $m))
1942                                 {
1943                                         $words = array_merge( $words, $m[0] );
1944                                         $stripped = array_merge( $stripped, $m[1] );
1945                                 }
1946                         }
1947                 }
1948                 wfProfileOut( __METHOD__ );
1949                 return array($words, $stripped);
1950         }
1951
1952         function orig () {
1953                 wfProfileIn( __METHOD__ );
1954                 $orig = new _HWLDF_WordAccumulator;
1955
1956                 foreach ($this->edits as $edit) {
1957                         if ($edit->type == 'copy')
1958                         $orig->addWords($edit->orig);
1959                         elseif ($edit->orig)
1960                         $orig->addWords($edit->orig, 'del');
1961                 }
1962                 $lines = $orig->getLines();
1963                 wfProfileOut( __METHOD__ );
1964                 return $lines;
1965         }
1966
1967         function closing () {
1968                 wfProfileIn( __METHOD__ );
1969                 $closing = new _HWLDF_WordAccumulator;
1970
1971                 foreach ($this->edits as $edit) {
1972                         if ($edit->type == 'copy')
1973                         $closing->addWords($edit->closing);
1974                         elseif ($edit->closing)
1975                         $closing->addWords($edit->closing, 'ins');
1976                 }
1977                 $lines = $closing->getLines();
1978                 wfProfileOut( __METHOD__ );
1979                 return $lines;
1980         }
1981 }
1982
1983 /**
1984  * Wikipedia Table style diff formatter.
1985  * @todo document
1986  * @private
1987  * @ingroup DifferenceEngine
1988  */
1989 class TableDiffFormatter extends DiffFormatter {
1990         function TableDiffFormatter() {
1991                 $this->leading_context_lines = 2;
1992                 $this->trailing_context_lines = 2;
1993         }
1994
1995         public static function escapeWhiteSpace( $msg ) {
1996                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
1997                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
1998                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
1999                 return $msg;
2000         }
2001
2002         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
2003                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
2004                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
2005                 return $r;
2006         }
2007
2008         function _start_block( $header ) {
2009                 echo $header;
2010         }
2011
2012         function _end_block() {
2013         }
2014
2015         function _lines( $lines, $prefix=' ', $color='white' ) {
2016         }
2017
2018         # HTML-escape parameter before calling this
2019         function addedLine( $line ) {
2020                 return $this->wrapLine( '+', 'diff-addedline', $line );
2021         }
2022
2023         # HTML-escape parameter before calling this
2024         function deletedLine( $line ) {
2025                 return $this->wrapLine( '-', 'diff-deletedline', $line );
2026         }
2027
2028         # HTML-escape parameter before calling this
2029         function contextLine( $line ) {
2030                 return $this->wrapLine( ' ', 'diff-context', $line );
2031         }
2032
2033         private function wrapLine( $marker, $class, $line ) {
2034                 if( $line !== '' ) {
2035                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
2036                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
2037                 }
2038                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
2039         }
2040
2041         function emptyLine() {
2042                 return '<td colspan="2">&nbsp;</td>';
2043         }
2044
2045         function _added( $lines ) {
2046                 foreach ($lines as $line) {
2047                         echo '<tr>' . $this->emptyLine() .
2048                         $this->addedLine( '<ins class="diffchange">' .
2049                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
2050                 }
2051         }
2052
2053         function _deleted($lines) {
2054                 foreach ($lines as $line) {
2055                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
2056                         htmlspecialchars ( $line ) . '</del>' ) .
2057                         $this->emptyLine() . "</tr>\n";
2058                 }
2059         }
2060
2061         function _context( $lines ) {
2062                 foreach ($lines as $line) {
2063                         echo '<tr>' .
2064                         $this->contextLine( htmlspecialchars ( $line ) ) .
2065                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
2066                 }
2067         }
2068
2069         function _changed( $orig, $closing ) {
2070                 wfProfileIn( __METHOD__ );
2071
2072                 $diff = new WordLevelDiff( $orig, $closing );
2073                 $del = $diff->orig();
2074                 $add = $diff->closing();
2075
2076                 # Notice that WordLevelDiff returns HTML-escaped output.
2077                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2078
2079                 while ( $line = array_shift( $del ) ) {
2080                         $aline = array_shift( $add );
2081                         echo '<tr>' . $this->deletedLine( $line ) .
2082                         $this->addedLine( $aline ) . "</tr>\n";
2083                 }
2084                 foreach ($add as $line) {       # If any leftovers
2085                         echo '<tr>' . $this->emptyLine() .
2086                         $this->addedLine( $line ) . "</tr>\n";
2087                 }
2088                 wfProfileOut( __METHOD__ );
2089         }
2090 }