includes/DifferenceEngine.php

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