includes/diff/DifferenceEngine.php

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