includes/diff/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         var $mCacheHit = false; // Was the diff fetched from cache?
  31         var $htmldiff;
  32
  33         protected $unhide = false;
  34         /**#@-*/
  35
  36         /**
  37          * Constructor
  38          * @param $titleObj Title object that the diff is associated with
  39          * @param $old Integer: old ID we want to show and diff with.
  40          * @param $new String: either 'prev' or 'next'.
  41          * @param $rcid Integer: ??? FIXME (default 0)
  42          * @param $refreshCache boolean If set, refreshes the diff cache
  43          * @param $htmldiff boolean If set, output using HTMLDiff instead of raw wikicode diff
  44          * @param $unhide boolean If set, allow viewing deleted revs
  45          */
  46         function __construct( $titleObj = null, $old = 0, $new = 0, $rcid = 0,
  47                 $refreshCache = false, $htmldiff = false, $unhide = false )
  48         {
  49                 $this->mTitle = $titleObj;
  50                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  51
  52                 if ( 'prev' === $new ) {
  53                         # Show diff between revision $old and the previous one.
  54                         # Get previous one from DB.
  55                         $this->mNewid = intval($old);
  56                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  57                 } elseif ( 'next' === $new ) {
  58                         # Show diff between revision $old and the next one.
  59                         # Get next one from DB.
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  62                         if ( false === $this->mNewid ) {
  63                                 # if no result, NewId points to the newest old revision. The only newer
  64                                 # revision is cur, which is "0".
  65                                 $this->mNewid = 0;
  66                         }
  67                 } else {
  68                         $this->mOldid = intval($old);
  69                         $this->mNewid = intval($new);
  70                         wfRunHooks( 'NewDifferenceEngine', array(&$titleObj, &$this->mOldid, &$this->mNewid, $old, $new) );
  71                 }
  72                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  73                 $this->mRefreshCache = $refreshCache;
  74                 $this->htmldiff = $htmldiff;
  75                 $this->unhide = $unhide;
  76         }
  77
  78         function getTitle() {
  79                 return $this->mTitle;
  80         }
  81
  82         function wasCacheHit() {
  83                 return $this->mCacheHit;
  84         }
  85
  86         function getOldid() {
  87                 return $this->mOldid;
  88         }
  89
  90         function getNewid() {
  91                 return $this->mNewid;
  92         }
  93
  94         function showDiffPage( $diffOnly = false ) {
  95                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol, $wgEnableHtmlDiff;
  96                 wfProfileIn( __METHOD__ );
  97
  98
  99                 # If external diffs are enabled both globally and for the user,
 100                 # we'll use the application/x-external-editor interface to call
 101                 # an external diff tool like kompare, kdiff3, etc.
 102                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
 103                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
 104                         $wgOut->disable();
 105                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
 106                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
 107                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
 108                         $special=$wgLang->getNsText(NS_SPECIAL);
 109                         $control=<<<CONTROL
 110                         [Process]
 111                         Type=Diff text
 112                         Engine=MediaWiki
 113                         Script={$wgServer}{$wgScript}
 114                         Special namespace={$special}
 115
 116                         [File]
 117                         Extension=wiki
 118                         URL=$url1
 119
 120                         [File 2]
 121                         Extension=wiki
 122                         URL=$url2
 123 CONTROL;
 124                         echo($control);
 125                         return;
 126                 }
 127
 128                 $wgOut->setArticleFlag( false );
 129                 if ( !$this->loadRevisionData() ) {
 130                         $t = $this->mTitle->getPrefixedText();
 131                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 132                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 133                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 134                         wfProfileOut( __METHOD__ );
 135                         return;
 136                 }
 137
 138                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 139
 140                 if ( $this->mNewRev->isCurrent() ) {
 141                         $wgOut->setArticleFlag( true );
 142                 }
 143
 144                 # mOldid is false if the difference engine is called with a "vague" query for
 145                 # a diff between a version V and its previous version V' AND the version V
 146                 # is the first version of that article. In that case, V' does not exist.
 147                 if ( $this->mOldid === false ) {
 148                         $this->showFirstRevision();
 149                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 150                         wfProfileOut( __METHOD__ );
 151                         return;
 152                 }
 153
 154                 $wgOut->suppressQuickbar();
 155
 156                 $oldTitle = $this->mOldPage->getPrefixedText();
 157                 $newTitle = $this->mNewPage->getPrefixedText();
 158                 if( $oldTitle == $newTitle ) {
 159                         $wgOut->setPageTitle( $newTitle );
 160                 } else {
 161                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 162                 }
 163                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 164                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 165
 166                 if ( !$this->mOldPage->userCanRead() || !$this->mNewPage->userCanRead() ) {
 167                         $wgOut->loginToUse();
 168                         $wgOut->output();
 169                         $wgOut->disable();
 170                         wfProfileOut( __METHOD__ );
 171                         return;
 172                 }
 173
 174                 $sk = $wgUser->getSkin();
 175
 176                 // Check if page is editable
 177                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 178                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed( 'rollback' ) ) {
 179                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 180                 } else {
 181                         $rollback = '';
 182                 }
 183
 184                 // Prepare a change patrol link, if applicable
 185                 if( $wgUseRCPatrol && $this->mTitle->userCan('patrol') ) {
 186                         // If we've been given an explicit change identifier, use it; saves time
 187                         if( $this->mRcidMarkPatrolled ) {
 188                                 $rcid = $this->mRcidMarkPatrolled;
 189                                 $rc = RecentChange::newFromId( $rcid );
 190                                 // Already patrolled?
 191                                 $rcid = is_object($rc) && !$rc->getAttribute('rc_patrolled') ? $rcid : 0;
 192                         } else {
 193                                 // Look for an unpatrolled change corresponding to this diff
 194                                 $db = wfGetDB( DB_SLAVE );
 195                                 $change = RecentChange::newFromConds(
 196                                         array(
 197                                         // Redundant user,timestamp condition so we can use the existing index
 198                                                 'rc_user_text'  => $this->mNewRev->getRawUserText(),
 199                                                 'rc_timestamp'  => $db->timestamp( $this->mNewRev->getTimestamp() ),
 200                                                 'rc_this_oldid' => $this->mNewid,
 201                                                 'rc_last_oldid' => $this->mOldid,
 202                                                 'rc_patrolled'  => 0
 203                                         ),
 204                                         __METHOD__
 205                                 );
 206                                 if( $change instanceof RecentChange ) {
 207                                         $rcid = $change->mAttribs['rc_id'];
 208                                         $this->mRcidMarkPatrolled = $rcid;
 209                                 } else {
 210                                         // None found
 211                                         $rcid = 0;
 212                                 }
 213                         }
 214                         // Build the link
 215                         if( $rcid ) {
 216                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj( $this->mTitle,
 217                                         wfMsgHtml( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$rcid}" ) . ']</span>';
 218                         } else {
 219                                 $patrol = '';
 220                         }
 221                 } else {
 222                         $patrol = '';
 223                 }
 224
 225                 $diffOnlyArg = '';
 226                 # Carry over 'diffonly' param via navigation links
 227                 if( $diffOnly != $wgUser->getBoolOption('diffonly') ) {
 228                         $diffOnlyArg = '&diffonly='.$diffOnly;
 229                 }
 230                 $htmldiffarg = $this->htmlDiffArgument();
 231                 # Make "previous revision link"
 232                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 233                         "diff=prev&oldid={$this->mOldid}{$htmldiffarg}{$diffOnlyArg}", '', '', 'id="differences-prevlink"' );
 234                 # Make "next revision link"
 235                 if( $this->mNewRev->isCurrent() ) {
 236                         $nextlink = '&nbsp;';
 237                 } else {
 238                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 239                                 "diff=next&oldid={$this->mNewid}{$htmldiffarg}{$diffOnlyArg}", '', '', 'id="differences-nextlink"' );
 240                 }
 241
 242                 $oldminor = '';
 243                 $newminor = '';
 244
 245                 if( $this->mOldRev->isMinor() ) {
 246                         $oldminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 247                 }
 248                 if( $this->mNewRev->isMinor() ) {
 249                         $newminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 250                 }
 251
 252                 $rdel = ''; $ldel = '';
 253                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 254                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 255                                 // If revision was hidden from sysops
 256                                 $ldel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 257                         } else {
 258                                 $query = array( 'target' => $this->mOldRev->mTitle->getPrefixedDbkey(),
 259                                         'oldid' => $this->mOldRev->getId()
 260                                 );
 261                                 $ldel = $sk->revDeleteLink( $query, $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 262                         }
 263                         $ldel = "&nbsp;&nbsp;&nbsp;$ldel ";
 264                         // We don't currently handle well changing the top revision's settings
 265                         if( $this->mNewRev->isCurrent() ) {
 266                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 267                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 268                                 // If revision was hidden from sysops
 269                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 270                         } else {
 271                                 $query = array( 'target' =>  $this->mNewRev->mTitle->getPrefixedDbkey(),
 272                                         'oldid' => $this->mNewRev->getId()
 273                                 );
 274                                 $rdel = $sk->revDeleteLink( $query, $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 275                         }
 276                         $rdel = "&nbsp;&nbsp;&nbsp;$rdel ";
 277                 }
 278
 279                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 280                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, !$this->unhide ) . "</div>" .
 281                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, !$this->unhide ).$ldel."</div>" .
 282                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 283                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 284                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, !$this->unhide ) . " $rollback</div>" .
 285                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, !$this->unhide ).$rdel."</div>" .
 286                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 287
 288                 # Check if this user can see the revisions
 289                 $allowed = $this->mOldRev->userCan(Revision::DELETED_TEXT)
 290                         && $this->mNewRev->userCan(Revision::DELETED_TEXT);
 291                 $deleted = $this->mOldRev->isDeleted(Revision::DELETED_TEXT)
 292                         || $this->mNewRev->isDeleted(Revision::DELETED_TEXT);
 293                 # Output the diff if allowed...
 294                 if( $deleted && (!$this->unhide || !$allowed) ) {
 295                         $this->showDiffStyle();
 296                         $multi = $this->getMultiNotice();
 297                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 298                         if( !$allowed ) {
 299                                 # Give explanation for why revision is not visible
 300                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 301                                         array( 'rev-deleted-no-diff' ) );
 302                         } else {
 303                                 # Give explanation and add a link to view the diff...
 304                                 $link = $this->mTitle->getFullUrl( "diff={$this->mNewid}&oldid={$this->mOldid}".
 305                                         '&unhide=1&token='.urlencode( $wgUser->editToken($this->mNewid) ) );
 306                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 307                                         array( 'rev-deleted-unhide-diff', $link ) );
 308                         }
 309                 } else if( $wgEnableHtmlDiff && $this->htmldiff ) {
 310                         $multi = $this->getMultiNotice();
 311                         $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'wikicodecomparison' ),
 312                                 'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=0', '', '', 'id="differences-switchtype"' ).'</div>');
 313                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 314                         $this->renderHtmlDiff();
 315                 } else {
 316                         if( $wgEnableHtmlDiff ) {
 317                                 $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'visualcomparison' ),
 318                                         'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=1', '', '', 'id="differences-switchtype"' ).'</div>');
 319                         }
 320                         $this->showDiff( $oldHeader, $newHeader );
 321                         if( !$diffOnly ) {
 322                                 $this->renderNewRevision();
 323                         }
 324                 }
 325                 wfProfileOut( __METHOD__ );
 326         }
 327
 328         /**
 329          * Show the new revision of the page.
 330          */
 331         function renderNewRevision() {
 332                 global $wgOut, $wgUser;
 333                 wfProfileIn( __METHOD__ );
 334
 335                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 336                 # Add deleted rev tag if needed
 337                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 338                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 339                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 340                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 341                 }
 342
 343                 if( !$this->mNewRev->isCurrent() ) {
 344                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 345                 }
 346
 347                 $this->loadNewText();
 348                 if( is_object( $this->mNewRev ) ) {
 349                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 350                 }
 351
 352                 if( $this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage() ) {
 353                         // Stolen from Article::view --AG 2007-10-11
 354                         // Give hooks a chance to customise the output
 355                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 356                                 // Wrap the whole lot in a <pre> and don't parse
 357                                 $m = array();
 358                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 359                                 $wgOut->addHTML( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 360                                 $wgOut->addHTML( htmlspecialchars( $this->mNewtext ) );
 361                                 $wgOut->addHTML( "\n</pre>\n" );
 362                         }
 363                 } else {
 364                         $wgOut->addWikiTextTidy( $this->mNewtext );
 365                 }
 366
 367                 if( is_object( $this->mNewRev ) && !$this->mNewRev->isCurrent() ) {
 368                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 369                 }
 370                 # Add redundant patrol link on bottom...
 371                 if( $this->mRcidMarkPatrolled && $this->mTitle->quickUserCan('patrol') ) {
 372                         $sk = $wgUser->getSkin();
 373                         $wgOut->addHTML(
 374                                 "<div class='patrollink'>[" . $sk->makeKnownLinkObj( $this->mTitle,
 375                                         wfMsgHtml( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) .
 376                                 ']</div>'
 377                          );
 378                 }
 379
 380                 wfProfileOut( __METHOD__ );
 381         }
 382
 383
 384         function renderHtmlDiff() {
 385                 global $wgOut, $wgParser, $wgDebugComments;
 386                 wfProfileIn( __METHOD__ );
 387
 388                 $this->showDiffStyle();
 389
 390                 $wgOut->addHTML( '<h2>'.wfMsgHtml( 'visual-comparison' )."</h2>\n" );
 391                 #add deleted rev tag if needed
 392                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 393                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 394                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 395                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 396                 }
 397
 398                 if( !$this->mNewRev->isCurrent() ) {
 399                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 400                 }
 401
 402                 $this->loadText();
 403
 404                 // Old revision
 405                 if( is_object( $this->mOldRev ) ) {
 406                         $wgOut->setRevisionId( $this->mOldRev->getId() );
 407                 }
 408
 409                 $popts = $wgOut->parserOptions();
 410                 $oldTidy = $popts->setTidy( true );
 411                 $popts->setEditSection( false );
 412
 413                 $parserOutput = $wgParser->parse( $this->mOldtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 414                 $popts->setTidy( $oldTidy );
 415
 416                 //only for new?
 417                 //$wgOut->addParserOutputNoText( $parserOutput );
 418                 $oldHtml = $parserOutput->getText();
 419                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$oldHtml ) );
 420
 421                 // New revision
 422                 if( is_object( $this->mNewRev ) ) {
 423                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 424                 }
 425
 426                 $popts = $wgOut->parserOptions();
 427                 $oldTidy = $popts->setTidy( true );
 428
 429                 $parserOutput = $wgParser->parse( $this->mNewtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 430                 $popts->setTidy( $oldTidy );
 431
 432                 $wgOut->addParserOutputNoText( $parserOutput );
 433                 $newHtml = $parserOutput->getText();
 434                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$newHtml ) );
 435
 436                 unset($parserOutput, $popts);
 437
 438                 $differ = new HTMLDiffer(new DelegatingContentHandler($wgOut));
 439                 $differ->htmlDiff($oldHtml, $newHtml);
 440                 if ( $wgDebugComments ) {
 441                         $wgOut->addHTML( "\n<!-- HtmlDiff Debug Output:\n" . HTMLDiffer::getDebugOutput() . " End Debug -->" );
 442                 }
 443
 444                 wfProfileOut( __METHOD__ );
 445         }
 446
 447         /**
 448          * Show the first revision of an article. Uses normal diff headers in
 449          * contrast to normal "old revision" display style.
 450          */
 451         function showFirstRevision() {
 452                 global $wgOut, $wgUser;
 453                 wfProfileIn( __METHOD__ );
 454
 455                 # Get article text from the DB
 456                 #
 457                 if ( ! $this->loadNewText() ) {
 458                         $t = $this->mTitle->getPrefixedText();
 459                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 460                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 461                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 462                         wfProfileOut( __METHOD__ );
 463                         return;
 464                 }
 465                 if ( $this->mNewRev->isCurrent() ) {
 466                         $wgOut->setArticleFlag( true );
 467                 }
 468
 469                 # Check if user is allowed to look at this page. If not, bail out.
 470                 #
 471                 if ( !$this->mTitle->userCanRead() ) {
 472                         $wgOut->loginToUse();
 473                         $wgOut->output();
 474                         wfProfileOut( __METHOD__ );
 475                         throw new MWException("Permission Error: you do not have access to view this page");
 476                 }
 477
 478                 # Prepare the header box
 479                 #
 480                 $sk = $wgUser->getSkin();
 481
 482                 $next = $this->mTitle->getNextRevisionID( $this->mNewid );
 483                 if( !$next ) {
 484                         $nextlink = '';
 485                 } else {
 486                         $nextlink = '<br/>' . $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 487                                 'diff=next&oldid=' . $this->mNewid.$this->htmlDiffArgument(), '', '', 'id="differences-nextlink"' );
 488                 }
 489                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\">" .
 490                         $sk->revUserTools( $this->mNewRev ) . "<br/>" . $sk->revComment( $this->mNewRev ) . $nextlink . "</div>\n";
 491
 492                 $wgOut->addHTML( $header );
 493
 494                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 495                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 496
 497                 wfProfileOut( __METHOD__ );
 498         }
 499
 500         function htmlDiffArgument(){
 501                 global $wgEnableHtmlDiff;
 502                 if($wgEnableHtmlDiff){
 503                         if($this->htmldiff){
 504                                 return '&htmldiff=1';
 505                         }else{
 506                                 return '&htmldiff=0';
 507                         }
 508                 }else{
 509                         return '';
 510                 }
 511         }
 512
 513         /**
 514          * Get the diff text, send it to $wgOut
 515          * Returns false if the diff could not be generated, otherwise returns true
 516          */
 517         function showDiff( $otitle, $ntitle ) {
 518                 global $wgOut;
 519                 $diff = $this->getDiff( $otitle, $ntitle );
 520                 if ( $diff === false ) {
 521                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 522                         return false;
 523                 } else {
 524                         $this->showDiffStyle();
 525                         $wgOut->addHTML( $diff );
 526                         return true;
 527                 }
 528         }
 529
 530         /**
 531          * Add style sheets and supporting JS for diff display.
 532          */
 533         function showDiffStyle() {
 534                 global $wgStylePath, $wgStyleVersion, $wgOut;
 535                 $wgOut->addStyle( 'common/diff.css' );
 536
 537                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 538                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 539         }
 540
 541         /**
 542          * Get complete diff table, including header
 543          *
 544          * @param Title $otitle Old title
 545          * @param Title $ntitle New title
 546          * @return mixed
 547          */
 548         function getDiff( $otitle, $ntitle ) {
 549                 $body = $this->getDiffBody();
 550                 if ( $body === false ) {
 551                         return false;
 552                 } else {
 553                         $multi = $this->getMultiNotice();
 554                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 555                 }
 556         }
 557
 558         /**
 559          * Get the diff table body, without header
 560          *
 561          * @return mixed
 562          */
 563         function getDiffBody() {
 564                 global $wgMemc;
 565                 wfProfileIn( __METHOD__ );
 566                 $this->mCacheHit = true;
 567                 // Check if the diff should be hidden from this user
 568                 if ( !$this->loadRevisionData() )
 569                         return '';
 570                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 571                         return '';
 572                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 573                         return '';
 574                 } else if ( $this->mOldRev && $this->mNewRev && $this->mOldRev->getID() == $this->mNewRev->getID() ) {
 575                         return '';
 576                 }
 577                 // Cacheable?
 578                 $key = false;
 579                 if ( $this->mOldid && $this->mNewid ) {
 580                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 581                         // Try cache
 582                         if ( !$this->mRefreshCache ) {
 583                                 $difftext = $wgMemc->get( $key );
 584                                 if ( $difftext ) {
 585                                         wfIncrStats( 'diff_cache_hit' );
 586                                         $difftext = $this->localiseLineNumbers( $difftext );
 587                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 588                                         wfProfileOut( __METHOD__ );
 589                                         return $difftext;
 590                                 }
 591                         } // don't try to load but save the result
 592                 }
 593                 $this->mCacheHit = false;
 594
 595                 // Loadtext is permission safe, this just clears out the diff
 596                 if ( !$this->loadText() ) {
 597                         wfProfileOut( __METHOD__ );
 598                         return false;
 599                 }
 600
 601                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 602
 603                 // Save to cache for 7 days
 604                 if ( !wfRunHooks( 'AbortDiffCache', array( &$this ) ) ) {
 605                         wfIncrStats( 'diff_uncacheable' );
 606                 } else if ( $key !== false && $difftext !== false ) {
 607                         wfIncrStats( 'diff_cache_miss' );
 608                         $wgMemc->set( $key, $difftext, 7*86400 );
 609                 } else {
 610                         wfIncrStats( 'diff_uncacheable' );
 611                 }
 612                 // Replace line numbers with the text in the user's language
 613                 if ( $difftext !== false ) {
 614                         $difftext = $this->localiseLineNumbers( $difftext );
 615                 }
 616                 wfProfileOut( __METHOD__ );
 617                 return $difftext;
 618         }
 619
 620         /**
 621          * Make sure the proper modules are loaded before we try to
 622          * make the diff
 623          */
 624         private function initDiffEngines() {
 625                 global $wgExternalDiffEngine;
 626                 if ( $wgExternalDiffEngine == 'wikidiff' && !function_exists( 'wikidiff_do_diff' ) {
 627                         wfProfileIn( __METHOD__ . '-php_wikidiff.so' );
 628                         @dl( 'php_wikidiff.so' );
 629                         wfProfileOut( __METHOD__ . '-php_wikidiff.so' );
 630                 }
 631                 else if ( $wgExternalDiffEngine == 'wikidiff2' && !function_exists( 'wikidiff2_do_diff' ) ) {
 632                         wfProfileIn( __METHOD__ . '-php_wikidiff2.so' );
 633                         @dl( 'php_wikidiff2.so' );
 634                         wfProfileOut( __METHOD__ . '-php_wikidiff2.so' );
 635                 }
 636         }
 637
 638         /**
 639          * Generate a diff, no caching
 640          * $otext and $ntext must be already segmented
 641          */
 642         function generateDiffBody( $otext, $ntext ) {
 643                 global $wgExternalDiffEngine, $wgContLang;
 644
 645                 $otext = str_replace( "\r\n", "\n", $otext );
 646                 $ntext = str_replace( "\r\n", "\n", $ntext );
 647
 648                 $this->initDiffEngines();
 649
 650                 if ( $wgExternalDiffEngine == 'wikidiff' && function_exists( 'wikidiff_do_diff' ) ) {
 651                         # For historical reasons, external diff engine expects
 652                         # input text to be HTML-escaped already
 653                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 654                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 655                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 656                         $this->debug( 'wikidiff1' );
 657                 }
 658
 659                 if ( $wgExternalDiffEngine == 'wikidiff2' && function_exists( 'wikidiff2_do_diff' ) ) {
 660                         # Better external diff engine, the 2 may some day be dropped
 661                         # This one does the escaping and segmenting itself
 662                         wfProfileIn( 'wikidiff2_do_diff' );
 663                         $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 664                         $text .= $this->debug( 'wikidiff2' );
 665                         wfProfileOut( 'wikidiff2_do_diff' );
 666                         return $text;
 667                 }
 668                 if ( $wgExternalDiffEngine != 'wikidiff3' && $wgExternalDiffEngine !== false ) {
 669                         # Diff via the shell
 670                         global $wgTmpDirectory;
 671                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 672                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 673
 674                         $tempFile1 = fopen( $tempName1, "w" );
 675                         if ( !$tempFile1 ) {
 676                                 wfProfileOut( __METHOD__ );
 677                                 return false;
 678                         }
 679                         $tempFile2 = fopen( $tempName2, "w" );
 680                         if ( !$tempFile2 ) {
 681                                 wfProfileOut( __METHOD__ );
 682                                 return false;
 683                         }
 684                         fwrite( $tempFile1, $otext );
 685                         fwrite( $tempFile2, $ntext );
 686                         fclose( $tempFile1 );
 687                         fclose( $tempFile2 );
 688                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 689                         wfProfileIn( __METHOD__ . "-shellexec" );
 690                         $difftext = wfShellExec( $cmd );
 691                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 692                         wfProfileOut( __METHOD__ . "-shellexec" );
 693                         unlink( $tempName1 );
 694                         unlink( $tempName2 );
 695                         return $difftext;
 696                 }
 697
 698                 # Native PHP diff
 699                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 700                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 701                 $diffs = new Diff( $ota, $nta );
 702                 $formatter = new TableDiffFormatter();
 703                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 704                 $this->debug();
 705         }
 706
 707         /**
 708          * Generate a debug comment indicating diff generating time,
 709          * server node, and generator backend.
 710          */
 711         protected function debug( $generator="internal" ) {
 712                 global $wgShowHostnames;
 713                 $data = array( $generator );
 714                 if( $wgShowHostnames ) {
 715                         $data[] = wfHostname();
 716                 }
 717                 $data[] = wfTimestamp( TS_DB );
 718                 return "<!-- diff generator: " .
 719                 implode( " ",
 720                 array_map(
 721                                         "htmlspecialchars",
 722                 $data ) ) .
 723                         " -->\n";
 724         }
 725
 726         /**
 727          * Replace line numbers with the text in the user's language
 728          */
 729         function localiseLineNumbers( $text ) {
 730                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 731                 array( &$this, 'localiseLineNumbersCb' ), $text );
 732         }
 733
 734         function localiseLineNumbersCb( $matches ) {
 735                 global $wgLang;
 736                 return wfMsgExt( 'lineno', array (), $wgLang->formatNum( $matches[1] ) );
 737         }
 738
 739
 740         /**
 741          * If there are revisions between the ones being compared, return a note saying so.
 742          */
 743         function getMultiNotice() {
 744                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 745                 return '';
 746
 747                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 748                         // Comparing two different pages? Count would be meaningless.
 749                         return '';
 750                 }
 751
 752                 $oldid = $this->mOldRev->getId();
 753                 $newid = $this->mNewRev->getId();
 754                 if ( $oldid > $newid ) {
 755                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 756                 }
 757
 758                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 759                 if ( !$n )
 760                 return '';
 761
 762                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 763         }
 764
 765
 766         /**
 767          * Add the header to a diff body
 768          */
 769         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 770                 $colspan = 1;
 771                 $header = "<table class='diff'>";
 772                 if( $diff ) { // Safari/Chrome show broken output if cols not used
 773                         $header .= "
 774                         <col class='diff-marker' />
 775                         <col class='diff-content' />
 776                         <col class='diff-marker' />
 777                         <col class='diff-content' />";
 778                         $colspan = 2;
 779                 }
 780                 $header .= "
 781                 <tr valign='top'>
 782                 <td colspan='$colspan' class='diff-otitle'>{$otitle}</td>
 783                 <td colspan='$colspan' class='diff-ntitle'>{$ntitle}</td>
 784                 </tr>";
 785
 786                 if ( $multi != '' )
 787                 $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 788
 789                 return $header . $diff . "</table>";
 790         }
 791
 792         /**
 793          * Use specified text instead of loading from the database
 794          */
 795         function setText( $oldText, $newText ) {
 796                 $this->mOldtext = $oldText;
 797                 $this->mNewtext = $newText;
 798                 $this->mTextLoaded = 2;
 799                 $this->mRevisionsLoaded = true;
 800         }
 801
 802         /**
 803          * Load revision metadata for the specified articles. If newid is 0, then compare
 804          * the old article in oldid to the current article; if oldid is 0, then
 805          * compare the current article to the immediately previous one (ignoring the
 806          * value of newid).
 807          *
 808          * If oldid is false, leave the corresponding revision object set
 809          * to false. This is impossible via ordinary user input, and is provided for
 810          * API convenience.
 811          */
 812         function loadRevisionData() {
 813                 global $wgLang, $wgUser;
 814                 if ( $this->mRevisionsLoaded ) {
 815                         return true;
 816                 } else {
 817                         // Whether it succeeds or fails, we don't want to try again
 818                         $this->mRevisionsLoaded = true;
 819                 }
 820
 821                 // Load the new revision object
 822                 $this->mNewRev = $this->mNewid
 823                         ? Revision::newFromId( $this->mNewid )
 824                         : Revision::newFromTitle( $this->mTitle );
 825                 if( !$this->mNewRev instanceof Revision )
 826                         return false;
 827
 828                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 829                 $this->mNewid = $this->mNewRev->getId();
 830
 831                 // Check if page is editable
 832                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 833
 834                 // Set assorted variables
 835                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 836                 $this->mNewPage = $this->mNewRev->getTitle();
 837                 if( $this->mNewRev->isCurrent() ) {
 838                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 839                         $this->mPagetitle = wfMsgHTML( 'currentrev-asof', $timestamp );
 840                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 841
 842                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 843                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 844                 } else {
 845                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 846                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 847                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 848
 849                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 850                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 851                 }
 852                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 853                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 854                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 855                         $this->mNewtitle = "<span class='history-deleted'>{$this->mNewtitle}</span>";
 856                 }
 857
 858                 // Load the old revision object
 859                 $this->mOldRev = false;
 860                 if( $this->mOldid ) {
 861                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 862                 } elseif ( $this->mOldid === 0 ) {
 863                         $rev = $this->mNewRev->getPrevious();
 864                         if( $rev ) {
 865                                 $this->mOldid = $rev->getId();
 866                                 $this->mOldRev = $rev;
 867                         } else {
 868                                 // No previous revision; mark to show as first-version only.
 869                                 $this->mOldid = false;
 870                                 $this->mOldRev = false;
 871                         }
 872                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 873
 874                 if( is_null( $this->mOldRev ) ) {
 875                         return false;
 876                 }
 877
 878                 if ( $this->mOldRev ) {
 879                         $this->mOldPage = $this->mOldRev->getTitle();
 880
 881                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 882                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 883                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 884                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 885
 886                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 887                         . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 888                         // Add an "undo" link
 889                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 890                         $htmlLink = htmlspecialchars( wfMsg( 'editundo' ) );
 891                         $htmlTitle = $wgUser->getSkin()->tooltip( 'undo' );
 892                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 893                                 $this->mNewtitle .= " (<a href='$newUndo' $htmlTitle>" . $htmlLink . "</a>)";
 894                         }
 895
 896                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 897                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 898                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 899                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 900                         }
 901                 }
 902
 903                 return true;
 904         }
 905
 906         /**
 907          * Load the text of the revisions, as well as revision data.
 908          */
 909         function loadText() {
 910                 if ( $this->mTextLoaded == 2 ) {
 911                         return true;
 912                 } else {
 913                         // Whether it succeeds or fails, we don't want to try again
 914                         $this->mTextLoaded = 2;
 915                 }
 916
 917                 if ( !$this->loadRevisionData() ) {
 918                         return false;
 919                 }
 920                 if ( $this->mOldRev ) {
 921                         $this->mOldtext = $this->mOldRev->getText( Revision::FOR_THIS_USER );
 922                         if ( $this->mOldtext === false ) {
 923                                 return false;
 924                         }
 925                 }
 926                 if ( $this->mNewRev ) {
 927                         $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 928                         if ( $this->mNewtext === false ) {
 929                                 return false;
 930                         }
 931                 }
 932                 return true;
 933         }
 934
 935         /**
 936          * Load the text of the new revision, not the old one
 937          */
 938         function loadNewText() {
 939                 if ( $this->mTextLoaded >= 1 ) {
 940                         return true;
 941                 } else {
 942                         $this->mTextLoaded = 1;
 943                 }
 944                 if ( !$this->loadRevisionData() ) {
 945                         return false;
 946                 }
 947                 $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 948                 return true;
 949         }
 950
 951
 952 }
 953
 954 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 955 //
 956 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 957 // You may copy this code freely under the conditions of the GPL.
 958 //
 959
 960 define('USE_ASSERTS', function_exists('assert'));
 961
 962 /**
 963  * @todo document
 964  * @private
 965  * @ingroup DifferenceEngine
 966  */
 967 class _DiffOp {
 968         var $type;
 969         var $orig;
 970         var $closing;
 971
 972         function reverse() {
 973                 trigger_error('pure virtual', E_USER_ERROR);
 974         }
 975
 976         function norig() {
 977                 return $this->orig ? sizeof($this->orig) : 0;
 978         }
 979
 980         function nclosing() {
 981                 return $this->closing ? sizeof($this->closing) : 0;
 982         }
 983 }
 984
 985 /**
 986  * @todo document
 987  * @private
 988  * @ingroup DifferenceEngine
 989  */
 990 class _DiffOp_Copy extends _DiffOp {
 991         var $type = 'copy';
 992
 993         function _DiffOp_Copy ($orig, $closing = false) {
 994                 if (!is_array($closing))
 995                 $closing = $orig;
 996                 $this->orig = $orig;
 997                 $this->closing = $closing;
 998         }
 999
1000         function reverse() {
1001                 return new _DiffOp_Copy($this->closing, $this->orig);
1002         }
1003 }
1004
1005 /**
1006  * @todo document
1007  * @private
1008  * @ingroup DifferenceEngine
1009  */
1010 class _DiffOp_Delete extends _DiffOp {
1011         var $type = 'delete';
1012
1013         function _DiffOp_Delete ($lines) {
1014                 $this->orig = $lines;
1015                 $this->closing = false;
1016         }
1017
1018         function reverse() {
1019                 return new _DiffOp_Add($this->orig);
1020         }
1021 }
1022
1023 /**
1024  * @todo document
1025  * @private
1026  * @ingroup DifferenceEngine
1027  */
1028 class _DiffOp_Add extends _DiffOp {
1029         var $type = 'add';
1030
1031         function _DiffOp_Add ($lines) {
1032                 $this->closing = $lines;
1033                 $this->orig = false;
1034         }
1035
1036         function reverse() {
1037                 return new _DiffOp_Delete($this->closing);
1038         }
1039 }
1040
1041 /**
1042  * @todo document
1043  * @private
1044  * @ingroup DifferenceEngine
1045  */
1046 class _DiffOp_Change extends _DiffOp {
1047         var $type = 'change';
1048
1049         function _DiffOp_Change ($orig, $closing) {
1050                 $this->orig = $orig;
1051                 $this->closing = $closing;
1052         }
1053
1054         function reverse() {
1055                 return new _DiffOp_Change($this->closing, $this->orig);
1056         }
1057 }
1058
1059 /**
1060  * Class used internally by Diff to actually compute the diffs.
1061  *
1062  * The algorithm used here is mostly lifted from the perl module
1063  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
1064  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
1065  *
1066  * More ideas are taken from:
1067  *       http://www.ics.uci.edu/~eppstein/161/960229.html
1068  *
1069  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
1070  * diffutils-2.7, which can be found at:
1071  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
1072  *
1073  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
1074  * are my own.
1075  *
1076  * Line length limits for robustness added by Tim Starling, 2005-08-31
1077  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
1078  *
1079  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
1080  * @private
1081  * @ingroup DifferenceEngine
1082  */
1083 class _DiffEngine {
1084
1085         const MAX_XREF_LENGTH =  10000;
1086
1087         function diff ($from_lines, $to_lines){
1088                 wfProfileIn( __METHOD__ );
1089
1090                 // Diff and store locally
1091                 $this->diff_local($from_lines, $to_lines);
1092
1093                 // Merge edits when possible
1094                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
1095                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
1096
1097                 // Compute the edit operations.
1098                 $n_from = sizeof($from_lines);
1099                 $n_to = sizeof($to_lines);
1100
1101                 $edits = array();
1102                 $xi = $yi = 0;
1103                 while ($xi < $n_from || $yi < $n_to) {
1104                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
1105                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
1106
1107                         // Skip matching "snake".
1108                         $copy = array();
1109                         while ( $xi < $n_from && $yi < $n_to
1110                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
1111                                 $copy[] = $from_lines[$xi++];
1112                                 ++$yi;
1113                         }
1114                         if ($copy)
1115                         $edits[] = new _DiffOp_Copy($copy);
1116
1117                         // Find deletes & adds.
1118                         $delete = array();
1119                         while ($xi < $n_from && $this->xchanged[$xi])
1120                         $delete[] = $from_lines[$xi++];
1121
1122                         $add = array();
1123                         while ($yi < $n_to && $this->ychanged[$yi])
1124                         $add[] = $to_lines[$yi++];
1125
1126                         if ($delete && $add)
1127                         $edits[] = new _DiffOp_Change($delete, $add);
1128                         elseif ($delete)
1129                         $edits[] = new _DiffOp_Delete($delete);
1130                         elseif ($add)
1131                         $edits[] = new _DiffOp_Add($add);
1132                 }
1133                 wfProfileOut( __METHOD__ );
1134                 return $edits;
1135         }
1136
1137         function diff_local ($from_lines, $to_lines) {
1138                 global $wgExternalDiffEngine;
1139                 wfProfileIn( __METHOD__);
1140
1141                 if($wgExternalDiffEngine == 'wikidiff3'){
1142                         // wikidiff3
1143                         $wikidiff3 = new WikiDiff3();
1144                         $wikidiff3->diff($from_lines, $to_lines);
1145                         $this->xchanged = $wikidiff3->removed;
1146                         $this->ychanged = $wikidiff3->added;
1147                         unset($wikidiff3);
1148                 }else{
1149                         // old diff
1150                         $n_from = sizeof($from_lines);
1151                         $n_to = sizeof($to_lines);
1152                         $this->xchanged = $this->ychanged = array();
1153                         $this->xv = $this->yv = array();
1154                         $this->xind = $this->yind = array();
1155                         unset($this->seq);
1156                         unset($this->in_seq);
1157                         unset($this->lcs);
1158
1159                         // Skip leading common lines.
1160                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
1161                                 if ($from_lines[$skip] !== $to_lines[$skip])
1162                                 break;
1163                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
1164                         }
1165                         // Skip trailing common lines.
1166                         $xi = $n_from; $yi = $n_to;
1167                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
1168                                 if ($from_lines[$xi] !== $to_lines[$yi])
1169                                 break;
1170                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
1171                         }
1172
1173                         // Ignore lines which do not exist in both files.
1174                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1175                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
1176                         }
1177
1178                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
1179                                 $line = $to_lines[$yi];
1180                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
1181                                 continue;
1182                                 $yhash[$this->_line_hash($line)] = 1;
1183                                 $this->yv[] = $line;
1184                                 $this->yind[] = $yi;
1185                         }
1186                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1187                                 $line = $from_lines[$xi];
1188                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
1189                                 continue;
1190                                 $this->xv[] = $line;
1191                                 $this->xind[] = $xi;
1192                         }
1193
1194                         // Find the LCS.
1195                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
1196                 }
1197                 wfProfileOut( __METHOD__ );
1198         }
1199
1200         /**
1201          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1202          */
1203         function _line_hash( $line ) {
1204                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1205                         return md5( $line );
1206                 } else {
1207                         return $line;
1208                 }
1209         }
1210
1211         /* Divide the Largest Common Subsequence (LCS) of the sequences
1212          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1213          * sized segments.
1214          *
1215          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1216          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1217          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1218          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1219          * that (X0, Y0) == (XOFF, YOFF) and
1220          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1221          *
1222          * This function assumes that the first lines of the specified portions
1223          * of the two files do not match, and likewise that the last lines do not
1224          * match.  The caller must trim matching lines from the beginning and end
1225          * of the portions it is going to specify.
1226          */
1227         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1228                 $flip = false;
1229
1230                 if ($xlim - $xoff > $ylim - $yoff) {
1231                         // Things seems faster (I'm not sure I understand why)
1232                         // when the shortest sequence in X.
1233                         $flip = true;
1234                         list ($xoff, $xlim, $yoff, $ylim)
1235                         = array( $yoff, $ylim, $xoff, $xlim);
1236                 }
1237
1238                 if ($flip)
1239                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1240                 $ymatches[$this->xv[$i]][] = $i;
1241                 else
1242                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1243                 $ymatches[$this->yv[$i]][] = $i;
1244
1245                 $this->lcs = 0;
1246                 $this->seq[0]= $yoff - 1;
1247                 $this->in_seq = array();
1248                 $ymids[0] = array();
1249
1250                 $numer = $xlim - $xoff + $nchunks - 1;
1251                 $x = $xoff;
1252                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1253                         if ($chunk > 0)
1254                         for ($i = 0; $i <= $this->lcs; $i++)
1255                         $ymids[$i][$chunk-1] = $this->seq[$i];
1256
1257                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1258                         for ( ; $x < $x1; $x++) {
1259                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1260                                 if (empty($ymatches[$line]))
1261                                 continue;
1262                                 $matches = $ymatches[$line];
1263                                 reset($matches);
1264                                 while (list ($junk, $y) = each($matches))
1265                                 if (empty($this->in_seq[$y])) {
1266                                         $k = $this->_lcs_pos($y);
1267                                         USE_ASSERTS && assert($k > 0);
1268                                         $ymids[$k] = $ymids[$k-1];
1269                                         break;
1270                                 }
1271                                 while (list ( /* $junk */, $y) = each($matches)) {
1272                                         if ($y > $this->seq[$k-1]) {
1273                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1274                                                 // Optimization: this is a common case:
1275                                                 //      next match is just replacing previous match.
1276                                                 $this->in_seq[$this->seq[$k]] = false;
1277                                                 $this->seq[$k] = $y;
1278                                                 $this->in_seq[$y] = 1;
1279                                         } else if (empty($this->in_seq[$y])) {
1280                                                 $k = $this->_lcs_pos($y);
1281                                                 USE_ASSERTS && assert($k > 0);
1282                                                 $ymids[$k] = $ymids[$k-1];
1283                                         }
1284                                 }
1285                         }
1286                 }
1287
1288                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1289                 $ymid = $ymids[$this->lcs];
1290                 for ($n = 0; $n < $nchunks - 1; $n++) {
1291                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1292                         $y1 = $ymid[$n] + 1;
1293                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1294                 }
1295                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1296
1297                 return array($this->lcs, $seps);
1298         }
1299
1300         function _lcs_pos ($ypos) {
1301                 $end = $this->lcs;
1302                 if ($end == 0 || $ypos > $this->seq[$end]) {
1303                         $this->seq[++$this->lcs] = $ypos;
1304                         $this->in_seq[$ypos] = 1;
1305                         return $this->lcs;
1306                 }
1307
1308                 $beg = 1;
1309                 while ($beg < $end) {
1310                         $mid = (int)(($beg + $end) / 2);
1311                         if ( $ypos > $this->seq[$mid] )
1312                         $beg = $mid + 1;
1313                         else
1314                         $end = $mid;
1315                 }
1316
1317                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1318
1319                 $this->in_seq[$this->seq[$end]] = false;
1320                 $this->seq[$end] = $ypos;
1321                 $this->in_seq[$ypos] = 1;
1322                 return $end;
1323         }
1324
1325         /* Find LCS of two sequences.
1326          *
1327          * The results are recorded in the vectors $this->{x,y}changed[], by
1328          * storing a 1 in the element for each line that is an insertion
1329          * or deletion (ie. is not in the LCS).
1330          *
1331          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1332          *
1333          * Note that XLIM, YLIM are exclusive bounds.
1334          * All line numbers are origin-0 and discarded lines are not counted.
1335          */
1336         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1337                 // Slide down the bottom initial diagonal.
1338                 while ($xoff < $xlim && $yoff < $ylim
1339                 && $this->xv[$xoff] == $this->yv[$yoff]) {
1340                         ++$xoff;
1341                         ++$yoff;
1342                 }
1343
1344                 // Slide up the top initial diagonal.
1345                 while ($xlim > $xoff && $ylim > $yoff
1346                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1347                         --$xlim;
1348                         --$ylim;
1349                 }
1350
1351                 if ($xoff == $xlim || $yoff == $ylim)
1352                 $lcs = 0;
1353                 else {
1354                         // This is ad hoc but seems to work well.
1355                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1356                         //$nchunks = max(2,min(8,(int)$nchunks));
1357                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1358                         list ($lcs, $seps)
1359                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1360                 }
1361
1362                 if ($lcs == 0) {
1363                         // X and Y sequences have no common subsequence:
1364                         // mark all changed.
1365                         while ($yoff < $ylim)
1366                         $this->ychanged[$this->yind[$yoff++]] = 1;
1367                         while ($xoff < $xlim)
1368                         $this->xchanged[$this->xind[$xoff++]] = 1;
1369                 } else {
1370                         // Use the partitions to split this problem into subproblems.
1371                         reset($seps);
1372                         $pt1 = $seps[0];
1373                         while ($pt2 = next($seps)) {
1374                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1375                                 $pt1 = $pt2;
1376                         }
1377                 }
1378         }
1379
1380         /* Adjust inserts/deletes of identical lines to join changes
1381          * as much as possible.
1382          *
1383          * We do something when a run of changed lines include a
1384          * line at one end and has an excluded, identical line at the other.
1385          * We are free to choose which identical line is included.
1386          * `compareseq' usually chooses the one at the beginning,
1387          * but usually it is cleaner to consider the following identical line
1388          * to be the "change".
1389          *
1390          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1391          */
1392         function _shift_boundaries ($lines, &$changed, $other_changed) {
1393                 wfProfileIn( __METHOD__ );
1394                 $i = 0;
1395                 $j = 0;
1396
1397                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1398                 $len = sizeof($lines);
1399                 $other_len = sizeof($other_changed);
1400
1401                 while (1) {
1402                         /*
1403                          * Scan forwards to find beginning of another run of changes.
1404                          * Also keep track of the corresponding point in the other file.
1405                          *
1406                          * Throughout this code, $i and $j are adjusted together so that
1407                          * the first $i elements of $changed and the first $j elements
1408                          * of $other_changed both contain the same number of zeros
1409                          * (unchanged lines).
1410                          * Furthermore, $j is always kept so that $j == $other_len or
1411                          * $other_changed[$j] == false.
1412                          */
1413                         while ($j < $other_len && $other_changed[$j])
1414                         $j++;
1415
1416                         while ($i < $len && ! $changed[$i]) {
1417                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1418                                 $i++; $j++;
1419                                 while ($j < $other_len && $other_changed[$j])
1420                                 $j++;
1421                         }
1422
1423                         if ($i == $len)
1424                         break;
1425
1426                         $start = $i;
1427
1428                         // Find the end of this run of changes.
1429                         while (++$i < $len && $changed[$i])
1430                         continue;
1431
1432                         do {
1433                                 /*
1434                                  * Record the length of this run of changes, so that
1435                                  * we can later determine whether the run has grown.
1436                                  */
1437                                 $runlength = $i - $start;
1438
1439                                 /*
1440                                  * Move the changed region back, so long as the
1441                                  * previous unchanged line matches the last changed one.
1442                                  * This merges with previous changed regions.
1443                                  */
1444                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1445                                         $changed[--$start] = 1;
1446                                         $changed[--$i] = false;
1447                                         while ($start > 0 && $changed[$start - 1])
1448                                         $start--;
1449                                         USE_ASSERTS && assert('$j > 0');
1450                                         while ($other_changed[--$j])
1451                                         continue;
1452                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1453                                 }
1454
1455                                 /*
1456                                  * Set CORRESPONDING to the end of the changed run, at the last
1457                                  * point where it corresponds to a changed run in the other file.
1458                                  * CORRESPONDING == LEN means no such point has been found.
1459                                  */
1460                                 $corresponding = $j < $other_len ? $i : $len;
1461
1462                                 /*
1463                                  * Move the changed region forward, so long as the
1464                                  * first changed line matches the following unchanged one.
1465                                  * This merges with following changed regions.
1466                                  * Do this second, so that if there are no merges,
1467                                  * the changed region is moved forward as far as possible.
1468                                  */
1469                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1470                                         $changed[$start++] = false;
1471                                         $changed[$i++] = 1;
1472                                         while ($i < $len && $changed[$i])
1473                                         $i++;
1474
1475                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1476                                         $j++;
1477                                         if ($j < $other_len && $other_changed[$j]) {
1478                                                 $corresponding = $i;
1479                                                 while ($j < $other_len && $other_changed[$j])
1480                                                 $j++;
1481                                         }
1482                                 }
1483                         } while ($runlength != $i - $start);
1484
1485                         /*
1486                          * If possible, move the fully-merged run of changes
1487                          * back to a corresponding run in the other file.
1488                          */
1489                         while ($corresponding < $i) {
1490                                 $changed[--$start] = 1;
1491                                 $changed[--$i] = 0;
1492                                 USE_ASSERTS && assert('$j > 0');
1493                                 while ($other_changed[--$j])
1494                                 continue;
1495                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1496                         }
1497                 }
1498                 wfProfileOut( __METHOD__ );
1499         }
1500 }
1501
1502 /**
1503  * Class representing a 'diff' between two sequences of strings.
1504  * @todo document
1505  * @private
1506  * @ingroup DifferenceEngine
1507  */
1508 class Diff
1509 {
1510         var $edits;
1511
1512         /**
1513          * Constructor.
1514          * Computes diff between sequences of strings.
1515          *
1516          * @param $from_lines array An array of strings.
1517          *                (Typically these are lines from a file.)
1518          * @param $to_lines array An array of strings.
1519          */
1520         function Diff($from_lines, $to_lines) {
1521                 $eng = new _DiffEngine;
1522                 $this->edits = $eng->diff($from_lines, $to_lines);
1523                 //$this->_check($from_lines, $to_lines);
1524         }
1525
1526         /**
1527          * Compute reversed Diff.
1528          *
1529          * SYNOPSIS:
1530          *
1531          *      $diff = new Diff($lines1, $lines2);
1532          *      $rev = $diff->reverse();
1533          * @return object A Diff object representing the inverse of the
1534          *                                original diff.
1535          */
1536         function reverse () {
1537                 $rev = $this;
1538                 $rev->edits = array();
1539                 foreach ($this->edits as $edit) {
1540                         $rev->edits[] = $edit->reverse();
1541                 }
1542                 return $rev;
1543         }
1544
1545         /**
1546          * Check for empty diff.
1547          *
1548          * @return bool True iff two sequences were identical.
1549          */
1550         function isEmpty () {
1551                 foreach ($this->edits as $edit) {
1552                         if ($edit->type != 'copy')
1553                         return false;
1554                 }
1555                 return true;
1556         }
1557
1558         /**
1559          * Compute the length of the Longest Common Subsequence (LCS).
1560          *
1561          * This is mostly for diagnostic purposed.
1562          *
1563          * @return int The length of the LCS.
1564          */
1565         function lcs () {
1566                 $lcs = 0;
1567                 foreach ($this->edits as $edit) {
1568                         if ($edit->type == 'copy')
1569                         $lcs += sizeof($edit->orig);
1570                 }
1571                 return $lcs;
1572         }
1573
1574         /**
1575          * Get the original set of lines.
1576          *
1577          * This reconstructs the $from_lines parameter passed to the
1578          * constructor.
1579          *
1580          * @return array The original sequence of strings.
1581          */
1582         function orig() {
1583                 $lines = array();
1584
1585                 foreach ($this->edits as $edit) {
1586                         if ($edit->orig)
1587                         array_splice($lines, sizeof($lines), 0, $edit->orig);
1588                 }
1589                 return $lines;
1590         }
1591
1592         /**
1593          * Get the closing set of lines.
1594          *
1595          * This reconstructs the $to_lines parameter passed to the
1596          * constructor.
1597          *
1598          * @return array The sequence of strings.
1599          */
1600         function closing() {
1601                 $lines = array();
1602
1603                 foreach ($this->edits as $edit) {
1604                         if ($edit->closing)
1605                         array_splice($lines, sizeof($lines), 0, $edit->closing);
1606                 }
1607                 return $lines;
1608         }
1609
1610         /**
1611          * Check a Diff for validity.
1612          *
1613          * This is here only for debugging purposes.
1614          */
1615         function _check ($from_lines, $to_lines) {
1616                 wfProfileIn( __METHOD__ );
1617                 if (serialize($from_lines) != serialize($this->orig()))
1618                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1619                 if (serialize($to_lines) != serialize($this->closing()))
1620                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1621
1622                 $rev = $this->reverse();
1623                 if (serialize($to_lines) != serialize($rev->orig()))
1624                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
1625                 if (serialize($from_lines) != serialize($rev->closing()))
1626                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1627
1628
1629                 $prevtype = 'none';
1630                 foreach ($this->edits as $edit) {
1631                         if ( $prevtype == $edit->type )
1632                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1633                         $prevtype = $edit->type;
1634                 }
1635
1636                 $lcs = $this->lcs();
1637                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1638                 wfProfileOut( __METHOD__ );
1639         }
1640 }
1641
1642 /**
1643  * @todo document, bad name.
1644  * @private
1645  * @ingroup DifferenceEngine
1646  */
1647 class MappedDiff extends Diff
1648 {
1649         /**
1650          * Constructor.
1651          *
1652          * Computes diff between sequences of strings.
1653          *
1654          * This can be used to compute things like
1655          * case-insensitve diffs, or diffs which ignore
1656          * changes in white-space.
1657          *
1658          * @param $from_lines array An array of strings.
1659          *      (Typically these are lines from a file.)
1660          *
1661          * @param $to_lines array An array of strings.
1662          *
1663          * @param $mapped_from_lines array This array should
1664          *      have the same size number of elements as $from_lines.
1665          *      The elements in $mapped_from_lines and
1666          *      $mapped_to_lines are what is actually compared
1667          *      when computing the diff.
1668          *
1669          * @param $mapped_to_lines array This array should
1670          *      have the same number of elements as $to_lines.
1671          */
1672         function MappedDiff($from_lines, $to_lines,
1673         $mapped_from_lines, $mapped_to_lines) {
1674                 wfProfileIn( __METHOD__ );
1675
1676                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1677                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1678
1679                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1680
1681                 $xi = $yi = 0;
1682                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1683                         $orig = &$this->edits[$i]->orig;
1684                         if (is_array($orig)) {
1685                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1686                                 $xi += sizeof($orig);
1687                         }
1688
1689                         $closing = &$this->edits[$i]->closing;
1690                         if (is_array($closing)) {
1691                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1692                                 $yi += sizeof($closing);
1693                         }
1694                 }
1695                 wfProfileOut( __METHOD__ );
1696         }
1697 }
1698
1699 /**
1700  * A class to format Diffs
1701  *
1702  * This class formats the diff in classic diff format.
1703  * It is intended that this class be customized via inheritance,
1704  * to obtain fancier outputs.
1705  * @todo document
1706  * @private
1707  * @ingroup DifferenceEngine
1708  */
1709 class DiffFormatter {
1710         /**
1711          * Number of leading context "lines" to preserve.
1712          *
1713          * This should be left at zero for this class, but subclasses
1714          * may want to set this to other values.
1715          */
1716         var $leading_context_lines = 0;
1717
1718         /**
1719          * Number of trailing context "lines" to preserve.
1720          *
1721          * This should be left at zero for this class, but subclasses
1722          * may want to set this to other values.
1723          */
1724         var $trailing_context_lines = 0;
1725
1726         /**
1727          * Format a diff.
1728          *
1729          * @param $diff object A Diff object.
1730          * @return string The formatted output.
1731          */
1732         function format($diff) {
1733                 wfProfileIn( __METHOD__ );
1734
1735                 $xi = $yi = 1;
1736                 $block = false;
1737                 $context = array();
1738
1739                 $nlead = $this->leading_context_lines;
1740                 $ntrail = $this->trailing_context_lines;
1741
1742                 $this->_start_diff();
1743
1744                 foreach ($diff->edits as $edit) {
1745                         if ($edit->type == 'copy') {
1746                                 if (is_array($block)) {
1747                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1748                                                 $block[] = $edit;
1749                                         }
1750                                         else{
1751                                                 if ($ntrail) {
1752                                                         $context = array_slice($edit->orig, 0, $ntrail);
1753                                                         $block[] = new _DiffOp_Copy($context);
1754                                                 }
1755                                                 $this->_block($x0, $ntrail + $xi - $x0,
1756                                                 $y0, $ntrail + $yi - $y0,
1757                                                 $block);
1758                                                 $block = false;
1759                                         }
1760                                 }
1761                                 $context = $edit->orig;
1762                         }
1763                         else {
1764                                 if (! is_array($block)) {
1765                                         $context = array_slice($context, sizeof($context) - $nlead);
1766                                         $x0 = $xi - sizeof($context);
1767                                         $y0 = $yi - sizeof($context);
1768                                         $block = array();
1769                                         if ($context)
1770                                         $block[] = new _DiffOp_Copy($context);
1771                                 }
1772                                 $block[] = $edit;
1773                         }
1774
1775                         if ($edit->orig)
1776                         $xi += sizeof($edit->orig);
1777                         if ($edit->closing)
1778                         $yi += sizeof($edit->closing);
1779                 }
1780
1781                 if (is_array($block))
1782                 $this->_block($x0, $xi - $x0,
1783                 $y0, $yi - $y0,
1784                 $block);
1785
1786                 $end = $this->_end_diff();
1787                 wfProfileOut( __METHOD__ );
1788                 return $end;
1789         }
1790
1791         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1792                 wfProfileIn( __METHOD__ );
1793                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1794                 foreach ($edits as $edit) {
1795                         if ($edit->type == 'copy')
1796                         $this->_context($edit->orig);
1797                         elseif ($edit->type == 'add')
1798                         $this->_added($edit->closing);
1799                         elseif ($edit->type == 'delete')
1800                         $this->_deleted($edit->orig);
1801                         elseif ($edit->type == 'change')
1802                         $this->_changed($edit->orig, $edit->closing);
1803                         else
1804                         trigger_error('Unknown edit type', E_USER_ERROR);
1805                 }
1806                 $this->_end_block();
1807                 wfProfileOut( __METHOD__ );
1808         }
1809
1810         function _start_diff() {
1811                 ob_start();
1812         }
1813
1814         function _end_diff() {
1815                 $val = ob_get_contents();
1816                 ob_end_clean();
1817                 return $val;
1818         }
1819
1820         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1821                 if ($xlen > 1)
1822                 $xbeg .= "," . ($xbeg + $xlen - 1);
1823                 if ($ylen > 1)
1824                 $ybeg .= "," . ($ybeg + $ylen - 1);
1825
1826                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1827         }
1828
1829         function _start_block($header) {
1830                 echo $header . "\n";
1831         }
1832
1833         function _end_block() {
1834         }
1835
1836         function _lines($lines, $prefix = ' ') {
1837                 foreach ($lines as $line)
1838                 echo "$prefix $line\n";
1839         }
1840
1841         function _context($lines) {
1842                 $this->_lines($lines);
1843         }
1844
1845         function _added($lines) {
1846                 $this->_lines($lines, '>');
1847         }
1848         function _deleted($lines) {
1849                 $this->_lines($lines, '<');
1850         }
1851
1852         function _changed($orig, $closing) {
1853                 $this->_deleted($orig);
1854                 echo "---\n";
1855                 $this->_added($closing);
1856         }
1857 }
1858
1859 /**
1860  * A formatter that outputs unified diffs
1861  * @ingroup DifferenceEngine
1862  */
1863
1864 class UnifiedDiffFormatter extends DiffFormatter {
1865         var $leading_context_lines = 2;
1866         var $trailing_context_lines = 2;
1867
1868         function _added($lines) {
1869                 $this->_lines($lines, '+');
1870         }
1871         function _deleted($lines) {
1872                 $this->_lines($lines, '-');
1873         }
1874         function _changed($orig, $closing) {
1875                 $this->_deleted($orig);
1876                 $this->_added($closing);
1877         }
1878         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1879                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1880         }
1881 }
1882
1883 /**
1884  * A pseudo-formatter that just passes along the Diff::$edits array
1885  * @ingroup DifferenceEngine
1886  */
1887 class ArrayDiffFormatter extends DiffFormatter {
1888         function format($diff) {
1889                 $oldline = 1;
1890                 $newline = 1;
1891                 $retval = array();
1892                 foreach($diff->edits as $edit)
1893                 switch($edit->type) {
1894                         case 'add':
1895                                 foreach($edit->closing as $l) {
1896                                         $retval[] = array(
1897                                                         'action' => 'add',
1898                                                         'new'=> $l,
1899                                                         'newline' => $newline++
1900                                         );
1901                                 }
1902                                 break;
1903                         case 'delete':
1904                                 foreach($edit->orig as $l) {
1905                                         $retval[] = array(
1906                                                         'action' => 'delete',
1907                                                         'old' => $l,
1908                                                         'oldline' => $oldline++,
1909                                         );
1910                                 }
1911                                 break;
1912                         case 'change':
1913                                 foreach($edit->orig as $i => $l) {
1914                                         $retval[] = array(
1915                                                         'action' => 'change',
1916                                                         'old' => $l,
1917                                                         'new' => @$edit->closing[$i],
1918                                                         'oldline' => $oldline++,
1919                                                         'newline' => $newline++,
1920                                         );
1921                                 }
1922                                 break;
1923                         case 'copy':
1924                                 $oldline += count($edit->orig);
1925                                 $newline += count($edit->orig);
1926                 }
1927                 return $retval;
1928         }
1929 }
1930
1931 /**
1932  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1933  *
1934  */
1935
1936 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1937
1938 /**
1939  * @todo document
1940  * @private
1941  * @ingroup DifferenceEngine
1942  */
1943 class _HWLDF_WordAccumulator {
1944         function _HWLDF_WordAccumulator () {
1945                 $this->_lines = array();
1946                 $this->_line = '';
1947                 $this->_group = '';
1948                 $this->_tag = '';
1949         }
1950
1951         function _flushGroup ($new_tag) {
1952                 if ($this->_group !== '') {
1953                         if ($this->_tag == 'ins')
1954                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1955                         htmlspecialchars ( $this->_group ) . '</ins>';
1956                         elseif ($this->_tag == 'del')
1957                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1958                         htmlspecialchars ( $this->_group ) . '</del>';
1959                         else
1960                         $this->_line .= htmlspecialchars ( $this->_group );
1961                 }
1962                 $this->_group = '';
1963                 $this->_tag = $new_tag;
1964         }
1965
1966         function _flushLine ($new_tag) {
1967                 $this->_flushGroup($new_tag);
1968                 if ($this->_line != '')
1969                 array_push ( $this->_lines, $this->_line );
1970                 else
1971                 # make empty lines visible by inserting an NBSP
1972                 array_push ( $this->_lines, NBSP );
1973                 $this->_line = '';
1974         }
1975
1976         function addWords ($words, $tag = '') {
1977                 if ($tag != $this->_tag)
1978                 $this->_flushGroup($tag);
1979
1980                 foreach ($words as $word) {
1981                         // new-line should only come as first char of word.
1982                         if ($word == '')
1983                         continue;
1984                         if ($word[0] == "\n") {
1985                                 $this->_flushLine($tag);
1986                                 $word = substr($word, 1);
1987                         }
1988                         assert(!strstr($word, "\n"));
1989                         $this->_group .= $word;
1990                 }
1991         }
1992
1993         function getLines() {
1994                 $this->_flushLine('~done');
1995                 return $this->_lines;
1996         }
1997 }
1998
1999 /**
2000  * @todo document
2001  * @private
2002  * @ingroup DifferenceEngine
2003  */
2004 class WordLevelDiff extends MappedDiff {
2005         const MAX_LINE_LENGTH = 10000;
2006
2007         function WordLevelDiff ($orig_lines, $closing_lines) {
2008                 wfProfileIn( __METHOD__ );
2009
2010                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
2011                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
2012
2013                 $this->MappedDiff($orig_words, $closing_words,
2014                 $orig_stripped, $closing_stripped);
2015                 wfProfileOut( __METHOD__ );
2016         }
2017
2018         function _split($lines) {
2019                 wfProfileIn( __METHOD__ );
2020
2021                 $words = array();
2022                 $stripped = array();
2023                 $first = true;
2024                 foreach ( $lines as $line ) {
2025                         # If the line is too long, just pretend the entire line is one big word
2026                         # This prevents resource exhaustion problems
2027                         if ( $first ) {
2028                                 $first = false;
2029                         } else {
2030                                 $words[] = "\n";
2031                                 $stripped[] = "\n";
2032                         }
2033                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
2034                                 $words[] = $line;
2035                                 $stripped[] = $line;
2036                         } else {
2037                                 $m = array();
2038                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
2039                                 $line, $m))
2040                                 {
2041                                         $words = array_merge( $words, $m[0] );
2042                                         $stripped = array_merge( $stripped, $m[1] );
2043                                 }
2044                         }
2045                 }
2046                 wfProfileOut( __METHOD__ );
2047                 return array($words, $stripped);
2048         }
2049
2050         function orig () {
2051                 wfProfileIn( __METHOD__ );
2052                 $orig = new _HWLDF_WordAccumulator;
2053
2054                 foreach ($this->edits as $edit) {
2055                         if ($edit->type == 'copy')
2056                         $orig->addWords($edit->orig);
2057                         elseif ($edit->orig)
2058                         $orig->addWords($edit->orig, 'del');
2059                 }
2060                 $lines = $orig->getLines();
2061                 wfProfileOut( __METHOD__ );
2062                 return $lines;
2063         }
2064
2065         function closing () {
2066                 wfProfileIn( __METHOD__ );
2067                 $closing = new _HWLDF_WordAccumulator;
2068
2069                 foreach ($this->edits as $edit) {
2070                         if ($edit->type == 'copy')
2071                         $closing->addWords($edit->closing);
2072                         elseif ($edit->closing)
2073                         $closing->addWords($edit->closing, 'ins');
2074                 }
2075                 $lines = $closing->getLines();
2076                 wfProfileOut( __METHOD__ );
2077                 return $lines;
2078         }
2079 }
2080
2081 /**
2082  * Wikipedia Table style diff formatter.
2083  * @todo document
2084  * @private
2085  * @ingroup DifferenceEngine
2086  */
2087 class TableDiffFormatter extends DiffFormatter {
2088         function TableDiffFormatter() {
2089                 $this->leading_context_lines = 2;
2090                 $this->trailing_context_lines = 2;
2091         }
2092
2093         public static function escapeWhiteSpace( $msg ) {
2094                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
2095                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
2096                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
2097                 return $msg;
2098         }
2099
2100         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
2101                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
2102                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
2103                 return $r;
2104         }
2105
2106         function _start_block( $header ) {
2107                 echo $header;
2108         }
2109
2110         function _end_block() {
2111         }
2112
2113         function _lines( $lines, $prefix=' ', $color='white' ) {
2114         }
2115
2116         # HTML-escape parameter before calling this
2117         function addedLine( $line ) {
2118                 return $this->wrapLine( '+', 'diff-addedline', $line );
2119         }
2120
2121         # HTML-escape parameter before calling this
2122         function deletedLine( $line ) {
2123                 return $this->wrapLine( '-', 'diff-deletedline', $line );
2124         }
2125
2126         # HTML-escape parameter before calling this
2127         function contextLine( $line ) {
2128                 return $this->wrapLine( ' ', 'diff-context', $line );
2129         }
2130
2131         private function wrapLine( $marker, $class, $line ) {
2132                 if( $line !== '' ) {
2133                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
2134                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
2135                 }
2136                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
2137         }
2138
2139         function emptyLine() {
2140                 return '<td colspan="2">&nbsp;</td>';
2141         }
2142
2143         function _added( $lines ) {
2144                 foreach ($lines as $line) {
2145                         echo '<tr>' . $this->emptyLine() .
2146                         $this->addedLine( '<ins class="diffchange">' .
2147                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
2148                 }
2149         }
2150
2151         function _deleted($lines) {
2152                 foreach ($lines as $line) {
2153                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
2154                         htmlspecialchars ( $line ) . '</del>' ) .
2155                         $this->emptyLine() . "</tr>\n";
2156                 }
2157         }
2158
2159         function _context( $lines ) {
2160                 foreach ($lines as $line) {
2161                         echo '<tr>' .
2162                         $this->contextLine( htmlspecialchars ( $line ) ) .
2163                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
2164                 }
2165         }
2166
2167         function _changed( $orig, $closing ) {
2168                 wfProfileIn( __METHOD__ );
2169
2170                 $diff = new WordLevelDiff( $orig, $closing );
2171                 $del = $diff->orig();
2172                 $add = $diff->closing();
2173
2174                 # Notice that WordLevelDiff returns HTML-escaped output.
2175                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2176
2177                 while ( $line = array_shift( $del ) ) {
2178                         $aline = array_shift( $add );
2179                         echo '<tr>' . $this->deletedLine( $line ) .
2180                         $this->addedLine( $aline ) . "</tr>\n";
2181                 }
2182                 foreach ($add as $line) {       # If any leftovers
2183                         echo '<tr>' . $this->emptyLine() .
2184                         $this->addedLine( $line ) . "</tr>\n";
2185                 }
2186                 wfProfileOut( __METHOD__ );
2187         }
2188 }