includes/diff/DifferenceEngine.php

   1 <?php
   2 /**
   3  * @defgroup DifferenceEngine DifferenceEngine
   4  */
   5
   6 /**
   7  * Constant to indicate diff cache compatibility.
   8  * Bump this when changing the diff formatting in a way that
   9  * fixes important bugs or such to force cached diff views to
  10  * clear.
  11  */
  12 define( 'MW_DIFF_VERSION', '1.11a' );
  13
  14 /**
  15  * @todo document
  16  * @ingroup DifferenceEngine
  17  */
  18 class DifferenceEngine {
  19         /**#@+
  20          * @private
  21          */
  22         var $mOldid, $mNewid, $mTitle;
  23         var $mOldtitle, $mNewtitle, $mPagetitle;
  24         var $mOldtext, $mNewtext;
  25         var $mOldPage, $mNewPage;
  26         var $mRcidMarkPatrolled;
  27         var $mOldRev, $mNewRev;
  28         var $mRevisionsLoaded = false; // Have the revisions been loaded
  29         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  30         var $mCacheHit = false; // Was the diff fetched from cache?
  31         var $htmldiff;
  32
  33         protected $unhide = false;
  34         /**#@-*/
  35
  36         /**
  37          * Constructor
  38          * @param $titleObj Title object that the diff is associated with
  39          * @param $old Integer: old ID we want to show and diff with.
  40          * @param $new String: either 'prev' or 'next'.
  41          * @param $rcid Integer: ??? FIXME (default 0)
  42          * @param $refreshCache boolean If set, refreshes the diff cache
  43          * @param $htmldiff boolean If set, output using HTMLDiff instead of raw wikicode diff
  44          * @param $unhide boolean If set, allow viewing deleted revs
  45          */
  46         function __construct( $titleObj = null, $old = 0, $new = 0, $rcid = 0,
  47                 $refreshCache = false, $htmldiff = false, $unhide = false )
  48         {
  49                 $this->mTitle = $titleObj;
  50                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  51
  52                 if ( 'prev' === $new ) {
  53                         # Show diff between revision $old and the previous one.
  54                         # Get previous one from DB.
  55                         $this->mNewid = intval($old);
  56                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  57                 } elseif ( 'next' === $new ) {
  58                         # Show diff between revision $old and the next one.
  59                         # Get next one from DB.
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  62                         if ( false === $this->mNewid ) {
  63                                 # if no result, NewId points to the newest old revision. The only newer
  64                                 # revision is cur, which is "0".
  65                                 $this->mNewid = 0;
  66                         }
  67                 } else {
  68                         $this->mOldid = intval($old);
  69                         $this->mNewid = intval($new);
  70                         wfRunHooks( 'NewDifferenceEngine', array(&$titleObj, &$this->mOldid, &$this->mNewid, $old, $new) );
  71                 }
  72                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  73                 $this->mRefreshCache = $refreshCache;
  74                 $this->htmldiff = $htmldiff;
  75                 $this->unhide = $unhide;
  76         }
  77
  78         function getTitle() {
  79                 return $this->mTitle;
  80         }
  81
  82         function wasCacheHit() {
  83                 return $this->mCacheHit;
  84         }
  85
  86         function getOldid() {
  87                 return $this->mOldid;
  88         }
  89
  90         function getNewid() {
  91                 return $this->mNewid;
  92         }
  93
  94         function showDiffPage( $diffOnly = false ) {
  95                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol, $wgEnableHtmlDiff;
  96                 wfProfileIn( __METHOD__ );
  97
  98
  99                 # If external diffs are enabled both globally and for the user,
 100                 # we'll use the application/x-external-editor interface to call
 101                 # an external diff tool like kompare, kdiff3, etc.
 102                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
 103                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
 104                         $wgOut->disable();
 105                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
 106                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
 107                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
 108                         $special=$wgLang->getNsText(NS_SPECIAL);
 109                         $control=<<<CONTROL
 110                         [Process]
 111                         Type=Diff text
 112                         Engine=MediaWiki
 113                         Script={$wgServer}{$wgScript}
 114                         Special namespace={$special}
 115
 116                         [File]
 117                         Extension=wiki
 118                         URL=$url1
 119
 120                         [File 2]
 121                         Extension=wiki
 122                         URL=$url2
 123 CONTROL;
 124                         echo($control);
 125                         return;
 126                 }
 127
 128                 $wgOut->setArticleFlag( false );
 129                 if ( !$this->loadRevisionData() ) {
 130                         $t = $this->mTitle->getPrefixedText();
 131                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 132                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 133                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 134                         wfProfileOut( __METHOD__ );
 135                         return;
 136                 }
 137
 138                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 139
 140                 if ( $this->mNewRev->isCurrent() ) {
 141                         $wgOut->setArticleFlag( true );
 142                 }
 143
 144                 # mOldid is false if the difference engine is called with a "vague" query for
 145                 # a diff between a version V and its previous version V' AND the version V
 146                 # is the first version of that article. In that case, V' does not exist.
 147                 if ( $this->mOldid === false ) {
 148                         $this->showFirstRevision();
 149                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 150                         wfProfileOut( __METHOD__ );
 151                         return;
 152                 }
 153
 154                 $wgOut->suppressQuickbar();
 155
 156                 $oldTitle = $this->mOldPage->getPrefixedText();
 157                 $newTitle = $this->mNewPage->getPrefixedText();
 158                 if( $oldTitle == $newTitle ) {
 159                         $wgOut->setPageTitle( $newTitle );
 160                 } else {
 161                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 162                 }
 163                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 164                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 165
 166                 if ( !$this->mOldPage->userCanRead() || !$this->mNewPage->userCanRead() ) {
 167                         $wgOut->loginToUse();
 168                         $wgOut->output();
 169                         $wgOut->disable();
 170                         wfProfileOut( __METHOD__ );
 171                         return;
 172                 }
 173
 174                 $sk = $wgUser->getSkin();
 175
 176                 // Check if page is editable
 177                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 178                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed( 'rollback' ) ) {
 179                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 180                 } else {
 181                         $rollback = '';
 182                 }
 183
 184                 // Prepare a change patrol link, if applicable
 185                 if( $wgUseRCPatrol && $this->mTitle->userCan('patrol') ) {
 186                         // If we've been given an explicit change identifier, use it; saves time
 187                         if( $this->mRcidMarkPatrolled ) {
 188                                 $rcid = $this->mRcidMarkPatrolled;
 189                                 $rc = RecentChange::newFromId( $rcid );
 190                                 // Already patrolled?
 191                                 $rcid = is_object($rc) && !$rc->getAttribute('rc_patrolled') ? $rcid : 0;
 192                         } else {
 193                                 // Look for an unpatrolled change corresponding to this diff
 194                                 $db = wfGetDB( DB_SLAVE );
 195                                 $change = RecentChange::newFromConds(
 196                                         array(
 197                                         // Redundant user,timestamp condition so we can use the existing index
 198                                                 'rc_user_text'  => $this->mNewRev->getRawUserText(),
 199                                                 'rc_timestamp'  => $db->timestamp( $this->mNewRev->getTimestamp() ),
 200                                                 'rc_this_oldid' => $this->mNewid,
 201                                                 'rc_last_oldid' => $this->mOldid,
 202                                                 'rc_patrolled'  => 0
 203                                         ),
 204                                         __METHOD__
 205                                 );
 206                                 if( $change instanceof RecentChange ) {
 207                                         $rcid = $change->mAttribs['rc_id'];
 208                                         $this->mRcidMarkPatrolled = $rcid;
 209                                 } else {
 210                                         // None found
 211                                         $rcid = 0;
 212                                 }
 213                         }
 214                         // Build the link
 215                         if( $rcid ) {
 216                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj( $this->mTitle,
 217                                         wfMsgHtml( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$rcid}" ) . ']</span>';
 218                         } else {
 219                                 $patrol = '';
 220                         }
 221                 } else {
 222                         $patrol = '';
 223                 }
 224
 225                 $diffOnlyArg = '';
 226                 # Carry over 'diffonly' param via navigation links
 227                 if( $diffOnly != $wgUser->getBoolOption('diffonly') ) {
 228                         $diffOnlyArg = '&diffonly='.$diffOnly;
 229                 }
 230                 $htmldiffarg = $this->htmlDiffArgument();
 231                 # Make "previous revision link"
 232                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 233                         "diff=prev&oldid={$this->mOldid}{$htmldiffarg}{$diffOnlyArg}", '', '', 'id="differences-prevlink"' );
 234                 # Make "next revision link"
 235                 if( $this->mNewRev->isCurrent() ) {
 236                         $nextlink = '&nbsp;';
 237                 } else {
 238                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 239                                 "diff=next&oldid={$this->mNewid}{$htmldiffarg}{$diffOnlyArg}", '', '', 'id="differences-nextlink"' );
 240                 }
 241
 242                 $oldminor = '';
 243                 $newminor = '';
 244
 245                 if( $this->mOldRev->isMinor() ) {
 246                         $oldminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 247                 }
 248                 if( $this->mNewRev->isMinor() ) {
 249                         $newminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 250                 }
 251
 252                 $rdel = ''; $ldel = '';
 253                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 254                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 255                                 // If revision was hidden from sysops
 256                                 $ldel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 257                         } else {
 258                                 $query = array( 'target' => $this->mOldRev->mTitle->getPrefixedDbkey(),
 259                                         'oldid' => $this->mOldRev->getId()
 260                                 );
 261                                 $ldel = $sk->revDeleteLink( $query, $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 262                         }
 263                         $ldel = "&nbsp;&nbsp;&nbsp;$ldel ";
 264                         // We don't currently handle well changing the top revision's settings
 265                         if( $this->mNewRev->isCurrent() ) {
 266                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 267                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 268                                 // If revision was hidden from sysops
 269                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 270                         } else {
 271                                 $query = array( 'target' =>  $this->mNewRev->mTitle->getPrefixedDbkey(),
 272                                         'oldid' => $this->mNewRev->getId()
 273                                 );
 274                                 $rdel = $sk->revDeleteLink( $query, $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 275                         }
 276                         $rdel = "&nbsp;&nbsp;&nbsp;$rdel ";
 277                 }
 278
 279                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 280                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, !$this->unhide ) . "</div>" .
 281                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, !$this->unhide ).$ldel."</div>" .
 282                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 283                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 284                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, !$this->unhide ) . " $rollback</div>" .
 285                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, !$this->unhide ).$rdel."</div>" .
 286                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 287
 288                 # Check if this user can see the revisions
 289                 $allowed = $this->mOldRev->userCan(Revision::DELETED_TEXT)
 290                         && $this->mNewRev->userCan(Revision::DELETED_TEXT);
 291                 $deleted = $this->mOldRev->isDeleted(Revision::DELETED_TEXT)
 292                         || $this->mNewRev->isDeleted(Revision::DELETED_TEXT);
 293                 # Output the diff if allowed...
 294                 if( $deleted && (!$this->unhide || !$allowed) ) {
 295                         $this->showDiffStyle();
 296                         $multi = $this->getMultiNotice();
 297                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 298                         if( !$allowed ) {
 299                                 # Give explanation for why revision is not visible
 300                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 301                                         array( 'rev-deleted-no-diff' ) );
 302                         } else {
 303                                 # Give explanation and add a link to view the diff...
 304                                 $link = $this->mTitle->getFullUrl( "diff={$this->mNewid}&oldid={$this->mOldid}".
 305                                         '&unhide=1&token='.urlencode( $wgUser->editToken($this->mNewid) ) );
 306                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 307                                         array( 'rev-deleted-unhide-diff', $link ) );
 308                         }
 309                 } else if( $wgEnableHtmlDiff && $this->htmldiff ) {
 310                         $multi = $this->getMultiNotice();
 311                         $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'wikicodecomparison' ),
 312                                 'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=0', '', '', 'id="differences-switchtype"' ).'</div>');
 313                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 314                         $this->renderHtmlDiff();
 315                 } else {
 316                         if( $wgEnableHtmlDiff ) {
 317                                 $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'visualcomparison' ),
 318                                         'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=1', '', '', 'id="differences-switchtype"' ).'</div>');
 319                         }
 320                         $this->showDiff( $oldHeader, $newHeader );
 321                         if( !$diffOnly ) {
 322                                 $this->renderNewRevision();
 323                         }
 324                 }
 325                 wfProfileOut( __METHOD__ );
 326         }
 327
 328         /**
 329          * Show the new revision of the page.
 330          */
 331         function renderNewRevision() {
 332                 global $wgOut, $wgUser;
 333                 wfProfileIn( __METHOD__ );
 334
 335                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 336                 # Add deleted rev tag if needed
 337                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 338                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 339                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 340                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 341                 }
 342
 343                 if( !$this->mNewRev->isCurrent() ) {
 344                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 345                 }
 346
 347                 $this->loadNewText();
 348                 if( is_object( $this->mNewRev ) ) {
 349                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 350                 }
 351
 352                 if( $this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage() ) {
 353                         // Stolen from Article::view --AG 2007-10-11
 354                         // Give hooks a chance to customise the output
 355                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 356                                 // Wrap the whole lot in a <pre> and don't parse
 357                                 $m = array();
 358                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 359                                 $wgOut->addHTML( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 360                                 $wgOut->addHTML( htmlspecialchars( $this->mNewtext ) );
 361                                 $wgOut->addHTML( "\n</pre>\n" );
 362                         }
 363                 } else {
 364                         $wgOut->addWikiTextTidy( $this->mNewtext );
 365                 }
 366
 367                 if( is_object( $this->mNewRev ) && !$this->mNewRev->isCurrent() ) {
 368                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 369                 }
 370                 # Add redundant patrol link on bottom...
 371                 if( $this->mRcidMarkPatrolled && $this->mTitle->quickUserCan('patrol') ) {
 372                         $sk = $wgUser->getSkin();
 373                         $wgOut->addHTML(
 374                                 "<div class='patrollink'>[" . $sk->makeKnownLinkObj( $this->mTitle,
 375                                         wfMsgHtml( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) .
 376                                 ']</div>'
 377                          );
 378                 }
 379
 380                 wfProfileOut( __METHOD__ );
 381         }
 382
 383
 384         function renderHtmlDiff() {
 385                 global $wgOut, $wgParser, $wgDebugComments;
 386                 wfProfileIn( __METHOD__ );
 387
 388                 $this->showDiffStyle();
 389
 390                 $wgOut->addHTML( '<h2>'.wfMsgHtml( 'visual-comparison' )."</h2>\n" );
 391                 #add deleted rev tag if needed
 392                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 393                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 394                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 395                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 396                 }
 397
 398                 if( !$this->mNewRev->isCurrent() ) {
 399                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 400                 }
 401
 402                 $this->loadText();
 403
 404                 // Old revision
 405                 if( is_object( $this->mOldRev ) ) {
 406                         $wgOut->setRevisionId( $this->mOldRev->getId() );
 407                 }
 408
 409                 $popts = $wgOut->parserOptions();
 410                 $oldTidy = $popts->setTidy( true );
 411                 $popts->setEditSection( false );
 412
 413                 $parserOutput = $wgParser->parse( $this->mOldtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 414                 $popts->setTidy( $oldTidy );
 415
 416                 //only for new?
 417                 //$wgOut->addParserOutputNoText( $parserOutput );
 418                 $oldHtml = $parserOutput->getText();
 419                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$oldHtml ) );
 420
 421                 // New revision
 422                 if( is_object( $this->mNewRev ) ) {
 423                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 424                 }
 425
 426                 $popts = $wgOut->parserOptions();
 427                 $oldTidy = $popts->setTidy( true );
 428
 429                 $parserOutput = $wgParser->parse( $this->mNewtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 430                 $popts->setTidy( $oldTidy );
 431
 432                 $wgOut->addParserOutputNoText( $parserOutput );
 433                 $newHtml = $parserOutput->getText();
 434                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$newHtml ) );
 435
 436                 unset($parserOutput, $popts);
 437
 438                 $differ = new HTMLDiffer(new DelegatingContentHandler($wgOut));
 439                 $differ->htmlDiff($oldHtml, $newHtml);
 440                 if ( $wgDebugComments ) {
 441                         $wgOut->addHTML( "\n<!-- HtmlDiff Debug Output:\n" . HTMLDiffer::getDebugOutput() . " End Debug -->" );
 442                 }
 443
 444                 wfProfileOut( __METHOD__ );
 445         }
 446
 447         /**
 448          * Show the first revision of an article. Uses normal diff headers in
 449          * contrast to normal "old revision" display style.
 450          */
 451         function showFirstRevision() {
 452                 global $wgOut, $wgUser;
 453                 wfProfileIn( __METHOD__ );
 454
 455                 # Get article text from the DB
 456                 #
 457                 if ( ! $this->loadNewText() ) {
 458                         $t = $this->mTitle->getPrefixedText();
 459                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 460                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 461                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 462                         wfProfileOut( __METHOD__ );
 463                         return;
 464                 }
 465                 if ( $this->mNewRev->isCurrent() ) {
 466                         $wgOut->setArticleFlag( true );
 467                 }
 468
 469                 # Check if user is allowed to look at this page. If not, bail out.
 470                 #
 471                 if ( !$this->mTitle->userCanRead() ) {
 472                         $wgOut->loginToUse();
 473                         $wgOut->output();
 474                         wfProfileOut( __METHOD__ );
 475                         throw new MWException("Permission Error: you do not have access to view this page");
 476                 }
 477
 478                 # Prepare the header box
 479                 #
 480                 $sk = $wgUser->getSkin();
 481
 482                 $next = $this->mTitle->getNextRevisionID( $this->mNewid );
 483                 if( !$next ) {
 484                         $nextlink = '';
 485                 } else {
 486                         $nextlink = '<br/>' . $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 487                                 'diff=next&oldid=' . $this->mNewid.$this->htmlDiffArgument(), '', '', 'id="differences-nextlink"' );
 488                 }
 489                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\">" .
 490                         $sk->revUserTools( $this->mNewRev ) . "<br/>" . $sk->revComment( $this->mNewRev ) . $nextlink . "</div>\n";
 491
 492                 $wgOut->addHTML( $header );
 493
 494                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 495                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 496
 497                 wfProfileOut( __METHOD__ );
 498         }
 499
 500         function htmlDiffArgument(){
 501                 global $wgEnableHtmlDiff;
 502                 if($wgEnableHtmlDiff){
 503                         if($this->htmldiff){
 504                                 return '&htmldiff=1';
 505                         }else{
 506                                 return '&htmldiff=0';
 507                         }
 508                 }else{
 509                         return '';
 510                 }
 511         }
 512
 513         /**
 514          * Get the diff text, send it to $wgOut
 515          * Returns false if the diff could not be generated, otherwise returns true
 516          */
 517         function showDiff( $otitle, $ntitle ) {
 518                 global $wgOut;
 519                 $diff = $this->getDiff( $otitle, $ntitle );
 520                 if ( $diff === false ) {
 521                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 522                         return false;
 523                 } else {
 524                         $this->showDiffStyle();
 525                         $wgOut->addHTML( $diff );
 526                         return true;
 527                 }
 528         }
 529
 530         /**
 531          * Add style sheets and supporting JS for diff display.
 532          */
 533         function showDiffStyle() {
 534                 global $wgStylePath, $wgStyleVersion, $wgOut;
 535                 $wgOut->addStyle( 'common/diff.css' );
 536
 537                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 538                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 539         }
 540
 541         /**
 542          * Get complete diff table, including header
 543          *
 544          * @param Title $otitle Old title
 545          * @param Title $ntitle New title
 546          * @return mixed
 547          */
 548         function getDiff( $otitle, $ntitle ) {
 549                 $body = $this->getDiffBody();
 550                 if ( $body === false ) {
 551                         return false;
 552                 } else {
 553                         $multi = $this->getMultiNotice();
 554                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 555                 }
 556         }
 557
 558         /**
 559          * Get the diff table body, without header
 560          *
 561          * @return mixed
 562          */
 563         function getDiffBody() {
 564                 global $wgMemc;
 565                 wfProfileIn( __METHOD__ );
 566                 $this->mCacheHit = true;
 567                 // Check if the diff should be hidden from this user
 568                 if ( !$this->loadRevisionData() )
 569                         return '';
 570                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 571                         return '';
 572                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 573                         return '';
 574                 } else if ( $this->mOldRev && $this->mNewRev && $this->mOldRev->getID() == $this->mNewRev->getID() ) {
 575                         return '';
 576                 }
 577                 // Cacheable?
 578                 $key = false;
 579                 if ( $this->mOldid && $this->mNewid ) {
 580                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 581                         // Try cache
 582                         if ( !$this->mRefreshCache ) {
 583                                 $difftext = $wgMemc->get( $key );
 584                                 if ( $difftext ) {
 585                                         wfIncrStats( 'diff_cache_hit' );
 586                                         $difftext = $this->localiseLineNumbers( $difftext );
 587                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 588                                         wfProfileOut( __METHOD__ );
 589                                         return $difftext;
 590                                 }
 591                         } // don't try to load but save the result
 592                 }
 593                 $this->mCacheHit = false;
 594
 595                 // Loadtext is permission safe, this just clears out the diff
 596                 if ( !$this->loadText() ) {
 597                         wfProfileOut( __METHOD__ );
 598                         return false;
 599                 }
 600
 601                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 602
 603                 // Save to cache for 7 days
 604                 if ( !wfRunHooks( 'AbortDiffCache', array( &$this ) ) ) {
 605                         wfIncrStats( 'diff_uncacheable' );
 606                 } else if ( $key !== false && $difftext !== false ) {
 607                         wfIncrStats( 'diff_cache_miss' );
 608                         $wgMemc->set( $key, $difftext, 7*86400 );
 609                 } else {
 610                         wfIncrStats( 'diff_uncacheable' );
 611                 }
 612                 // Replace line numbers with the text in the user's language
 613                 if ( $difftext !== false ) {
 614                         $difftext = $this->localiseLineNumbers( $difftext );
 615                 }
 616                 wfProfileOut( __METHOD__ );
 617                 return $difftext;
 618         }
 619
 620         /**
 621          * Make sure the proper modules are loaded before we try to
 622          * make the diff
 623          */
 624         private function initDiffEngines() {
 625                 global $wgExternalDiffEngine;
 626                 if ( $wgExternalDiffEngine == 'wikidiff' && !function_exists( 'wikidiff_do_diff' ) ) {
 627                         wfProfileIn( __METHOD__ . '-php_wikidiff.so' );
 628                         wfSuppressWarnings();
 629                         dl( 'php_wikidiff.so' );
 630                         wfRestoreWarnings();
 631                         wfProfileOut( __METHOD__ . '-php_wikidiff.so' );
 632                 }
 633                 else if ( $wgExternalDiffEngine == 'wikidiff2' && !function_exists( 'wikidiff2_do_diff' ) ) {
 634                         wfProfileIn( __METHOD__ . '-php_wikidiff2.so' );
 635                         wfSuppressWarnings();
 636                         dl( 'php_wikidiff2.so' );
 637                         wfRestoreWarnings();
 638                         wfProfileOut( __METHOD__ . '-php_wikidiff2.so' );
 639                 }
 640         }
 641
 642         /**
 643          * Generate a diff, no caching
 644          * $otext and $ntext must be already segmented
 645          */
 646         function generateDiffBody( $otext, $ntext ) {
 647                 global $wgExternalDiffEngine, $wgContLang;
 648
 649                 $otext = str_replace( "\r\n", "\n", $otext );
 650                 $ntext = str_replace( "\r\n", "\n", $ntext );
 651
 652                 $this->initDiffEngines();
 653
 654                 if ( $wgExternalDiffEngine == 'wikidiff' && function_exists( 'wikidiff_do_diff' ) ) {
 655                         # For historical reasons, external diff engine expects
 656                         # input text to be HTML-escaped already
 657                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 658                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 659                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 660                         $this->debug( 'wikidiff1' );
 661                 }
 662
 663                 if ( $wgExternalDiffEngine == 'wikidiff2' && function_exists( 'wikidiff2_do_diff' ) ) {
 664                         # Better external diff engine, the 2 may some day be dropped
 665                         # This one does the escaping and segmenting itself
 666                         wfProfileIn( 'wikidiff2_do_diff' );
 667                         $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 668                         $text .= $this->debug( 'wikidiff2' );
 669                         wfProfileOut( 'wikidiff2_do_diff' );
 670                         return $text;
 671                 }
 672                 if ( $wgExternalDiffEngine != 'wikidiff3' && $wgExternalDiffEngine !== false ) {
 673                         # Diff via the shell
 674                         global $wgTmpDirectory;
 675                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 676                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 677
 678                         $tempFile1 = fopen( $tempName1, "w" );
 679                         if ( !$tempFile1 ) {
 680                                 wfProfileOut( __METHOD__ );
 681                                 return false;
 682                         }
 683                         $tempFile2 = fopen( $tempName2, "w" );
 684                         if ( !$tempFile2 ) {
 685                                 wfProfileOut( __METHOD__ );
 686                                 return false;
 687                         }
 688                         fwrite( $tempFile1, $otext );
 689                         fwrite( $tempFile2, $ntext );
 690                         fclose( $tempFile1 );
 691                         fclose( $tempFile2 );
 692                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 693                         wfProfileIn( __METHOD__ . "-shellexec" );
 694                         $difftext = wfShellExec( $cmd );
 695                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 696                         wfProfileOut( __METHOD__ . "-shellexec" );
 697                         unlink( $tempName1 );
 698                         unlink( $tempName2 );
 699                         return $difftext;
 700                 }
 701
 702                 # Native PHP diff
 703                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 704                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 705                 $diffs = new Diff( $ota, $nta );
 706                 $formatter = new TableDiffFormatter();
 707                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 708                 $this->debug();
 709         }
 710
 711         /**
 712          * Generate a debug comment indicating diff generating time,
 713          * server node, and generator backend.
 714          */
 715         protected function debug( $generator="internal" ) {
 716                 global $wgShowHostnames;
 717                 $data = array( $generator );
 718                 if( $wgShowHostnames ) {
 719                         $data[] = wfHostname();
 720                 }
 721                 $data[] = wfTimestamp( TS_DB );
 722                 return "<!-- diff generator: " .
 723                 implode( " ",
 724                 array_map(
 725                                         "htmlspecialchars",
 726                 $data ) ) .
 727                         " -->\n";
 728         }
 729
 730         /**
 731          * Replace line numbers with the text in the user's language
 732          */
 733         function localiseLineNumbers( $text ) {
 734                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 735                 array( &$this, 'localiseLineNumbersCb' ), $text );
 736         }
 737
 738         function localiseLineNumbersCb( $matches ) {
 739                 global $wgLang;
 740                 return wfMsgExt( 'lineno', array (), $wgLang->formatNum( $matches[1] ) );
 741         }
 742
 743
 744         /**
 745          * If there are revisions between the ones being compared, return a note saying so.
 746          */
 747         function getMultiNotice() {
 748                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 749                 return '';
 750
 751                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 752                         // Comparing two different pages? Count would be meaningless.
 753                         return '';
 754                 }
 755
 756                 $oldid = $this->mOldRev->getId();
 757                 $newid = $this->mNewRev->getId();
 758                 if ( $oldid > $newid ) {
 759                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 760                 }
 761
 762                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 763                 if ( !$n )
 764                 return '';
 765
 766                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 767         }
 768
 769
 770         /**
 771          * Add the header to a diff body
 772          */
 773         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 774                 $colspan = 1;
 775                 $header = "<table class='diff'>";
 776                 if( $diff ) { // Safari/Chrome show broken output if cols not used
 777                         $header .= "
 778                         <col class='diff-marker' />
 779                         <col class='diff-content' />
 780                         <col class='diff-marker' />
 781                         <col class='diff-content' />";
 782                         $colspan = 2;
 783                 }
 784                 $header .= "
 785                 <tr valign='top'>
 786                 <td colspan='$colspan' class='diff-otitle'>{$otitle}</td>
 787                 <td colspan='$colspan' class='diff-ntitle'>{$ntitle}</td>
 788                 </tr>";
 789
 790                 if ( $multi != '' )
 791                 $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 792
 793                 return $header . $diff . "</table>";
 794         }
 795
 796         /**
 797          * Use specified text instead of loading from the database
 798          */
 799         function setText( $oldText, $newText ) {
 800                 $this->mOldtext = $oldText;
 801                 $this->mNewtext = $newText;
 802                 $this->mTextLoaded = 2;
 803                 $this->mRevisionsLoaded = true;
 804         }
 805
 806         /**
 807          * Load revision metadata for the specified articles. If newid is 0, then compare
 808          * the old article in oldid to the current article; if oldid is 0, then
 809          * compare the current article to the immediately previous one (ignoring the
 810          * value of newid).
 811          *
 812          * If oldid is false, leave the corresponding revision object set
 813          * to false. This is impossible via ordinary user input, and is provided for
 814          * API convenience.
 815          */
 816         function loadRevisionData() {
 817                 global $wgLang, $wgUser;
 818                 if ( $this->mRevisionsLoaded ) {
 819                         return true;
 820                 } else {
 821                         // Whether it succeeds or fails, we don't want to try again
 822                         $this->mRevisionsLoaded = true;
 823                 }
 824
 825                 // Load the new revision object
 826                 $this->mNewRev = $this->mNewid
 827                         ? Revision::newFromId( $this->mNewid )
 828                         : Revision::newFromTitle( $this->mTitle );
 829                 if( !$this->mNewRev instanceof Revision )
 830                         return false;
 831
 832                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 833                 $this->mNewid = $this->mNewRev->getId();
 834
 835                 // Check if page is editable
 836                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 837
 838                 // Set assorted variables
 839                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 840                 $this->mNewPage = $this->mNewRev->getTitle();
 841                 if( $this->mNewRev->isCurrent() ) {
 842                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 843                         $this->mPagetitle = wfMsgHTML( 'currentrev-asof', $timestamp );
 844                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 845
 846                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 847                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 848                 } else {
 849                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 850                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 851                         $this->mPagetitle = wfMsgHTML( 'revisionasof', $timestamp );
 852
 853                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 854                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 855                 }
 856                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 857                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 858                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 859                         $this->mNewtitle = "<span class='history-deleted'>{$this->mNewtitle}</span>";
 860                 }
 861
 862                 // Load the old revision object
 863                 $this->mOldRev = false;
 864                 if( $this->mOldid ) {
 865                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 866                 } elseif ( $this->mOldid === 0 ) {
 867                         $rev = $this->mNewRev->getPrevious();
 868                         if( $rev ) {
 869                                 $this->mOldid = $rev->getId();
 870                                 $this->mOldRev = $rev;
 871                         } else {
 872                                 // No previous revision; mark to show as first-version only.
 873                                 $this->mOldid = false;
 874                                 $this->mOldRev = false;
 875                         }
 876                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 877
 878                 if( is_null( $this->mOldRev ) ) {
 879                         return false;
 880                 }
 881
 882                 if ( $this->mOldRev ) {
 883                         $this->mOldPage = $this->mOldRev->getTitle();
 884
 885                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 886                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 887                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 888                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 889
 890                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 891                         . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 892                         // Add an "undo" link
 893                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 894                         $htmlLink = htmlspecialchars( wfMsg( 'editundo' ) );
 895                         $htmlTitle = $wgUser->getSkin()->tooltip( 'undo' );
 896                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 897                                 $this->mNewtitle .= " (<a href='$newUndo' $htmlTitle>" . $htmlLink . "</a>)";
 898                         }
 899
 900                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 901                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 902                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 903                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 904                         }
 905                 }
 906
 907                 return true;
 908         }
 909
 910         /**
 911          * Load the text of the revisions, as well as revision data.
 912          */
 913         function loadText() {
 914                 if ( $this->mTextLoaded == 2 ) {
 915                         return true;
 916                 } else {
 917                         // Whether it succeeds or fails, we don't want to try again
 918                         $this->mTextLoaded = 2;
 919                 }
 920
 921                 if ( !$this->loadRevisionData() ) {
 922                         return false;
 923                 }
 924                 if ( $this->mOldRev ) {
 925                         $this->mOldtext = $this->mOldRev->getText( Revision::FOR_THIS_USER );
 926                         if ( $this->mOldtext === false ) {
 927                                 return false;
 928                         }
 929                 }
 930                 if ( $this->mNewRev ) {
 931                         $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 932                         if ( $this->mNewtext === false ) {
 933                                 return false;
 934                         }
 935                 }
 936                 return true;
 937         }
 938
 939         /**
 940          * Load the text of the new revision, not the old one
 941          */
 942         function loadNewText() {
 943                 if ( $this->mTextLoaded >= 1 ) {
 944                         return true;
 945                 } else {
 946                         $this->mTextLoaded = 1;
 947                 }
 948                 if ( !$this->loadRevisionData() ) {
 949                         return false;
 950                 }
 951                 $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 952                 return true;
 953         }
 954
 955
 956 }
 957
 958 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 959 //
 960 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 961 // You may copy this code freely under the conditions of the GPL.
 962 //
 963
 964 define('USE_ASSERTS', function_exists('assert'));
 965
 966 /**
 967  * @todo document
 968  * @private
 969  * @ingroup DifferenceEngine
 970  */
 971 class _DiffOp {
 972         var $type;
 973         var $orig;
 974         var $closing;
 975
 976         function reverse() {
 977                 trigger_error('pure virtual', E_USER_ERROR);
 978         }
 979
 980         function norig() {
 981                 return $this->orig ? sizeof($this->orig) : 0;
 982         }
 983
 984         function nclosing() {
 985                 return $this->closing ? sizeof($this->closing) : 0;
 986         }
 987 }
 988
 989 /**
 990  * @todo document
 991  * @private
 992  * @ingroup DifferenceEngine
 993  */
 994 class _DiffOp_Copy extends _DiffOp {
 995         var $type = 'copy';
 996
 997         function _DiffOp_Copy ($orig, $closing = false) {
 998                 if (!is_array($closing))
 999                 $closing = $orig;
1000                 $this->orig = $orig;
1001                 $this->closing = $closing;
1002         }
1003
1004         function reverse() {
1005                 return new _DiffOp_Copy($this->closing, $this->orig);
1006         }
1007 }
1008
1009 /**
1010  * @todo document
1011  * @private
1012  * @ingroup DifferenceEngine
1013  */
1014 class _DiffOp_Delete extends _DiffOp {
1015         var $type = 'delete';
1016
1017         function _DiffOp_Delete ($lines) {
1018                 $this->orig = $lines;
1019                 $this->closing = false;
1020         }
1021
1022         function reverse() {
1023                 return new _DiffOp_Add($this->orig);
1024         }
1025 }
1026
1027 /**
1028  * @todo document
1029  * @private
1030  * @ingroup DifferenceEngine
1031  */
1032 class _DiffOp_Add extends _DiffOp {
1033         var $type = 'add';
1034
1035         function _DiffOp_Add ($lines) {
1036                 $this->closing = $lines;
1037                 $this->orig = false;
1038         }
1039
1040         function reverse() {
1041                 return new _DiffOp_Delete($this->closing);
1042         }
1043 }
1044
1045 /**
1046  * @todo document
1047  * @private
1048  * @ingroup DifferenceEngine
1049  */
1050 class _DiffOp_Change extends _DiffOp {
1051         var $type = 'change';
1052
1053         function _DiffOp_Change ($orig, $closing) {
1054                 $this->orig = $orig;
1055                 $this->closing = $closing;
1056         }
1057
1058         function reverse() {
1059                 return new _DiffOp_Change($this->closing, $this->orig);
1060         }
1061 }
1062
1063 /**
1064  * Class used internally by Diff to actually compute the diffs.
1065  *
1066  * The algorithm used here is mostly lifted from the perl module
1067  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
1068  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
1069  *
1070  * More ideas are taken from:
1071  *       http://www.ics.uci.edu/~eppstein/161/960229.html
1072  *
1073  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
1074  * diffutils-2.7, which can be found at:
1075  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
1076  *
1077  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
1078  * are my own.
1079  *
1080  * Line length limits for robustness added by Tim Starling, 2005-08-31
1081  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
1082  *
1083  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
1084  * @private
1085  * @ingroup DifferenceEngine
1086  */
1087 class _DiffEngine {
1088
1089         const MAX_XREF_LENGTH =  10000;
1090
1091         function diff ($from_lines, $to_lines){
1092                 wfProfileIn( __METHOD__ );
1093
1094                 // Diff and store locally
1095                 $this->diff_local($from_lines, $to_lines);
1096
1097                 // Merge edits when possible
1098                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
1099                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
1100
1101                 // Compute the edit operations.
1102                 $n_from = sizeof($from_lines);
1103                 $n_to = sizeof($to_lines);
1104
1105                 $edits = array();
1106                 $xi = $yi = 0;
1107                 while ($xi < $n_from || $yi < $n_to) {
1108                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
1109                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
1110
1111                         // Skip matching "snake".
1112                         $copy = array();
1113                         while ( $xi < $n_from && $yi < $n_to
1114                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
1115                                 $copy[] = $from_lines[$xi++];
1116                                 ++$yi;
1117                         }
1118                         if ($copy)
1119                         $edits[] = new _DiffOp_Copy($copy);
1120
1121                         // Find deletes & adds.
1122                         $delete = array();
1123                         while ($xi < $n_from && $this->xchanged[$xi])
1124                         $delete[] = $from_lines[$xi++];
1125
1126                         $add = array();
1127                         while ($yi < $n_to && $this->ychanged[$yi])
1128                         $add[] = $to_lines[$yi++];
1129
1130                         if ($delete && $add)
1131                         $edits[] = new _DiffOp_Change($delete, $add);
1132                         elseif ($delete)
1133                         $edits[] = new _DiffOp_Delete($delete);
1134                         elseif ($add)
1135                         $edits[] = new _DiffOp_Add($add);
1136                 }
1137                 wfProfileOut( __METHOD__ );
1138                 return $edits;
1139         }
1140
1141         function diff_local ($from_lines, $to_lines) {
1142                 global $wgExternalDiffEngine;
1143                 wfProfileIn( __METHOD__);
1144
1145                 if($wgExternalDiffEngine == 'wikidiff3'){
1146                         // wikidiff3
1147                         $wikidiff3 = new WikiDiff3();
1148                         $wikidiff3->diff($from_lines, $to_lines);
1149                         $this->xchanged = $wikidiff3->removed;
1150                         $this->ychanged = $wikidiff3->added;
1151                         unset($wikidiff3);
1152                 }else{
1153                         // old diff
1154                         $n_from = sizeof($from_lines);
1155                         $n_to = sizeof($to_lines);
1156                         $this->xchanged = $this->ychanged = array();
1157                         $this->xv = $this->yv = array();
1158                         $this->xind = $this->yind = array();
1159                         unset($this->seq);
1160                         unset($this->in_seq);
1161                         unset($this->lcs);
1162
1163                         // Skip leading common lines.
1164                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
1165                                 if ($from_lines[$skip] !== $to_lines[$skip])
1166                                 break;
1167                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
1168                         }
1169                         // Skip trailing common lines.
1170                         $xi = $n_from; $yi = $n_to;
1171                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
1172                                 if ($from_lines[$xi] !== $to_lines[$yi])
1173                                 break;
1174                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
1175                         }
1176
1177                         // Ignore lines which do not exist in both files.
1178                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1179                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
1180                         }
1181
1182                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
1183                                 $line = $to_lines[$yi];
1184                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
1185                                 continue;
1186                                 $yhash[$this->_line_hash($line)] = 1;
1187                                 $this->yv[] = $line;
1188                                 $this->yind[] = $yi;
1189                         }
1190                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1191                                 $line = $from_lines[$xi];
1192                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
1193                                 continue;
1194                                 $this->xv[] = $line;
1195                                 $this->xind[] = $xi;
1196                         }
1197
1198                         // Find the LCS.
1199                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
1200                 }
1201                 wfProfileOut( __METHOD__ );
1202         }
1203
1204         /**
1205          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1206          */
1207         function _line_hash( $line ) {
1208                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1209                         return md5( $line );
1210                 } else {
1211                         return $line;
1212                 }
1213         }
1214
1215         /* Divide the Largest Common Subsequence (LCS) of the sequences
1216          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1217          * sized segments.
1218          *
1219          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1220          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1221          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1222          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1223          * that (X0, Y0) == (XOFF, YOFF) and
1224          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1225          *
1226          * This function assumes that the first lines of the specified portions
1227          * of the two files do not match, and likewise that the last lines do not
1228          * match.  The caller must trim matching lines from the beginning and end
1229          * of the portions it is going to specify.
1230          */
1231         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1232                 $flip = false;
1233
1234                 if ($xlim - $xoff > $ylim - $yoff) {
1235                         // Things seems faster (I'm not sure I understand why)
1236                         // when the shortest sequence in X.
1237                         $flip = true;
1238                         list ($xoff, $xlim, $yoff, $ylim)
1239                         = array( $yoff, $ylim, $xoff, $xlim);
1240                 }
1241
1242                 if ($flip)
1243                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1244                 $ymatches[$this->xv[$i]][] = $i;
1245                 else
1246                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1247                 $ymatches[$this->yv[$i]][] = $i;
1248
1249                 $this->lcs = 0;
1250                 $this->seq[0]= $yoff - 1;
1251                 $this->in_seq = array();
1252                 $ymids[0] = array();
1253
1254                 $numer = $xlim - $xoff + $nchunks - 1;
1255                 $x = $xoff;
1256                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1257                         if ($chunk > 0)
1258                         for ($i = 0; $i <= $this->lcs; $i++)
1259                         $ymids[$i][$chunk-1] = $this->seq[$i];
1260
1261                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1262                         for ( ; $x < $x1; $x++) {
1263                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1264                                 if (empty($ymatches[$line]))
1265                                 continue;
1266                                 $matches = $ymatches[$line];
1267                                 reset($matches);
1268                                 while (list ($junk, $y) = each($matches))
1269                                 if (empty($this->in_seq[$y])) {
1270                                         $k = $this->_lcs_pos($y);
1271                                         USE_ASSERTS && assert($k > 0);
1272                                         $ymids[$k] = $ymids[$k-1];
1273                                         break;
1274                                 }
1275                                 while (list ( /* $junk */, $y) = each($matches)) {
1276                                         if ($y > $this->seq[$k-1]) {
1277                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1278                                                 // Optimization: this is a common case:
1279                                                 //      next match is just replacing previous match.
1280                                                 $this->in_seq[$this->seq[$k]] = false;
1281                                                 $this->seq[$k] = $y;
1282                                                 $this->in_seq[$y] = 1;
1283                                         } else if (empty($this->in_seq[$y])) {
1284                                                 $k = $this->_lcs_pos($y);
1285                                                 USE_ASSERTS && assert($k > 0);
1286                                                 $ymids[$k] = $ymids[$k-1];
1287                                         }
1288                                 }
1289                         }
1290                 }
1291
1292                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1293                 $ymid = $ymids[$this->lcs];
1294                 for ($n = 0; $n < $nchunks - 1; $n++) {
1295                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1296                         $y1 = $ymid[$n] + 1;
1297                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1298                 }
1299                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1300
1301                 return array($this->lcs, $seps);
1302         }
1303
1304         function _lcs_pos ($ypos) {
1305                 $end = $this->lcs;
1306                 if ($end == 0 || $ypos > $this->seq[$end]) {
1307                         $this->seq[++$this->lcs] = $ypos;
1308                         $this->in_seq[$ypos] = 1;
1309                         return $this->lcs;
1310                 }
1311
1312                 $beg = 1;
1313                 while ($beg < $end) {
1314                         $mid = (int)(($beg + $end) / 2);
1315                         if ( $ypos > $this->seq[$mid] )
1316                         $beg = $mid + 1;
1317                         else
1318                         $end = $mid;
1319                 }
1320
1321                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1322
1323                 $this->in_seq[$this->seq[$end]] = false;
1324                 $this->seq[$end] = $ypos;
1325                 $this->in_seq[$ypos] = 1;
1326                 return $end;
1327         }
1328
1329         /* Find LCS of two sequences.
1330          *
1331          * The results are recorded in the vectors $this->{x,y}changed[], by
1332          * storing a 1 in the element for each line that is an insertion
1333          * or deletion (ie. is not in the LCS).
1334          *
1335          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1336          *
1337          * Note that XLIM, YLIM are exclusive bounds.
1338          * All line numbers are origin-0 and discarded lines are not counted.
1339          */
1340         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1341                 // Slide down the bottom initial diagonal.
1342                 while ($xoff < $xlim && $yoff < $ylim
1343                 && $this->xv[$xoff] == $this->yv[$yoff]) {
1344                         ++$xoff;
1345                         ++$yoff;
1346                 }
1347
1348                 // Slide up the top initial diagonal.
1349                 while ($xlim > $xoff && $ylim > $yoff
1350                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1351                         --$xlim;
1352                         --$ylim;
1353                 }
1354
1355                 if ($xoff == $xlim || $yoff == $ylim)
1356                 $lcs = 0;
1357                 else {
1358                         // This is ad hoc but seems to work well.
1359                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1360                         //$nchunks = max(2,min(8,(int)$nchunks));
1361                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1362                         list ($lcs, $seps)
1363                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1364                 }
1365
1366                 if ($lcs == 0) {
1367                         // X and Y sequences have no common subsequence:
1368                         // mark all changed.
1369                         while ($yoff < $ylim)
1370                         $this->ychanged[$this->yind[$yoff++]] = 1;
1371                         while ($xoff < $xlim)
1372                         $this->xchanged[$this->xind[$xoff++]] = 1;
1373                 } else {
1374                         // Use the partitions to split this problem into subproblems.
1375                         reset($seps);
1376                         $pt1 = $seps[0];
1377                         while ($pt2 = next($seps)) {
1378                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1379                                 $pt1 = $pt2;
1380                         }
1381                 }
1382         }
1383
1384         /* Adjust inserts/deletes of identical lines to join changes
1385          * as much as possible.
1386          *
1387          * We do something when a run of changed lines include a
1388          * line at one end and has an excluded, identical line at the other.
1389          * We are free to choose which identical line is included.
1390          * `compareseq' usually chooses the one at the beginning,
1391          * but usually it is cleaner to consider the following identical line
1392          * to be the "change".
1393          *
1394          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1395          */
1396         function _shift_boundaries ($lines, &$changed, $other_changed) {
1397                 wfProfileIn( __METHOD__ );
1398                 $i = 0;
1399                 $j = 0;
1400
1401                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1402                 $len = sizeof($lines);
1403                 $other_len = sizeof($other_changed);
1404
1405                 while (1) {
1406                         /*
1407                          * Scan forwards to find beginning of another run of changes.
1408                          * Also keep track of the corresponding point in the other file.
1409                          *
1410                          * Throughout this code, $i and $j are adjusted together so that
1411                          * the first $i elements of $changed and the first $j elements
1412                          * of $other_changed both contain the same number of zeros
1413                          * (unchanged lines).
1414                          * Furthermore, $j is always kept so that $j == $other_len or
1415                          * $other_changed[$j] == false.
1416                          */
1417                         while ($j < $other_len && $other_changed[$j])
1418                         $j++;
1419
1420                         while ($i < $len && ! $changed[$i]) {
1421                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1422                                 $i++; $j++;
1423                                 while ($j < $other_len && $other_changed[$j])
1424                                 $j++;
1425                         }
1426
1427                         if ($i == $len)
1428                         break;
1429
1430                         $start = $i;
1431
1432                         // Find the end of this run of changes.
1433                         while (++$i < $len && $changed[$i])
1434                         continue;
1435
1436                         do {
1437                                 /*
1438                                  * Record the length of this run of changes, so that
1439                                  * we can later determine whether the run has grown.
1440                                  */
1441                                 $runlength = $i - $start;
1442
1443                                 /*
1444                                  * Move the changed region back, so long as the
1445                                  * previous unchanged line matches the last changed one.
1446                                  * This merges with previous changed regions.
1447                                  */
1448                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1449                                         $changed[--$start] = 1;
1450                                         $changed[--$i] = false;
1451                                         while ($start > 0 && $changed[$start - 1])
1452                                         $start--;
1453                                         USE_ASSERTS && assert('$j > 0');
1454                                         while ($other_changed[--$j])
1455                                         continue;
1456                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1457                                 }
1458
1459                                 /*
1460                                  * Set CORRESPONDING to the end of the changed run, at the last
1461                                  * point where it corresponds to a changed run in the other file.
1462                                  * CORRESPONDING == LEN means no such point has been found.
1463                                  */
1464                                 $corresponding = $j < $other_len ? $i : $len;
1465
1466                                 /*
1467                                  * Move the changed region forward, so long as the
1468                                  * first changed line matches the following unchanged one.
1469                                  * This merges with following changed regions.
1470                                  * Do this second, so that if there are no merges,
1471                                  * the changed region is moved forward as far as possible.
1472                                  */
1473                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1474                                         $changed[$start++] = false;
1475                                         $changed[$i++] = 1;
1476                                         while ($i < $len && $changed[$i])
1477                                         $i++;
1478
1479                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1480                                         $j++;
1481                                         if ($j < $other_len && $other_changed[$j]) {
1482                                                 $corresponding = $i;
1483                                                 while ($j < $other_len && $other_changed[$j])
1484                                                 $j++;
1485                                         }
1486                                 }
1487                         } while ($runlength != $i - $start);
1488
1489                         /*
1490                          * If possible, move the fully-merged run of changes
1491                          * back to a corresponding run in the other file.
1492                          */
1493                         while ($corresponding < $i) {
1494                                 $changed[--$start] = 1;
1495                                 $changed[--$i] = 0;
1496                                 USE_ASSERTS && assert('$j > 0');
1497                                 while ($other_changed[--$j])
1498                                 continue;
1499                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1500                         }
1501                 }
1502                 wfProfileOut( __METHOD__ );
1503         }
1504 }
1505
1506 /**
1507  * Class representing a 'diff' between two sequences of strings.
1508  * @todo document
1509  * @private
1510  * @ingroup DifferenceEngine
1511  */
1512 class Diff
1513 {
1514         var $edits;
1515
1516         /**
1517          * Constructor.
1518          * Computes diff between sequences of strings.
1519          *
1520          * @param $from_lines array An array of strings.
1521          *                (Typically these are lines from a file.)
1522          * @param $to_lines array An array of strings.
1523          */
1524         function Diff($from_lines, $to_lines) {
1525                 $eng = new _DiffEngine;
1526                 $this->edits = $eng->diff($from_lines, $to_lines);
1527                 //$this->_check($from_lines, $to_lines);
1528         }
1529
1530         /**
1531          * Compute reversed Diff.
1532          *
1533          * SYNOPSIS:
1534          *
1535          *      $diff = new Diff($lines1, $lines2);
1536          *      $rev = $diff->reverse();
1537          * @return object A Diff object representing the inverse of the
1538          *                                original diff.
1539          */
1540         function reverse () {
1541                 $rev = $this;
1542                 $rev->edits = array();
1543                 foreach ($this->edits as $edit) {
1544                         $rev->edits[] = $edit->reverse();
1545                 }
1546                 return $rev;
1547         }
1548
1549         /**
1550          * Check for empty diff.
1551          *
1552          * @return bool True iff two sequences were identical.
1553          */
1554         function isEmpty () {
1555                 foreach ($this->edits as $edit) {
1556                         if ($edit->type != 'copy')
1557                         return false;
1558                 }
1559                 return true;
1560         }
1561
1562         /**
1563          * Compute the length of the Longest Common Subsequence (LCS).
1564          *
1565          * This is mostly for diagnostic purposed.
1566          *
1567          * @return int The length of the LCS.
1568          */
1569         function lcs () {
1570                 $lcs = 0;
1571                 foreach ($this->edits as $edit) {
1572                         if ($edit->type == 'copy')
1573                         $lcs += sizeof($edit->orig);
1574                 }
1575                 return $lcs;
1576         }
1577
1578         /**
1579          * Get the original set of lines.
1580          *
1581          * This reconstructs the $from_lines parameter passed to the
1582          * constructor.
1583          *
1584          * @return array The original sequence of strings.
1585          */
1586         function orig() {
1587                 $lines = array();
1588
1589                 foreach ($this->edits as $edit) {
1590                         if ($edit->orig)
1591                         array_splice($lines, sizeof($lines), 0, $edit->orig);
1592                 }
1593                 return $lines;
1594         }
1595
1596         /**
1597          * Get the closing set of lines.
1598          *
1599          * This reconstructs the $to_lines parameter passed to the
1600          * constructor.
1601          *
1602          * @return array The sequence of strings.
1603          */
1604         function closing() {
1605                 $lines = array();
1606
1607                 foreach ($this->edits as $edit) {
1608                         if ($edit->closing)
1609                         array_splice($lines, sizeof($lines), 0, $edit->closing);
1610                 }
1611                 return $lines;
1612         }
1613
1614         /**
1615          * Check a Diff for validity.
1616          *
1617          * This is here only for debugging purposes.
1618          */
1619         function _check ($from_lines, $to_lines) {
1620                 wfProfileIn( __METHOD__ );
1621                 if (serialize($from_lines) != serialize($this->orig()))
1622                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1623                 if (serialize($to_lines) != serialize($this->closing()))
1624                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1625
1626                 $rev = $this->reverse();
1627                 if (serialize($to_lines) != serialize($rev->orig()))
1628                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
1629                 if (serialize($from_lines) != serialize($rev->closing()))
1630                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1631
1632
1633                 $prevtype = 'none';
1634                 foreach ($this->edits as $edit) {
1635                         if ( $prevtype == $edit->type )
1636                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1637                         $prevtype = $edit->type;
1638                 }
1639
1640                 $lcs = $this->lcs();
1641                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1642                 wfProfileOut( __METHOD__ );
1643         }
1644 }
1645
1646 /**
1647  * @todo document, bad name.
1648  * @private
1649  * @ingroup DifferenceEngine
1650  */
1651 class MappedDiff extends Diff
1652 {
1653         /**
1654          * Constructor.
1655          *
1656          * Computes diff between sequences of strings.
1657          *
1658          * This can be used to compute things like
1659          * case-insensitve diffs, or diffs which ignore
1660          * changes in white-space.
1661          *
1662          * @param $from_lines array An array of strings.
1663          *      (Typically these are lines from a file.)
1664          *
1665          * @param $to_lines array An array of strings.
1666          *
1667          * @param $mapped_from_lines array This array should
1668          *      have the same size number of elements as $from_lines.
1669          *      The elements in $mapped_from_lines and
1670          *      $mapped_to_lines are what is actually compared
1671          *      when computing the diff.
1672          *
1673          * @param $mapped_to_lines array This array should
1674          *      have the same number of elements as $to_lines.
1675          */
1676         function MappedDiff($from_lines, $to_lines,
1677         $mapped_from_lines, $mapped_to_lines) {
1678                 wfProfileIn( __METHOD__ );
1679
1680                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1681                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1682
1683                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1684
1685                 $xi = $yi = 0;
1686                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1687                         $orig = &$this->edits[$i]->orig;
1688                         if (is_array($orig)) {
1689                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1690                                 $xi += sizeof($orig);
1691                         }
1692
1693                         $closing = &$this->edits[$i]->closing;
1694                         if (is_array($closing)) {
1695                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1696                                 $yi += sizeof($closing);
1697                         }
1698                 }
1699                 wfProfileOut( __METHOD__ );
1700         }
1701 }
1702
1703 /**
1704  * A class to format Diffs
1705  *
1706  * This class formats the diff in classic diff format.
1707  * It is intended that this class be customized via inheritance,
1708  * to obtain fancier outputs.
1709  * @todo document
1710  * @private
1711  * @ingroup DifferenceEngine
1712  */
1713 class DiffFormatter {
1714         /**
1715          * Number of leading context "lines" to preserve.
1716          *
1717          * This should be left at zero for this class, but subclasses
1718          * may want to set this to other values.
1719          */
1720         var $leading_context_lines = 0;
1721
1722         /**
1723          * Number of trailing context "lines" to preserve.
1724          *
1725          * This should be left at zero for this class, but subclasses
1726          * may want to set this to other values.
1727          */
1728         var $trailing_context_lines = 0;
1729
1730         /**
1731          * Format a diff.
1732          *
1733          * @param $diff object A Diff object.
1734          * @return string The formatted output.
1735          */
1736         function format($diff) {
1737                 wfProfileIn( __METHOD__ );
1738
1739                 $xi = $yi = 1;
1740                 $block = false;
1741                 $context = array();
1742
1743                 $nlead = $this->leading_context_lines;
1744                 $ntrail = $this->trailing_context_lines;
1745
1746                 $this->_start_diff();
1747
1748                 foreach ($diff->edits as $edit) {
1749                         if ($edit->type == 'copy') {
1750                                 if (is_array($block)) {
1751                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1752                                                 $block[] = $edit;
1753                                         }
1754                                         else{
1755                                                 if ($ntrail) {
1756                                                         $context = array_slice($edit->orig, 0, $ntrail);
1757                                                         $block[] = new _DiffOp_Copy($context);
1758                                                 }
1759                                                 $this->_block($x0, $ntrail + $xi - $x0,
1760                                                 $y0, $ntrail + $yi - $y0,
1761                                                 $block);
1762                                                 $block = false;
1763                                         }
1764                                 }
1765                                 $context = $edit->orig;
1766                         }
1767                         else {
1768                                 if (! is_array($block)) {
1769                                         $context = array_slice($context, sizeof($context) - $nlead);
1770                                         $x0 = $xi - sizeof($context);
1771                                         $y0 = $yi - sizeof($context);
1772                                         $block = array();
1773                                         if ($context)
1774                                         $block[] = new _DiffOp_Copy($context);
1775                                 }
1776                                 $block[] = $edit;
1777                         }
1778
1779                         if ($edit->orig)
1780                         $xi += sizeof($edit->orig);
1781                         if ($edit->closing)
1782                         $yi += sizeof($edit->closing);
1783                 }
1784
1785                 if (is_array($block))
1786                 $this->_block($x0, $xi - $x0,
1787                 $y0, $yi - $y0,
1788                 $block);
1789
1790                 $end = $this->_end_diff();
1791                 wfProfileOut( __METHOD__ );
1792                 return $end;
1793         }
1794
1795         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1796                 wfProfileIn( __METHOD__ );
1797                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1798                 foreach ($edits as $edit) {
1799                         if ($edit->type == 'copy')
1800                         $this->_context($edit->orig);
1801                         elseif ($edit->type == 'add')
1802                         $this->_added($edit->closing);
1803                         elseif ($edit->type == 'delete')
1804                         $this->_deleted($edit->orig);
1805                         elseif ($edit->type == 'change')
1806                         $this->_changed($edit->orig, $edit->closing);
1807                         else
1808                         trigger_error('Unknown edit type', E_USER_ERROR);
1809                 }
1810                 $this->_end_block();
1811                 wfProfileOut( __METHOD__ );
1812         }
1813
1814         function _start_diff() {
1815                 ob_start();
1816         }
1817
1818         function _end_diff() {
1819                 $val = ob_get_contents();
1820                 ob_end_clean();
1821                 return $val;
1822         }
1823
1824         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1825                 if ($xlen > 1)
1826                 $xbeg .= "," . ($xbeg + $xlen - 1);
1827                 if ($ylen > 1)
1828                 $ybeg .= "," . ($ybeg + $ylen - 1);
1829
1830                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1831         }
1832
1833         function _start_block($header) {
1834                 echo $header . "\n";
1835         }
1836
1837         function _end_block() {
1838         }
1839
1840         function _lines($lines, $prefix = ' ') {
1841                 foreach ($lines as $line)
1842                 echo "$prefix $line\n";
1843         }
1844
1845         function _context($lines) {
1846                 $this->_lines($lines);
1847         }
1848
1849         function _added($lines) {
1850                 $this->_lines($lines, '>');
1851         }
1852         function _deleted($lines) {
1853                 $this->_lines($lines, '<');
1854         }
1855
1856         function _changed($orig, $closing) {
1857                 $this->_deleted($orig);
1858                 echo "---\n";
1859                 $this->_added($closing);
1860         }
1861 }
1862
1863 /**
1864  * A formatter that outputs unified diffs
1865  * @ingroup DifferenceEngine
1866  */
1867
1868 class UnifiedDiffFormatter extends DiffFormatter {
1869         var $leading_context_lines = 2;
1870         var $trailing_context_lines = 2;
1871
1872         function _added($lines) {
1873                 $this->_lines($lines, '+');
1874         }
1875         function _deleted($lines) {
1876                 $this->_lines($lines, '-');
1877         }
1878         function _changed($orig, $closing) {
1879                 $this->_deleted($orig);
1880                 $this->_added($closing);
1881         }
1882         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1883                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1884         }
1885 }
1886
1887 /**
1888  * A pseudo-formatter that just passes along the Diff::$edits array
1889  * @ingroup DifferenceEngine
1890  */
1891 class ArrayDiffFormatter extends DiffFormatter {
1892         function format($diff) {
1893                 $oldline = 1;
1894                 $newline = 1;
1895                 $retval = array();
1896                 foreach($diff->edits as $edit)
1897                 switch($edit->type) {
1898                         case 'add':
1899                                 foreach($edit->closing as $l) {
1900                                         $retval[] = array(
1901                                                         'action' => 'add',
1902                                                         'new'=> $l,
1903                                                         'newline' => $newline++
1904                                         );
1905                                 }
1906                                 break;
1907                         case 'delete':
1908                                 foreach($edit->orig as $l) {
1909                                         $retval[] = array(
1910                                                         'action' => 'delete',
1911                                                         'old' => $l,
1912                                                         'oldline' => $oldline++,
1913                                         );
1914                                 }
1915                                 break;
1916                         case 'change':
1917                                 foreach($edit->orig as $i => $l) {
1918                                         $retval[] = array(
1919                                                         'action' => 'change',
1920                                                         'old' => $l,
1921                                                         'new' => @$edit->closing[$i],
1922                                                         'oldline' => $oldline++,
1923                                                         'newline' => $newline++,
1924                                         );
1925                                 }
1926                                 break;
1927                         case 'copy':
1928                                 $oldline += count($edit->orig);
1929                                 $newline += count($edit->orig);
1930                 }
1931                 return $retval;
1932         }
1933 }
1934
1935 /**
1936  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1937  *
1938  */
1939
1940 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1941
1942 /**
1943  * @todo document
1944  * @private
1945  * @ingroup DifferenceEngine
1946  */
1947 class _HWLDF_WordAccumulator {
1948         function _HWLDF_WordAccumulator () {
1949                 $this->_lines = array();
1950                 $this->_line = '';
1951                 $this->_group = '';
1952                 $this->_tag = '';
1953         }
1954
1955         function _flushGroup ($new_tag) {
1956                 if ($this->_group !== '') {
1957                         if ($this->_tag == 'ins')
1958                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1959                         htmlspecialchars ( $this->_group ) . '</ins>';
1960                         elseif ($this->_tag == 'del')
1961                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1962                         htmlspecialchars ( $this->_group ) . '</del>';
1963                         else
1964                         $this->_line .= htmlspecialchars ( $this->_group );
1965                 }
1966                 $this->_group = '';
1967                 $this->_tag = $new_tag;
1968         }
1969
1970         function _flushLine ($new_tag) {
1971                 $this->_flushGroup($new_tag);
1972                 if ($this->_line != '')
1973                 array_push ( $this->_lines, $this->_line );
1974                 else
1975                 # make empty lines visible by inserting an NBSP
1976                 array_push ( $this->_lines, NBSP );
1977                 $this->_line = '';
1978         }
1979
1980         function addWords ($words, $tag = '') {
1981                 if ($tag != $this->_tag)
1982                 $this->_flushGroup($tag);
1983
1984                 foreach ($words as $word) {
1985                         // new-line should only come as first char of word.
1986                         if ($word == '')
1987                         continue;
1988                         if ($word[0] == "\n") {
1989                                 $this->_flushLine($tag);
1990                                 $word = substr($word, 1);
1991                         }
1992                         assert(!strstr($word, "\n"));
1993                         $this->_group .= $word;
1994                 }
1995         }
1996
1997         function getLines() {
1998                 $this->_flushLine('~done');
1999                 return $this->_lines;
2000         }
2001 }
2002
2003 /**
2004  * @todo document
2005  * @private
2006  * @ingroup DifferenceEngine
2007  */
2008 class WordLevelDiff extends MappedDiff {
2009         const MAX_LINE_LENGTH = 10000;
2010
2011         function WordLevelDiff ($orig_lines, $closing_lines) {
2012                 wfProfileIn( __METHOD__ );
2013
2014                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
2015                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
2016
2017                 $this->MappedDiff($orig_words, $closing_words,
2018                 $orig_stripped, $closing_stripped);
2019                 wfProfileOut( __METHOD__ );
2020         }
2021
2022         function _split($lines) {
2023                 wfProfileIn( __METHOD__ );
2024
2025                 $words = array();
2026                 $stripped = array();
2027                 $first = true;
2028                 foreach ( $lines as $line ) {
2029                         # If the line is too long, just pretend the entire line is one big word
2030                         # This prevents resource exhaustion problems
2031                         if ( $first ) {
2032                                 $first = false;
2033                         } else {
2034                                 $words[] = "\n";
2035                                 $stripped[] = "\n";
2036                         }
2037                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
2038                                 $words[] = $line;
2039                                 $stripped[] = $line;
2040                         } else {
2041                                 $m = array();
2042                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
2043                                 $line, $m))
2044                                 {
2045                                         $words = array_merge( $words, $m[0] );
2046                                         $stripped = array_merge( $stripped, $m[1] );
2047                                 }
2048                         }
2049                 }
2050                 wfProfileOut( __METHOD__ );
2051                 return array($words, $stripped);
2052         }
2053
2054         function orig () {
2055                 wfProfileIn( __METHOD__ );
2056                 $orig = new _HWLDF_WordAccumulator;
2057
2058                 foreach ($this->edits as $edit) {
2059                         if ($edit->type == 'copy')
2060                         $orig->addWords($edit->orig);
2061                         elseif ($edit->orig)
2062                         $orig->addWords($edit->orig, 'del');
2063                 }
2064                 $lines = $orig->getLines();
2065                 wfProfileOut( __METHOD__ );
2066                 return $lines;
2067         }
2068
2069         function closing () {
2070                 wfProfileIn( __METHOD__ );
2071                 $closing = new _HWLDF_WordAccumulator;
2072
2073                 foreach ($this->edits as $edit) {
2074                         if ($edit->type == 'copy')
2075                         $closing->addWords($edit->closing);
2076                         elseif ($edit->closing)
2077                         $closing->addWords($edit->closing, 'ins');
2078                 }
2079                 $lines = $closing->getLines();
2080                 wfProfileOut( __METHOD__ );
2081                 return $lines;
2082         }
2083 }
2084
2085 /**
2086  * Wikipedia Table style diff formatter.
2087  * @todo document
2088  * @private
2089  * @ingroup DifferenceEngine
2090  */
2091 class TableDiffFormatter extends DiffFormatter {
2092         function TableDiffFormatter() {
2093                 $this->leading_context_lines = 2;
2094                 $this->trailing_context_lines = 2;
2095         }
2096
2097         public static function escapeWhiteSpace( $msg ) {
2098                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
2099                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
2100                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
2101                 return $msg;
2102         }
2103
2104         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
2105                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
2106                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
2107                 return $r;
2108         }
2109
2110         function _start_block( $header ) {
2111                 echo $header;
2112         }
2113
2114         function _end_block() {
2115         }
2116
2117         function _lines( $lines, $prefix=' ', $color='white' ) {
2118         }
2119
2120         # HTML-escape parameter before calling this
2121         function addedLine( $line ) {
2122                 return $this->wrapLine( '+', 'diff-addedline', $line );
2123         }
2124
2125         # HTML-escape parameter before calling this
2126         function deletedLine( $line ) {
2127                 return $this->wrapLine( '-', 'diff-deletedline', $line );
2128         }
2129
2130         # HTML-escape parameter before calling this
2131         function contextLine( $line ) {
2132                 return $this->wrapLine( ' ', 'diff-context', $line );
2133         }
2134
2135         private function wrapLine( $marker, $class, $line ) {
2136                 if( $line !== '' ) {
2137                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
2138                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
2139                 }
2140                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
2141         }
2142
2143         function emptyLine() {
2144                 return '<td colspan="2">&nbsp;</td>';
2145         }
2146
2147         function _added( $lines ) {
2148                 foreach ($lines as $line) {
2149                         echo '<tr>' . $this->emptyLine() .
2150                         $this->addedLine( '<ins class="diffchange">' .
2151                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
2152                 }
2153         }
2154
2155         function _deleted($lines) {
2156                 foreach ($lines as $line) {
2157                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
2158                         htmlspecialchars ( $line ) . '</del>' ) .
2159                         $this->emptyLine() . "</tr>\n";
2160                 }
2161         }
2162
2163         function _context( $lines ) {
2164                 foreach ($lines as $line) {
2165                         echo '<tr>' .
2166                         $this->contextLine( htmlspecialchars ( $line ) ) .
2167                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
2168                 }
2169         }
2170
2171         function _changed( $orig, $closing ) {
2172                 wfProfileIn( __METHOD__ );
2173
2174                 $diff = new WordLevelDiff( $orig, $closing );
2175                 $del = $diff->orig();
2176                 $add = $diff->closing();
2177
2178                 # Notice that WordLevelDiff returns HTML-escaped output.
2179                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2180
2181                 while ( $line = array_shift( $del ) ) {
2182                         $aline = array_shift( $add );
2183                         echo '<tr>' . $this->deletedLine( $line ) .
2184                         $this->addedLine( $aline ) . "</tr>\n";
2185                 }
2186                 foreach ($add as $line) {       # If any leftovers
2187                         echo '<tr>' . $this->emptyLine() .
2188                         $this->addedLine( $line ) . "</tr>\n";
2189                 }
2190                 wfProfileOut( __METHOD__ );
2191         }
2192 }