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