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(
 259                                         'type' => 'revision',
 260                                         'target' => $this->mOldRev->mTitle->getPrefixedDbkey(),
 261                                         'ids' => $this->mOldRev->getId()
 262                                 );
 263                                 $ldel = $sk->revDeleteLink( $query, $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 264                         }
 265                         $ldel = "&nbsp;&nbsp;&nbsp;$ldel ";
 266                         // We don't currently handle well changing the top revision's settings
 267                         if( $this->mNewRev->isCurrent() ) {
 268                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 269                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 270                                 // If revision was hidden from sysops
 271                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 272                         } else {
 273                                 $query = array(
 274                                         'type' => 'revision',
 275                                         'target' =>  $this->mNewRev->mTitle->getPrefixedDbkey(),
 276                                         'ids' => $this->mNewRev->getId()
 277                                 );
 278                                 $rdel = $sk->revDeleteLink( $query, $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 279                         }
 280                         $rdel = "&nbsp;&nbsp;&nbsp;$rdel ";
 281                 }
 282
 283                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 284                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, !$this->unhide ) . "</div>" .
 285                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, !$this->unhide ).$ldel."</div>" .
 286                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 287                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 288                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, !$this->unhide ) . " $rollback</div>" .
 289                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, !$this->unhide ).$rdel."</div>" .
 290                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 291
 292                 # Check if this user can see the revisions
 293                 $allowed = $this->mOldRev->userCan(Revision::DELETED_TEXT)
 294                         && $this->mNewRev->userCan(Revision::DELETED_TEXT);
 295                 $deleted = $this->mOldRev->isDeleted(Revision::DELETED_TEXT)
 296                         || $this->mNewRev->isDeleted(Revision::DELETED_TEXT);
 297                 # Output the diff if allowed...
 298                 if( $deleted && (!$this->unhide || !$allowed) ) {
 299                         $this->showDiffStyle();
 300                         $multi = $this->getMultiNotice();
 301                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 302                         if( !$allowed ) {
 303                                 # Give explanation for why revision is not visible
 304                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 305                                         array( 'rev-deleted-no-diff' ) );
 306                         } else {
 307                                 # Give explanation and add a link to view the diff...
 308                                 $link = $this->mTitle->getFullUrl( "diff={$this->mNewid}&oldid={$this->mOldid}&unhide=1" );
 309                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 310                                         array( 'rev-deleted-unhide-diff', $link ) );
 311                         }
 312                 } else if( $wgEnableHtmlDiff && $this->htmldiff ) {
 313                         $multi = $this->getMultiNotice();
 314                         $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'wikicodecomparison' ),
 315                                 'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=0', '', '', 'id="differences-switchtype"' ).'</div>');
 316                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 317                         $this->renderHtmlDiff();
 318                 } else {
 319                         if( $wgEnableHtmlDiff ) {
 320                                 $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'visualcomparison' ),
 321                                         'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=1', '', '', 'id="differences-switchtype"' ).'</div>');
 322                         }
 323                         $this->showDiff( $oldHeader, $newHeader );
 324                         if( !$diffOnly ) {
 325                                 $this->renderNewRevision();
 326                         }
 327                 }
 328                 wfProfileOut( __METHOD__ );
 329         }
 330
 331         /**
 332          * Show the new revision of the page.
 333          */
 334         function renderNewRevision() {
 335                 global $wgOut, $wgUser;
 336                 wfProfileIn( __METHOD__ );
 337
 338                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 339                 # Add deleted rev tag if needed
 340                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 341                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 342                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 343                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 344                 }
 345
 346                 if( !$this->mNewRev->isCurrent() ) {
 347                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 348                 }
 349
 350                 $this->loadNewText();
 351                 if( is_object( $this->mNewRev ) ) {
 352                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 353                 }
 354
 355                 if( $this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage() ) {
 356                         // Stolen from Article::view --AG 2007-10-11
 357                         // Give hooks a chance to customise the output
 358                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 359                                 // Wrap the whole lot in a <pre> and don't parse
 360                                 $m = array();
 361                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 362                                 $wgOut->addHTML( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 363                                 $wgOut->addHTML( htmlspecialchars( $this->mNewtext ) );
 364                                 $wgOut->addHTML( "\n</pre>\n" );
 365                         }
 366                 } else {
 367                         $wgOut->addWikiTextTidy( $this->mNewtext );
 368                 }
 369
 370                 if( is_object( $this->mNewRev ) && !$this->mNewRev->isCurrent() ) {
 371                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 372                 }
 373                 # Add redundant patrol link on bottom...
 374                 if( $this->mRcidMarkPatrolled && $this->mTitle->quickUserCan('patrol') ) {
 375                         $sk = $wgUser->getSkin();
 376                         $wgOut->addHTML(
 377                                 "<div class='patrollink'>[" . $sk->makeKnownLinkObj( $this->mTitle,
 378                                         wfMsgHtml( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) .
 379                                 ']</div>'
 380                          );
 381                 }
 382
 383                 wfProfileOut( __METHOD__ );
 384         }
 385
 386
 387         function renderHtmlDiff() {
 388                 global $wgOut, $wgParser, $wgDebugComments;
 389                 wfProfileIn( __METHOD__ );
 390
 391                 $this->showDiffStyle();
 392
 393                 $wgOut->addHTML( '<h2>'.wfMsgHtml( 'visual-comparison' )."</h2>\n" );
 394                 #add deleted rev tag if needed
 395                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 396                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 397                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 398                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 399                 }
 400
 401                 if( !$this->mNewRev->isCurrent() ) {
 402                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 403                 }
 404
 405                 $this->loadText();
 406
 407                 // Old revision
 408                 if( is_object( $this->mOldRev ) ) {
 409                         $wgOut->setRevisionId( $this->mOldRev->getId() );
 410                 }
 411
 412                 $popts = $wgOut->parserOptions();
 413                 $oldTidy = $popts->setTidy( true );
 414                 $popts->setEditSection( false );
 415
 416                 $parserOutput = $wgParser->parse( $this->mOldtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 417                 $popts->setTidy( $oldTidy );
 418
 419                 //only for new?
 420                 //$wgOut->addParserOutputNoText( $parserOutput );
 421                 $oldHtml = $parserOutput->getText();
 422                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$oldHtml ) );
 423
 424                 // New revision
 425                 if( is_object( $this->mNewRev ) ) {
 426                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 427                 }
 428
 429                 $popts = $wgOut->parserOptions();
 430                 $oldTidy = $popts->setTidy( true );
 431
 432                 $parserOutput = $wgParser->parse( $this->mNewtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 433                 $popts->setTidy( $oldTidy );
 434
 435                 $wgOut->addParserOutputNoText( $parserOutput );
 436                 $newHtml = $parserOutput->getText();
 437                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$newHtml ) );
 438
 439                 unset($parserOutput, $popts);
 440
 441                 $differ = new HTMLDiffer(new DelegatingContentHandler($wgOut));
 442                 $differ->htmlDiff($oldHtml, $newHtml);
 443                 if ( $wgDebugComments ) {
 444                         $wgOut->addHTML( "\n<!-- HtmlDiff Debug Output:\n" . HTMLDiffer::getDebugOutput() . " End Debug -->" );
 445                 }
 446
 447                 wfProfileOut( __METHOD__ );
 448         }
 449
 450         /**
 451          * Show the first revision of an article. Uses normal diff headers in
 452          * contrast to normal "old revision" display style.
 453          */
 454         function showFirstRevision() {
 455                 global $wgOut, $wgUser;
 456                 wfProfileIn( __METHOD__ );
 457
 458                 # Get article text from the DB
 459                 #
 460                 if ( ! $this->loadNewText() ) {
 461                         $t = $this->mTitle->getPrefixedText();
 462                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 463                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 464                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 465                         wfProfileOut( __METHOD__ );
 466                         return;
 467                 }
 468                 if ( $this->mNewRev->isCurrent() ) {
 469                         $wgOut->setArticleFlag( true );
 470                 }
 471
 472                 # Check if user is allowed to look at this page. If not, bail out.
 473                 #
 474                 if ( !$this->mTitle->userCanRead() ) {
 475                         $wgOut->loginToUse();
 476                         $wgOut->output();
 477                         wfProfileOut( __METHOD__ );
 478                         throw new MWException("Permission Error: you do not have access to view this page");
 479                 }
 480
 481                 # Prepare the header box
 482                 #
 483                 $sk = $wgUser->getSkin();
 484
 485                 $next = $this->mTitle->getNextRevisionID( $this->mNewid );
 486                 if( !$next ) {
 487                         $nextlink = '';
 488                 } else {
 489                         $nextlink = '<br/>' . $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 490                                 'diff=next&oldid=' . $this->mNewid.$this->htmlDiffArgument(), '', '', 'id="differences-nextlink"' );
 491                 }
 492                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\">" .
 493                         $sk->revUserTools( $this->mNewRev ) . "<br/>" . $sk->revComment( $this->mNewRev ) . $nextlink . "</div>\n";
 494
 495                 $wgOut->addHTML( $header );
 496
 497                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 498                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 499
 500                 wfProfileOut( __METHOD__ );
 501         }
 502
 503         function htmlDiffArgument(){
 504                 global $wgEnableHtmlDiff;
 505                 if($wgEnableHtmlDiff){
 506                         if($this->htmldiff){
 507                                 return '&htmldiff=1';
 508                         }else{
 509                                 return '&htmldiff=0';
 510                         }
 511                 }else{
 512                         return '';
 513                 }
 514         }
 515
 516         /**
 517          * Get the diff text, send it to $wgOut
 518          * Returns false if the diff could not be generated, otherwise returns true
 519          */
 520         function showDiff( $otitle, $ntitle ) {
 521                 global $wgOut;
 522                 $diff = $this->getDiff( $otitle, $ntitle );
 523                 if ( $diff === false ) {
 524                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 525                         return false;
 526                 } else {
 527                         $this->showDiffStyle();
 528                         $wgOut->addHTML( $diff );
 529                         return true;
 530                 }
 531         }
 532
 533         /**
 534          * Add style sheets and supporting JS for diff display.
 535          */
 536         function showDiffStyle() {
 537                 global $wgStylePath, $wgStyleVersion, $wgOut;
 538                 $wgOut->addStyle( 'common/diff.css' );
 539
 540                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 541                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 542         }
 543
 544         /**
 545          * Get complete diff table, including header
 546          *
 547          * @param Title $otitle Old title
 548          * @param Title $ntitle New title
 549          * @return mixed
 550          */
 551         function getDiff( $otitle, $ntitle ) {
 552                 $body = $this->getDiffBody();
 553                 if ( $body === false ) {
 554                         return false;
 555                 } else {
 556                         $multi = $this->getMultiNotice();
 557                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 558                 }
 559         }
 560
 561         /**
 562          * Get the diff table body, without header
 563          *
 564          * @return mixed
 565          */
 566         function getDiffBody() {
 567                 global $wgMemc;
 568                 wfProfileIn( __METHOD__ );
 569                 $this->mCacheHit = true;
 570                 // Check if the diff should be hidden from this user
 571                 if ( !$this->loadRevisionData() )
 572                         return '';
 573                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 574                         return '';
 575                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 576                         return '';
 577                 } else if ( $this->mOldRev && $this->mNewRev && $this->mOldRev->getID() == $this->mNewRev->getID() ) {
 578                         return '';
 579                 }
 580                 // Cacheable?
 581                 $key = false;
 582                 if ( $this->mOldid && $this->mNewid ) {
 583                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 584                         // Try cache
 585                         if ( !$this->mRefreshCache ) {
 586                                 $difftext = $wgMemc->get( $key );
 587                                 if ( $difftext ) {
 588                                         wfIncrStats( 'diff_cache_hit' );
 589                                         $difftext = $this->localiseLineNumbers( $difftext );
 590                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 591                                         wfProfileOut( __METHOD__ );
 592                                         return $difftext;
 593                                 }
 594                         } // don't try to load but save the result
 595                 }
 596                 $this->mCacheHit = false;
 597
 598                 // Loadtext is permission safe, this just clears out the diff
 599                 if ( !$this->loadText() ) {
 600                         wfProfileOut( __METHOD__ );
 601                         return false;
 602                 }
 603
 604                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 605
 606                 // Save to cache for 7 days
 607                 if ( !wfRunHooks( 'AbortDiffCache', array( &$this ) ) ) {
 608                         wfIncrStats( 'diff_uncacheable' );
 609                 } else if ( $key !== false && $difftext !== false ) {
 610                         wfIncrStats( 'diff_cache_miss' );
 611                         $wgMemc->set( $key, $difftext, 7*86400 );
 612                 } else {
 613                         wfIncrStats( 'diff_uncacheable' );
 614                 }
 615                 // Replace line numbers with the text in the user's language
 616                 if ( $difftext !== false ) {
 617                         $difftext = $this->localiseLineNumbers( $difftext );
 618                 }
 619                 wfProfileOut( __METHOD__ );
 620                 return $difftext;
 621         }
 622
 623         /**
 624          * Make sure the proper modules are loaded before we try to
 625          * make the diff
 626          */
 627         private function initDiffEngines() {
 628                 global $wgExternalDiffEngine;
 629                 if ( $wgExternalDiffEngine == 'wikidiff' && !function_exists( 'wikidiff_do_diff' ) ) {
 630                         wfProfileIn( __METHOD__ . '-php_wikidiff.so' );
 631                         wfSuppressWarnings();
 632                         dl( 'php_wikidiff.so' );
 633                         wfRestoreWarnings();
 634                         wfProfileOut( __METHOD__ . '-php_wikidiff.so' );
 635                 }
 636                 else if ( $wgExternalDiffEngine == 'wikidiff2' && !function_exists( 'wikidiff2_do_diff' ) ) {
 637                         wfProfileIn( __METHOD__ . '-php_wikidiff2.so' );
 638                         wfSuppressWarnings();
 639                         dl( 'php_wikidiff2.so' );
 640                         wfRestoreWarnings();
 641                         wfProfileOut( __METHOD__ . '-php_wikidiff2.so' );
 642                 }
 643         }
 644
 645         /**
 646          * Generate a diff, no caching
 647          * $otext and $ntext must be already segmented
 648          */
 649         function generateDiffBody( $otext, $ntext ) {
 650                 global $wgExternalDiffEngine, $wgContLang;
 651
 652                 $otext = str_replace( "\r\n", "\n", $otext );
 653                 $ntext = str_replace( "\r\n", "\n", $ntext );
 654
 655                 $this->initDiffEngines();
 656
 657                 if ( $wgExternalDiffEngine == 'wikidiff' && function_exists( 'wikidiff_do_diff' ) ) {
 658                         # For historical reasons, external diff engine expects
 659                         # input text to be HTML-escaped already
 660                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 661                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 662                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 663                         $this->debug( 'wikidiff1' );
 664                 }
 665
 666                 if ( $wgExternalDiffEngine == 'wikidiff2' && function_exists( 'wikidiff2_do_diff' ) ) {
 667                         # Better external diff engine, the 2 may some day be dropped
 668                         # This one does the escaping and segmenting itself
 669                         wfProfileIn( 'wikidiff2_do_diff' );
 670                         $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 671                         $text .= $this->debug( 'wikidiff2' );
 672                         wfProfileOut( 'wikidiff2_do_diff' );
 673                         return $text;
 674                 }
 675                 if ( $wgExternalDiffEngine != 'wikidiff3' && $wgExternalDiffEngine !== false ) {
 676                         # Diff via the shell
 677                         global $wgTmpDirectory;
 678                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 679                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 680
 681                         $tempFile1 = fopen( $tempName1, "w" );
 682                         if ( !$tempFile1 ) {
 683                                 wfProfileOut( __METHOD__ );
 684                                 return false;
 685                         }
 686                         $tempFile2 = fopen( $tempName2, "w" );
 687                         if ( !$tempFile2 ) {
 688                                 wfProfileOut( __METHOD__ );
 689                                 return false;
 690                         }
 691                         fwrite( $tempFile1, $otext );
 692                         fwrite( $tempFile2, $ntext );
 693                         fclose( $tempFile1 );
 694                         fclose( $tempFile2 );
 695                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 696                         wfProfileIn( __METHOD__ . "-shellexec" );
 697                         $difftext = wfShellExec( $cmd );
 698                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 699                         wfProfileOut( __METHOD__ . "-shellexec" );
 700                         unlink( $tempName1 );
 701                         unlink( $tempName2 );
 702                         return $difftext;
 703                 }
 704
 705                 # Native PHP diff
 706                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 707                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 708                 $diffs = new Diff( $ota, $nta );
 709                 $formatter = new TableDiffFormatter();
 710                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 711                 $this->debug();
 712         }
 713
 714         /**
 715          * Generate a debug comment indicating diff generating time,
 716          * server node, and generator backend.
 717          */
 718         protected function debug( $generator="internal" ) {
 719                 global $wgShowHostnames;
 720                 $data = array( $generator );
 721                 if( $wgShowHostnames ) {
 722                         $data[] = wfHostname();
 723                 }
 724                 $data[] = wfTimestamp( TS_DB );
 725                 return "<!-- diff generator: " .
 726                 implode( " ",
 727                 array_map(
 728                                         "htmlspecialchars",
 729                 $data ) ) .
 730                         " -->\n";
 731         }
 732
 733         /**
 734          * Replace line numbers with the text in the user's language
 735          */
 736         function localiseLineNumbers( $text ) {
 737                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 738                 array( &$this, 'localiseLineNumbersCb' ), $text );
 739         }
 740
 741         function localiseLineNumbersCb( $matches ) {
 742                 global $wgLang;
 743                 return wfMsgExt( 'lineno', 'escape', $wgLang->formatNum( $matches[1] ) );
 744         }
 745
 746
 747         /**
 748          * If there are revisions between the ones being compared, return a note saying so.
 749          */
 750         function getMultiNotice() {
 751                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 752                 return '';
 753
 754                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 755                         // Comparing two different pages? Count would be meaningless.
 756                         return '';
 757                 }
 758
 759                 $oldid = $this->mOldRev->getId();
 760                 $newid = $this->mNewRev->getId();
 761                 if ( $oldid > $newid ) {
 762                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 763                 }
 764
 765                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 766                 if ( !$n )
 767                 return '';
 768
 769                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 770         }
 771
 772
 773         /**
 774          * Add the header to a diff body
 775          */
 776         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 777                 $colspan = 1;
 778                 $header = "<table class='diff'>";
 779                 if( $diff ) { // Safari/Chrome show broken output if cols not used
 780                         $header .= "
 781                         <col class='diff-marker' />
 782                         <col class='diff-content' />
 783                         <col class='diff-marker' />
 784                         <col class='diff-content' />";
 785                         $colspan = 2;
 786                 }
 787                 $header .= "
 788                 <tr valign='top'>
 789                 <td colspan='$colspan' class='diff-otitle'>{$otitle}</td>
 790                 <td colspan='$colspan' class='diff-ntitle'>{$ntitle}</td>
 791                 </tr>";
 792
 793                 if ( $multi != '' )
 794                 $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 795
 796                 return $header . $diff . "</table>";
 797         }
 798
 799         /**
 800          * Use specified text instead of loading from the database
 801          */
 802         function setText( $oldText, $newText ) {
 803                 $this->mOldtext = $oldText;
 804                 $this->mNewtext = $newText;
 805                 $this->mTextLoaded = 2;
 806                 $this->mRevisionsLoaded = true;
 807         }
 808
 809         /**
 810          * Load revision metadata for the specified articles. If newid is 0, then compare
 811          * the old article in oldid to the current article; if oldid is 0, then
 812          * compare the current article to the immediately previous one (ignoring the
 813          * value of newid).
 814          *
 815          * If oldid is false, leave the corresponding revision object set
 816          * to false. This is impossible via ordinary user input, and is provided for
 817          * API convenience.
 818          */
 819         function loadRevisionData() {
 820                 global $wgLang, $wgUser;
 821                 if ( $this->mRevisionsLoaded ) {
 822                         return true;
 823                 } else {
 824                         // Whether it succeeds or fails, we don't want to try again
 825                         $this->mRevisionsLoaded = true;
 826                 }
 827
 828                 // Load the new revision object
 829                 $this->mNewRev = $this->mNewid
 830                         ? Revision::newFromId( $this->mNewid )
 831                         : Revision::newFromTitle( $this->mTitle );
 832                 if( !$this->mNewRev instanceof Revision )
 833                         return false;
 834
 835                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 836                 $this->mNewid = $this->mNewRev->getId();
 837
 838                 // Check if page is editable
 839                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 840
 841                 // Set assorted variables
 842                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 843                 $dateofrev = $wgLang->date( $this->mNewRev->getTimestamp(), true );
 844                 $timeofrev = $wgLang->time( $this->mNewRev->getTimestamp(), true );
 845                 $this->mNewPage = $this->mNewRev->getTitle();
 846                 if( $this->mNewRev->isCurrent() ) {
 847                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 848                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev-asof', $timestamp ) );
 849                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 850
 851                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 852                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 853                 } else {
 854                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 855                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 856                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp, $dateofrev, $timeofrev ) );
 857
 858                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 859                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 860                 }
 861                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 862                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 863                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 864                         $this->mNewtitle = "<span class='history-deleted'>{$this->mNewtitle}</span>";
 865                 }
 866
 867                 // Load the old revision object
 868                 $this->mOldRev = false;
 869                 if( $this->mOldid ) {
 870                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 871                 } elseif ( $this->mOldid === 0 ) {
 872                         $rev = $this->mNewRev->getPrevious();
 873                         if( $rev ) {
 874                                 $this->mOldid = $rev->getId();
 875                                 $this->mOldRev = $rev;
 876                         } else {
 877                                 // No previous revision; mark to show as first-version only.
 878                                 $this->mOldid = false;
 879                                 $this->mOldRev = false;
 880                         }
 881                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 882
 883                 if( is_null( $this->mOldRev ) ) {
 884                         return false;
 885                 }
 886
 887                 if ( $this->mOldRev ) {
 888                         $this->mOldPage = $this->mOldRev->getTitle();
 889
 890                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 891                         $dateofrev = $wgLang->date( $this->mOldRev->getTimestamp(), true );
 892                         $timeofrev = $wgLang->time( $this->mOldRev->getTimestamp(), true );
 893                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 894                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 895                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t, $dateofrev, $timeofrev ) );
 896
 897                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 898                         . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 899                         // Add an "undo" link
 900                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 901                         $htmlLink = htmlspecialchars( wfMsg( 'editundo' ) );
 902                         $htmlTitle = $wgUser->getSkin()->tooltip( 'undo' );
 903                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 904                                 $this->mNewtitle .= " (<a href='$newUndo' $htmlTitle>" . $htmlLink . "</a>)";
 905                         }
 906
 907                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 908                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 909                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 910                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 911                         }
 912                 }
 913
 914                 return true;
 915         }
 916
 917         /**
 918          * Load the text of the revisions, as well as revision data.
 919          */
 920         function loadText() {
 921                 if ( $this->mTextLoaded == 2 ) {
 922                         return true;
 923                 } else {
 924                         // Whether it succeeds or fails, we don't want to try again
 925                         $this->mTextLoaded = 2;
 926                 }
 927
 928                 if ( !$this->loadRevisionData() ) {
 929                         return false;
 930                 }
 931                 if ( $this->mOldRev ) {
 932                         $this->mOldtext = $this->mOldRev->getText( Revision::FOR_THIS_USER );
 933                         if ( $this->mOldtext === false ) {
 934                                 return false;
 935                         }
 936                 }
 937                 if ( $this->mNewRev ) {
 938                         $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 939                         if ( $this->mNewtext === false ) {
 940                                 return false;
 941                         }
 942                 }
 943                 return true;
 944         }
 945
 946         /**
 947          * Load the text of the new revision, not the old one
 948          */
 949         function loadNewText() {
 950                 if ( $this->mTextLoaded >= 1 ) {
 951                         return true;
 952                 } else {
 953                         $this->mTextLoaded = 1;
 954                 }
 955                 if ( !$this->loadRevisionData() ) {
 956                         return false;
 957                 }
 958                 $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 959                 return true;
 960         }
 961
 962
 963 }
 964
 965 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 966 //
 967 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 968 // You may copy this code freely under the conditions of the GPL.
 969 //
 970
 971 define('USE_ASSERTS', function_exists('assert'));
 972
 973 /**
 974  * @todo document
 975  * @private
 976  * @ingroup DifferenceEngine
 977  */
 978 class _DiffOp {
 979         var $type;
 980         var $orig;
 981         var $closing;
 982
 983         function reverse() {
 984                 trigger_error('pure virtual', E_USER_ERROR);
 985         }
 986
 987         function norig() {
 988                 return $this->orig ? sizeof($this->orig) : 0;
 989         }
 990
 991         function nclosing() {
 992                 return $this->closing ? sizeof($this->closing) : 0;
 993         }
 994 }
 995
 996 /**
 997  * @todo document
 998  * @private
 999  * @ingroup DifferenceEngine
1000  */
1001 class _DiffOp_Copy extends _DiffOp {
1002         var $type = 'copy';
1003
1004         function _DiffOp_Copy ($orig, $closing = false) {
1005                 if (!is_array($closing))
1006                 $closing = $orig;
1007                 $this->orig = $orig;
1008                 $this->closing = $closing;
1009         }
1010
1011         function reverse() {
1012                 return new _DiffOp_Copy($this->closing, $this->orig);
1013         }
1014 }
1015
1016 /**
1017  * @todo document
1018  * @private
1019  * @ingroup DifferenceEngine
1020  */
1021 class _DiffOp_Delete extends _DiffOp {
1022         var $type = 'delete';
1023
1024         function _DiffOp_Delete ($lines) {
1025                 $this->orig = $lines;
1026                 $this->closing = false;
1027         }
1028
1029         function reverse() {
1030                 return new _DiffOp_Add($this->orig);
1031         }
1032 }
1033
1034 /**
1035  * @todo document
1036  * @private
1037  * @ingroup DifferenceEngine
1038  */
1039 class _DiffOp_Add extends _DiffOp {
1040         var $type = 'add';
1041
1042         function _DiffOp_Add ($lines) {
1043                 $this->closing = $lines;
1044                 $this->orig = false;
1045         }
1046
1047         function reverse() {
1048                 return new _DiffOp_Delete($this->closing);
1049         }
1050 }
1051
1052 /**
1053  * @todo document
1054  * @private
1055  * @ingroup DifferenceEngine
1056  */
1057 class _DiffOp_Change extends _DiffOp {
1058         var $type = 'change';
1059
1060         function _DiffOp_Change ($orig, $closing) {
1061                 $this->orig = $orig;
1062                 $this->closing = $closing;
1063         }
1064
1065         function reverse() {
1066                 return new _DiffOp_Change($this->closing, $this->orig);
1067         }
1068 }
1069
1070 /**
1071  * Class used internally by Diff to actually compute the diffs.
1072  *
1073  * The algorithm used here is mostly lifted from the perl module
1074  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
1075  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
1076  *
1077  * More ideas are taken from:
1078  *       http://www.ics.uci.edu/~eppstein/161/960229.html
1079  *
1080  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
1081  * diffutils-2.7, which can be found at:
1082  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
1083  *
1084  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
1085  * are my own.
1086  *
1087  * Line length limits for robustness added by Tim Starling, 2005-08-31
1088  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
1089  *
1090  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
1091  * @private
1092  * @ingroup DifferenceEngine
1093  */
1094 class _DiffEngine {
1095
1096         const MAX_XREF_LENGTH =  10000;
1097
1098         function diff ($from_lines, $to_lines){
1099                 wfProfileIn( __METHOD__ );
1100
1101                 // Diff and store locally
1102                 $this->diff_local($from_lines, $to_lines);
1103
1104                 // Merge edits when possible
1105                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
1106                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
1107
1108                 // Compute the edit operations.
1109                 $n_from = sizeof($from_lines);
1110                 $n_to = sizeof($to_lines);
1111
1112                 $edits = array();
1113                 $xi = $yi = 0;
1114                 while ($xi < $n_from || $yi < $n_to) {
1115                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
1116                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
1117
1118                         // Skip matching "snake".
1119                         $copy = array();
1120                         while ( $xi < $n_from && $yi < $n_to
1121                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
1122                                 $copy[] = $from_lines[$xi++];
1123                                 ++$yi;
1124                         }
1125                         if ($copy)
1126                         $edits[] = new _DiffOp_Copy($copy);
1127
1128                         // Find deletes & adds.
1129                         $delete = array();
1130                         while ($xi < $n_from && $this->xchanged[$xi])
1131                         $delete[] = $from_lines[$xi++];
1132
1133                         $add = array();
1134                         while ($yi < $n_to && $this->ychanged[$yi])
1135                         $add[] = $to_lines[$yi++];
1136
1137                         if ($delete && $add)
1138                         $edits[] = new _DiffOp_Change($delete, $add);
1139                         elseif ($delete)
1140                         $edits[] = new _DiffOp_Delete($delete);
1141                         elseif ($add)
1142                         $edits[] = new _DiffOp_Add($add);
1143                 }
1144                 wfProfileOut( __METHOD__ );
1145                 return $edits;
1146         }
1147
1148         function diff_local ($from_lines, $to_lines) {
1149                 global $wgExternalDiffEngine;
1150                 wfProfileIn( __METHOD__);
1151
1152                 if($wgExternalDiffEngine == 'wikidiff3'){
1153                         // wikidiff3
1154                         $wikidiff3 = new WikiDiff3();
1155                         $wikidiff3->diff($from_lines, $to_lines);
1156                         $this->xchanged = $wikidiff3->removed;
1157                         $this->ychanged = $wikidiff3->added;
1158                         unset($wikidiff3);
1159                 }else{
1160                         // old diff
1161                         $n_from = sizeof($from_lines);
1162                         $n_to = sizeof($to_lines);
1163                         $this->xchanged = $this->ychanged = array();
1164                         $this->xv = $this->yv = array();
1165                         $this->xind = $this->yind = array();
1166                         unset($this->seq);
1167                         unset($this->in_seq);
1168                         unset($this->lcs);
1169
1170                         // Skip leading common lines.
1171                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
1172                                 if ($from_lines[$skip] !== $to_lines[$skip])
1173                                 break;
1174                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
1175                         }
1176                         // Skip trailing common lines.
1177                         $xi = $n_from; $yi = $n_to;
1178                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
1179                                 if ($from_lines[$xi] !== $to_lines[$yi])
1180                                 break;
1181                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
1182                         }
1183
1184                         // Ignore lines which do not exist in both files.
1185                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1186                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
1187                         }
1188
1189                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
1190                                 $line = $to_lines[$yi];
1191                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
1192                                 continue;
1193                                 $yhash[$this->_line_hash($line)] = 1;
1194                                 $this->yv[] = $line;
1195                                 $this->yind[] = $yi;
1196                         }
1197                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1198                                 $line = $from_lines[$xi];
1199                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
1200                                 continue;
1201                                 $this->xv[] = $line;
1202                                 $this->xind[] = $xi;
1203                         }
1204
1205                         // Find the LCS.
1206                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
1207                 }
1208                 wfProfileOut( __METHOD__ );
1209         }
1210
1211         /**
1212          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1213          */
1214         function _line_hash( $line ) {
1215                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1216                         return md5( $line );
1217                 } else {
1218                         return $line;
1219                 }
1220         }
1221
1222         /* Divide the Largest Common Subsequence (LCS) of the sequences
1223          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1224          * sized segments.
1225          *
1226          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1227          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1228          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1229          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1230          * that (X0, Y0) == (XOFF, YOFF) and
1231          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1232          *
1233          * This function assumes that the first lines of the specified portions
1234          * of the two files do not match, and likewise that the last lines do not
1235          * match.  The caller must trim matching lines from the beginning and end
1236          * of the portions it is going to specify.
1237          */
1238         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1239                 $flip = false;
1240
1241                 if ($xlim - $xoff > $ylim - $yoff) {
1242                         // Things seems faster (I'm not sure I understand why)
1243                         // when the shortest sequence in X.
1244                         $flip = true;
1245                         list ($xoff, $xlim, $yoff, $ylim)
1246                         = array( $yoff, $ylim, $xoff, $xlim);
1247                 }
1248
1249                 if ($flip)
1250                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1251                 $ymatches[$this->xv[$i]][] = $i;
1252                 else
1253                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1254                 $ymatches[$this->yv[$i]][] = $i;
1255
1256                 $this->lcs = 0;
1257                 $this->seq[0]= $yoff - 1;
1258                 $this->in_seq = array();
1259                 $ymids[0] = array();
1260
1261                 $numer = $xlim - $xoff + $nchunks - 1;
1262                 $x = $xoff;
1263                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1264                         if ($chunk > 0)
1265                         for ($i = 0; $i <= $this->lcs; $i++)
1266                         $ymids[$i][$chunk-1] = $this->seq[$i];
1267
1268                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1269                         for ( ; $x < $x1; $x++) {
1270                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1271                                 if (empty($ymatches[$line]))
1272                                 continue;
1273                                 $matches = $ymatches[$line];
1274                                 reset($matches);
1275                                 while (list ($junk, $y) = each($matches))
1276                                 if (empty($this->in_seq[$y])) {
1277                                         $k = $this->_lcs_pos($y);
1278                                         USE_ASSERTS && assert($k > 0);
1279                                         $ymids[$k] = $ymids[$k-1];
1280                                         break;
1281                                 }
1282                                 while (list ( /* $junk */, $y) = each($matches)) {
1283                                         if ($y > $this->seq[$k-1]) {
1284                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1285                                                 // Optimization: this is a common case:
1286                                                 //      next match is just replacing previous match.
1287                                                 $this->in_seq[$this->seq[$k]] = false;
1288                                                 $this->seq[$k] = $y;
1289                                                 $this->in_seq[$y] = 1;
1290                                         } else if (empty($this->in_seq[$y])) {
1291                                                 $k = $this->_lcs_pos($y);
1292                                                 USE_ASSERTS && assert($k > 0);
1293                                                 $ymids[$k] = $ymids[$k-1];
1294                                         }
1295                                 }
1296                         }
1297                 }
1298
1299                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1300                 $ymid = $ymids[$this->lcs];
1301                 for ($n = 0; $n < $nchunks - 1; $n++) {
1302                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1303                         $y1 = $ymid[$n] + 1;
1304                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1305                 }
1306                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1307
1308                 return array($this->lcs, $seps);
1309         }
1310
1311         function _lcs_pos ($ypos) {
1312                 $end = $this->lcs;
1313                 if ($end == 0 || $ypos > $this->seq[$end]) {
1314                         $this->seq[++$this->lcs] = $ypos;
1315                         $this->in_seq[$ypos] = 1;
1316                         return $this->lcs;
1317                 }
1318
1319                 $beg = 1;
1320                 while ($beg < $end) {
1321                         $mid = (int)(($beg + $end) / 2);
1322                         if ( $ypos > $this->seq[$mid] )
1323                         $beg = $mid + 1;
1324                         else
1325                         $end = $mid;
1326                 }
1327
1328                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1329
1330                 $this->in_seq[$this->seq[$end]] = false;
1331                 $this->seq[$end] = $ypos;
1332                 $this->in_seq[$ypos] = 1;
1333                 return $end;
1334         }
1335
1336         /* Find LCS of two sequences.
1337          *
1338          * The results are recorded in the vectors $this->{x,y}changed[], by
1339          * storing a 1 in the element for each line that is an insertion
1340          * or deletion (ie. is not in the LCS).
1341          *
1342          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1343          *
1344          * Note that XLIM, YLIM are exclusive bounds.
1345          * All line numbers are origin-0 and discarded lines are not counted.
1346          */
1347         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1348                 // Slide down the bottom initial diagonal.
1349                 while ($xoff < $xlim && $yoff < $ylim
1350                 && $this->xv[$xoff] == $this->yv[$yoff]) {
1351                         ++$xoff;
1352                         ++$yoff;
1353                 }
1354
1355                 // Slide up the top initial diagonal.
1356                 while ($xlim > $xoff && $ylim > $yoff
1357                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1358                         --$xlim;
1359                         --$ylim;
1360                 }
1361
1362                 if ($xoff == $xlim || $yoff == $ylim)
1363                 $lcs = 0;
1364                 else {
1365                         // This is ad hoc but seems to work well.
1366                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1367                         //$nchunks = max(2,min(8,(int)$nchunks));
1368                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1369                         list ($lcs, $seps)
1370                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1371                 }
1372
1373                 if ($lcs == 0) {
1374                         // X and Y sequences have no common subsequence:
1375                         // mark all changed.
1376                         while ($yoff < $ylim)
1377                         $this->ychanged[$this->yind[$yoff++]] = 1;
1378                         while ($xoff < $xlim)
1379                         $this->xchanged[$this->xind[$xoff++]] = 1;
1380                 } else {
1381                         // Use the partitions to split this problem into subproblems.
1382                         reset($seps);
1383                         $pt1 = $seps[0];
1384                         while ($pt2 = next($seps)) {
1385                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1386                                 $pt1 = $pt2;
1387                         }
1388                 }
1389         }
1390
1391         /* Adjust inserts/deletes of identical lines to join changes
1392          * as much as possible.
1393          *
1394          * We do something when a run of changed lines include a
1395          * line at one end and has an excluded, identical line at the other.
1396          * We are free to choose which identical line is included.
1397          * `compareseq' usually chooses the one at the beginning,
1398          * but usually it is cleaner to consider the following identical line
1399          * to be the "change".
1400          *
1401          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1402          */
1403         function _shift_boundaries ($lines, &$changed, $other_changed) {
1404                 wfProfileIn( __METHOD__ );
1405                 $i = 0;
1406                 $j = 0;
1407
1408                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1409                 $len = sizeof($lines);
1410                 $other_len = sizeof($other_changed);
1411
1412                 while (1) {
1413                         /*
1414                          * Scan forwards to find beginning of another run of changes.
1415                          * Also keep track of the corresponding point in the other file.
1416                          *
1417                          * Throughout this code, $i and $j are adjusted together so that
1418                          * the first $i elements of $changed and the first $j elements
1419                          * of $other_changed both contain the same number of zeros
1420                          * (unchanged lines).
1421                          * Furthermore, $j is always kept so that $j == $other_len or
1422                          * $other_changed[$j] == false.
1423                          */
1424                         while ($j < $other_len && $other_changed[$j])
1425                         $j++;
1426
1427                         while ($i < $len && ! $changed[$i]) {
1428                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1429                                 $i++; $j++;
1430                                 while ($j < $other_len && $other_changed[$j])
1431                                 $j++;
1432                         }
1433
1434                         if ($i == $len)
1435                         break;
1436
1437                         $start = $i;
1438
1439                         // Find the end of this run of changes.
1440                         while (++$i < $len && $changed[$i])
1441                         continue;
1442
1443                         do {
1444                                 /*
1445                                  * Record the length of this run of changes, so that
1446                                  * we can later determine whether the run has grown.
1447                                  */
1448                                 $runlength = $i - $start;
1449
1450                                 /*
1451                                  * Move the changed region back, so long as the
1452                                  * previous unchanged line matches the last changed one.
1453                                  * This merges with previous changed regions.
1454                                  */
1455                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1456                                         $changed[--$start] = 1;
1457                                         $changed[--$i] = false;
1458                                         while ($start > 0 && $changed[$start - 1])
1459                                         $start--;
1460                                         USE_ASSERTS && assert('$j > 0');
1461                                         while ($other_changed[--$j])
1462                                         continue;
1463                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1464                                 }
1465
1466                                 /*
1467                                  * Set CORRESPONDING to the end of the changed run, at the last
1468                                  * point where it corresponds to a changed run in the other file.
1469                                  * CORRESPONDING == LEN means no such point has been found.
1470                                  */
1471                                 $corresponding = $j < $other_len ? $i : $len;
1472
1473                                 /*
1474                                  * Move the changed region forward, so long as the
1475                                  * first changed line matches the following unchanged one.
1476                                  * This merges with following changed regions.
1477                                  * Do this second, so that if there are no merges,
1478                                  * the changed region is moved forward as far as possible.
1479                                  */
1480                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1481                                         $changed[$start++] = false;
1482                                         $changed[$i++] = 1;
1483                                         while ($i < $len && $changed[$i])
1484                                         $i++;
1485
1486                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1487                                         $j++;
1488                                         if ($j < $other_len && $other_changed[$j]) {
1489                                                 $corresponding = $i;
1490                                                 while ($j < $other_len && $other_changed[$j])
1491                                                 $j++;
1492                                         }
1493                                 }
1494                         } while ($runlength != $i - $start);
1495
1496                         /*
1497                          * If possible, move the fully-merged run of changes
1498                          * back to a corresponding run in the other file.
1499                          */
1500                         while ($corresponding < $i) {
1501                                 $changed[--$start] = 1;
1502                                 $changed[--$i] = 0;
1503                                 USE_ASSERTS && assert('$j > 0');
1504                                 while ($other_changed[--$j])
1505                                 continue;
1506                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1507                         }
1508                 }
1509                 wfProfileOut( __METHOD__ );
1510         }
1511 }
1512
1513 /**
1514  * Class representing a 'diff' between two sequences of strings.
1515  * @todo document
1516  * @private
1517  * @ingroup DifferenceEngine
1518  */
1519 class Diff
1520 {
1521         var $edits;
1522
1523         /**
1524          * Constructor.
1525          * Computes diff between sequences of strings.
1526          *
1527          * @param $from_lines array An array of strings.
1528          *                (Typically these are lines from a file.)
1529          * @param $to_lines array An array of strings.
1530          */
1531         function Diff($from_lines, $to_lines) {
1532                 $eng = new _DiffEngine;
1533                 $this->edits = $eng->diff($from_lines, $to_lines);
1534                 //$this->_check($from_lines, $to_lines);
1535         }
1536
1537         /**
1538          * Compute reversed Diff.
1539          *
1540          * SYNOPSIS:
1541          *
1542          *      $diff = new Diff($lines1, $lines2);
1543          *      $rev = $diff->reverse();
1544          * @return object A Diff object representing the inverse of the
1545          *                                original diff.
1546          */
1547         function reverse () {
1548                 $rev = $this;
1549                 $rev->edits = array();
1550                 foreach ($this->edits as $edit) {
1551                         $rev->edits[] = $edit->reverse();
1552                 }
1553                 return $rev;
1554         }
1555
1556         /**
1557          * Check for empty diff.
1558          *
1559          * @return bool True iff two sequences were identical.
1560          */
1561         function isEmpty () {
1562                 foreach ($this->edits as $edit) {
1563                         if ($edit->type != 'copy')
1564                         return false;
1565                 }
1566                 return true;
1567         }
1568
1569         /**
1570          * Compute the length of the Longest Common Subsequence (LCS).
1571          *
1572          * This is mostly for diagnostic purposed.
1573          *
1574          * @return int The length of the LCS.
1575          */
1576         function lcs () {
1577                 $lcs = 0;
1578                 foreach ($this->edits as $edit) {
1579                         if ($edit->type == 'copy')
1580                         $lcs += sizeof($edit->orig);
1581                 }
1582                 return $lcs;
1583         }
1584
1585         /**
1586          * Get the original set of lines.
1587          *
1588          * This reconstructs the $from_lines parameter passed to the
1589          * constructor.
1590          *
1591          * @return array The original sequence of strings.
1592          */
1593         function orig() {
1594                 $lines = array();
1595
1596                 foreach ($this->edits as $edit) {
1597                         if ($edit->orig)
1598                         array_splice($lines, sizeof($lines), 0, $edit->orig);
1599                 }
1600                 return $lines;
1601         }
1602
1603         /**
1604          * Get the closing set of lines.
1605          *
1606          * This reconstructs the $to_lines parameter passed to the
1607          * constructor.
1608          *
1609          * @return array The sequence of strings.
1610          */
1611         function closing() {
1612                 $lines = array();
1613
1614                 foreach ($this->edits as $edit) {
1615                         if ($edit->closing)
1616                         array_splice($lines, sizeof($lines), 0, $edit->closing);
1617                 }
1618                 return $lines;
1619         }
1620
1621         /**
1622          * Check a Diff for validity.
1623          *
1624          * This is here only for debugging purposes.
1625          */
1626         function _check ($from_lines, $to_lines) {
1627                 wfProfileIn( __METHOD__ );
1628                 if (serialize($from_lines) != serialize($this->orig()))
1629                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1630                 if (serialize($to_lines) != serialize($this->closing()))
1631                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1632
1633                 $rev = $this->reverse();
1634                 if (serialize($to_lines) != serialize($rev->orig()))
1635                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
1636                 if (serialize($from_lines) != serialize($rev->closing()))
1637                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1638
1639
1640                 $prevtype = 'none';
1641                 foreach ($this->edits as $edit) {
1642                         if ( $prevtype == $edit->type )
1643                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1644                         $prevtype = $edit->type;
1645                 }
1646
1647                 $lcs = $this->lcs();
1648                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1649                 wfProfileOut( __METHOD__ );
1650         }
1651 }
1652
1653 /**
1654  * @todo document, bad name.
1655  * @private
1656  * @ingroup DifferenceEngine
1657  */
1658 class MappedDiff extends Diff
1659 {
1660         /**
1661          * Constructor.
1662          *
1663          * Computes diff between sequences of strings.
1664          *
1665          * This can be used to compute things like
1666          * case-insensitve diffs, or diffs which ignore
1667          * changes in white-space.
1668          *
1669          * @param $from_lines array An array of strings.
1670          *      (Typically these are lines from a file.)
1671          *
1672          * @param $to_lines array An array of strings.
1673          *
1674          * @param $mapped_from_lines array This array should
1675          *      have the same size number of elements as $from_lines.
1676          *      The elements in $mapped_from_lines and
1677          *      $mapped_to_lines are what is actually compared
1678          *      when computing the diff.
1679          *
1680          * @param $mapped_to_lines array This array should
1681          *      have the same number of elements as $to_lines.
1682          */
1683         function MappedDiff($from_lines, $to_lines,
1684         $mapped_from_lines, $mapped_to_lines) {
1685                 wfProfileIn( __METHOD__ );
1686
1687                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1688                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1689
1690                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1691
1692                 $xi = $yi = 0;
1693                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1694                         $orig = &$this->edits[$i]->orig;
1695                         if (is_array($orig)) {
1696                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1697                                 $xi += sizeof($orig);
1698                         }
1699
1700                         $closing = &$this->edits[$i]->closing;
1701                         if (is_array($closing)) {
1702                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1703                                 $yi += sizeof($closing);
1704                         }
1705                 }
1706                 wfProfileOut( __METHOD__ );
1707         }
1708 }
1709
1710 /**
1711  * A class to format Diffs
1712  *
1713  * This class formats the diff in classic diff format.
1714  * It is intended that this class be customized via inheritance,
1715  * to obtain fancier outputs.
1716  * @todo document
1717  * @private
1718  * @ingroup DifferenceEngine
1719  */
1720 class DiffFormatter {
1721         /**
1722          * Number of leading context "lines" to preserve.
1723          *
1724          * This should be left at zero for this class, but subclasses
1725          * may want to set this to other values.
1726          */
1727         var $leading_context_lines = 0;
1728
1729         /**
1730          * Number of trailing context "lines" to preserve.
1731          *
1732          * This should be left at zero for this class, but subclasses
1733          * may want to set this to other values.
1734          */
1735         var $trailing_context_lines = 0;
1736
1737         /**
1738          * Format a diff.
1739          *
1740          * @param $diff object A Diff object.
1741          * @return string The formatted output.
1742          */
1743         function format($diff) {
1744                 wfProfileIn( __METHOD__ );
1745
1746                 $xi = $yi = 1;
1747                 $block = false;
1748                 $context = array();
1749
1750                 $nlead = $this->leading_context_lines;
1751                 $ntrail = $this->trailing_context_lines;
1752
1753                 $this->_start_diff();
1754
1755                 foreach ($diff->edits as $edit) {
1756                         if ($edit->type == 'copy') {
1757                                 if (is_array($block)) {
1758                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1759                                                 $block[] = $edit;
1760                                         }
1761                                         else{
1762                                                 if ($ntrail) {
1763                                                         $context = array_slice($edit->orig, 0, $ntrail);
1764                                                         $block[] = new _DiffOp_Copy($context);
1765                                                 }
1766                                                 $this->_block($x0, $ntrail + $xi - $x0,
1767                                                 $y0, $ntrail + $yi - $y0,
1768                                                 $block);
1769                                                 $block = false;
1770                                         }
1771                                 }
1772                                 $context = $edit->orig;
1773                         }
1774                         else {
1775                                 if (! is_array($block)) {
1776                                         $context = array_slice($context, sizeof($context) - $nlead);
1777                                         $x0 = $xi - sizeof($context);
1778                                         $y0 = $yi - sizeof($context);
1779                                         $block = array();
1780                                         if ($context)
1781                                         $block[] = new _DiffOp_Copy($context);
1782                                 }
1783                                 $block[] = $edit;
1784                         }
1785
1786                         if ($edit->orig)
1787                         $xi += sizeof($edit->orig);
1788                         if ($edit->closing)
1789                         $yi += sizeof($edit->closing);
1790                 }
1791
1792                 if (is_array($block))
1793                 $this->_block($x0, $xi - $x0,
1794                 $y0, $yi - $y0,
1795                 $block);
1796
1797                 $end = $this->_end_diff();
1798                 wfProfileOut( __METHOD__ );
1799                 return $end;
1800         }
1801
1802         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1803                 wfProfileIn( __METHOD__ );
1804                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1805                 foreach ($edits as $edit) {
1806                         if ($edit->type == 'copy')
1807                         $this->_context($edit->orig);
1808                         elseif ($edit->type == 'add')
1809                         $this->_added($edit->closing);
1810                         elseif ($edit->type == 'delete')
1811                         $this->_deleted($edit->orig);
1812                         elseif ($edit->type == 'change')
1813                         $this->_changed($edit->orig, $edit->closing);
1814                         else
1815                         trigger_error('Unknown edit type', E_USER_ERROR);
1816                 }
1817                 $this->_end_block();
1818                 wfProfileOut( __METHOD__ );
1819         }
1820
1821         function _start_diff() {
1822                 ob_start();
1823         }
1824
1825         function _end_diff() {
1826                 $val = ob_get_contents();
1827                 ob_end_clean();
1828                 return $val;
1829         }
1830
1831         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1832                 if ($xlen > 1)
1833                 $xbeg .= "," . ($xbeg + $xlen - 1);
1834                 if ($ylen > 1)
1835                 $ybeg .= "," . ($ybeg + $ylen - 1);
1836
1837                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1838         }
1839
1840         function _start_block($header) {
1841                 echo $header . "\n";
1842         }
1843
1844         function _end_block() {
1845         }
1846
1847         function _lines($lines, $prefix = ' ') {
1848                 foreach ($lines as $line)
1849                 echo "$prefix $line\n";
1850         }
1851
1852         function _context($lines) {
1853                 $this->_lines($lines);
1854         }
1855
1856         function _added($lines) {
1857                 $this->_lines($lines, '>');
1858         }
1859         function _deleted($lines) {
1860                 $this->_lines($lines, '<');
1861         }
1862
1863         function _changed($orig, $closing) {
1864                 $this->_deleted($orig);
1865                 echo "---\n";
1866                 $this->_added($closing);
1867         }
1868 }
1869
1870 /**
1871  * A formatter that outputs unified diffs
1872  * @ingroup DifferenceEngine
1873  */
1874
1875 class UnifiedDiffFormatter extends DiffFormatter {
1876         var $leading_context_lines = 2;
1877         var $trailing_context_lines = 2;
1878
1879         function _added($lines) {
1880                 $this->_lines($lines, '+');
1881         }
1882         function _deleted($lines) {
1883                 $this->_lines($lines, '-');
1884         }
1885         function _changed($orig, $closing) {
1886                 $this->_deleted($orig);
1887                 $this->_added($closing);
1888         }
1889         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1890                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1891         }
1892 }
1893
1894 /**
1895  * A pseudo-formatter that just passes along the Diff::$edits array
1896  * @ingroup DifferenceEngine
1897  */
1898 class ArrayDiffFormatter extends DiffFormatter {
1899         function format($diff) {
1900                 $oldline = 1;
1901                 $newline = 1;
1902                 $retval = array();
1903                 foreach($diff->edits as $edit)
1904                 switch($edit->type) {
1905                         case 'add':
1906                                 foreach($edit->closing as $l) {
1907                                         $retval[] = array(
1908                                                         'action' => 'add',
1909                                                         'new'=> $l,
1910                                                         'newline' => $newline++
1911                                         );
1912                                 }
1913                                 break;
1914                         case 'delete':
1915                                 foreach($edit->orig as $l) {
1916                                         $retval[] = array(
1917                                                         'action' => 'delete',
1918                                                         'old' => $l,
1919                                                         'oldline' => $oldline++,
1920                                         );
1921                                 }
1922                                 break;
1923                         case 'change':
1924                                 foreach($edit->orig as $i => $l) {
1925                                         $retval[] = array(
1926                                                         'action' => 'change',
1927                                                         'old' => $l,
1928                                                         'new' => @$edit->closing[$i],
1929                                                         'oldline' => $oldline++,
1930                                                         'newline' => $newline++,
1931                                         );
1932                                 }
1933                                 break;
1934                         case 'copy':
1935                                 $oldline += count($edit->orig);
1936                                 $newline += count($edit->orig);
1937                 }
1938                 return $retval;
1939         }
1940 }
1941
1942 /**
1943  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1944  *
1945  */
1946
1947 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1948
1949 /**
1950  * @todo document
1951  * @private
1952  * @ingroup DifferenceEngine
1953  */
1954 class _HWLDF_WordAccumulator {
1955         function _HWLDF_WordAccumulator () {
1956                 $this->_lines = array();
1957                 $this->_line = '';
1958                 $this->_group = '';
1959                 $this->_tag = '';
1960         }
1961
1962         function _flushGroup ($new_tag) {
1963                 if ($this->_group !== '') {
1964                         if ($this->_tag == 'ins')
1965                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1966                         htmlspecialchars ( $this->_group ) . '</ins>';
1967                         elseif ($this->_tag == 'del')
1968                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1969                         htmlspecialchars ( $this->_group ) . '</del>';
1970                         else
1971                         $this->_line .= htmlspecialchars ( $this->_group );
1972                 }
1973                 $this->_group = '';
1974                 $this->_tag = $new_tag;
1975         }
1976
1977         function _flushLine ($new_tag) {
1978                 $this->_flushGroup($new_tag);
1979                 if ($this->_line != '')
1980                 array_push ( $this->_lines, $this->_line );
1981                 else
1982                 # make empty lines visible by inserting an NBSP
1983                 array_push ( $this->_lines, NBSP );
1984                 $this->_line = '';
1985         }
1986
1987         function addWords ($words, $tag = '') {
1988                 if ($tag != $this->_tag)
1989                 $this->_flushGroup($tag);
1990
1991                 foreach ($words as $word) {
1992                         // new-line should only come as first char of word.
1993                         if ($word == '')
1994                         continue;
1995                         if ($word[0] == "\n") {
1996                                 $this->_flushLine($tag);
1997                                 $word = substr($word, 1);
1998                         }
1999                         assert(!strstr($word, "\n"));
2000                         $this->_group .= $word;
2001                 }
2002         }
2003
2004         function getLines() {
2005                 $this->_flushLine('~done');
2006                 return $this->_lines;
2007         }
2008 }
2009
2010 /**
2011  * @todo document
2012  * @private
2013  * @ingroup DifferenceEngine
2014  */
2015 class WordLevelDiff extends MappedDiff {
2016         const MAX_LINE_LENGTH = 10000;
2017
2018         function WordLevelDiff ($orig_lines, $closing_lines) {
2019                 wfProfileIn( __METHOD__ );
2020
2021                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
2022                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
2023
2024                 $this->MappedDiff($orig_words, $closing_words,
2025                 $orig_stripped, $closing_stripped);
2026                 wfProfileOut( __METHOD__ );
2027         }
2028
2029         function _split($lines) {
2030                 wfProfileIn( __METHOD__ );
2031
2032                 $words = array();
2033                 $stripped = array();
2034                 $first = true;
2035                 foreach ( $lines as $line ) {
2036                         # If the line is too long, just pretend the entire line is one big word
2037                         # This prevents resource exhaustion problems
2038                         if ( $first ) {
2039                                 $first = false;
2040                         } else {
2041                                 $words[] = "\n";
2042                                 $stripped[] = "\n";
2043                         }
2044                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
2045                                 $words[] = $line;
2046                                 $stripped[] = $line;
2047                         } else {
2048                                 $m = array();
2049                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
2050                                 $line, $m))
2051                                 {
2052                                         $words = array_merge( $words, $m[0] );
2053                                         $stripped = array_merge( $stripped, $m[1] );
2054                                 }
2055                         }
2056                 }
2057                 wfProfileOut( __METHOD__ );
2058                 return array($words, $stripped);
2059         }
2060
2061         function orig () {
2062                 wfProfileIn( __METHOD__ );
2063                 $orig = new _HWLDF_WordAccumulator;
2064
2065                 foreach ($this->edits as $edit) {
2066                         if ($edit->type == 'copy')
2067                         $orig->addWords($edit->orig);
2068                         elseif ($edit->orig)
2069                         $orig->addWords($edit->orig, 'del');
2070                 }
2071                 $lines = $orig->getLines();
2072                 wfProfileOut( __METHOD__ );
2073                 return $lines;
2074         }
2075
2076         function closing () {
2077                 wfProfileIn( __METHOD__ );
2078                 $closing = new _HWLDF_WordAccumulator;
2079
2080                 foreach ($this->edits as $edit) {
2081                         if ($edit->type == 'copy')
2082                         $closing->addWords($edit->closing);
2083                         elseif ($edit->closing)
2084                         $closing->addWords($edit->closing, 'ins');
2085                 }
2086                 $lines = $closing->getLines();
2087                 wfProfileOut( __METHOD__ );
2088                 return $lines;
2089         }
2090 }
2091
2092 /**
2093  * Wikipedia Table style diff formatter.
2094  * @todo document
2095  * @private
2096  * @ingroup DifferenceEngine
2097  */
2098 class TableDiffFormatter extends DiffFormatter {
2099         function TableDiffFormatter() {
2100                 $this->leading_context_lines = 2;
2101                 $this->trailing_context_lines = 2;
2102         }
2103
2104         public static function escapeWhiteSpace( $msg ) {
2105                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
2106                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
2107                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
2108                 return $msg;
2109         }
2110
2111         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
2112                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
2113                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
2114                 return $r;
2115         }
2116
2117         function _start_block( $header ) {
2118                 echo $header;
2119         }
2120
2121         function _end_block() {
2122         }
2123
2124         function _lines( $lines, $prefix=' ', $color='white' ) {
2125         }
2126
2127         # HTML-escape parameter before calling this
2128         function addedLine( $line ) {
2129                 return $this->wrapLine( '+', 'diff-addedline', $line );
2130         }
2131
2132         # HTML-escape parameter before calling this
2133         function deletedLine( $line ) {
2134                 return $this->wrapLine( '-', 'diff-deletedline', $line );
2135         }
2136
2137         # HTML-escape parameter before calling this
2138         function contextLine( $line ) {
2139                 return $this->wrapLine( ' ', 'diff-context', $line );
2140         }
2141
2142         private function wrapLine( $marker, $class, $line ) {
2143                 if( $line !== '' ) {
2144                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
2145                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
2146                 }
2147                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
2148         }
2149
2150         function emptyLine() {
2151                 return '<td colspan="2">&nbsp;</td>';
2152         }
2153
2154         function _added( $lines ) {
2155                 foreach ($lines as $line) {
2156                         echo '<tr>' . $this->emptyLine() .
2157                         $this->addedLine( '<ins class="diffchange">' .
2158                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
2159                 }
2160         }
2161
2162         function _deleted($lines) {
2163                 foreach ($lines as $line) {
2164                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
2165                         htmlspecialchars ( $line ) . '</del>' ) .
2166                         $this->emptyLine() . "</tr>\n";
2167                 }
2168         }
2169
2170         function _context( $lines ) {
2171                 foreach ($lines as $line) {
2172                         echo '<tr>' .
2173                         $this->contextLine( htmlspecialchars ( $line ) ) .
2174                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
2175                 }
2176         }
2177
2178         function _changed( $orig, $closing ) {
2179                 wfProfileIn( __METHOD__ );
2180
2181                 $diff = new WordLevelDiff( $orig, $closing );
2182                 $del = $diff->orig();
2183                 $add = $diff->closing();
2184
2185                 # Notice that WordLevelDiff returns HTML-escaped output.
2186                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2187
2188                 while ( $line = array_shift( $del ) ) {
2189                         $aline = array_shift( $add );
2190                         echo '<tr>' . $this->deletedLine( $line ) .
2191                         $this->addedLine( $aline ) . "</tr>\n";
2192                 }
2193                 foreach ($add as $line) {       # If any leftovers
2194                         echo '<tr>' . $this->emptyLine() .
2195                         $this->addedLine( $line ) . "</tr>\n";
2196                 }
2197                 wfProfileOut( __METHOD__ );
2198         }
2199 }