includes/diff/DifferenceEngine.php

   1 <?php
   2 /**
   3  * @defgroup DifferenceEngine DifferenceEngine
   4  */
   5
   6 /**
   7  * Constant to indicate diff cache compatibility.
   8  * Bump this when changing the diff formatting in a way that
   9  * fixes important bugs or such to force cached diff views to
  10  * clear.
  11  */
  12 define( 'MW_DIFF_VERSION', '1.11a' );
  13
  14 /**
  15  * @todo document
  16  * @ingroup DifferenceEngine
  17  */
  18 class DifferenceEngine {
  19         /**#@+
  20          * @private
  21          */
  22         var $mOldid, $mNewid, $mTitle;
  23         var $mOldtitle, $mNewtitle, $mPagetitle;
  24         var $mOldtext, $mNewtext;
  25         var $mOldPage, $mNewPage;
  26         var $mRcidMarkPatrolled;
  27         var $mOldRev, $mNewRev;
  28         var $mRevisionsLoaded = false; // Have the revisions been loaded
  29         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  30         var $mCacheHit = false; // Was the diff fetched from cache?
  31         var $htmldiff;
  32
  33         protected $unhide = false;
  34         /**#@-*/
  35
  36         /**
  37          * Constructor
  38          * @param $titleObj Title object that the diff is associated with
  39          * @param $old Integer: old ID we want to show and diff with.
  40          * @param $new String: either 'prev' or 'next'.
  41          * @param $rcid Integer: ??? FIXME (default 0)
  42          * @param $refreshCache boolean If set, refreshes the diff cache
  43          * @param $htmldiff boolean If set, output using HTMLDiff instead of raw wikicode diff
  44          * @param $unhide boolean If set, allow viewing deleted revs
  45          */
  46         function __construct( $titleObj = null, $old = 0, $new = 0, $rcid = 0,
  47                 $refreshCache = false, $htmldiff = false, $unhide = false )
  48         {
  49                 $this->mTitle = $titleObj;
  50                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  51
  52                 if ( 'prev' === $new ) {
  53                         # Show diff between revision $old and the previous one.
  54                         # Get previous one from DB.
  55                         $this->mNewid = intval($old);
  56                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  57                 } elseif ( 'next' === $new ) {
  58                         # Show diff between revision $old and the next one.
  59                         # Get next one from DB.
  60                         $this->mOldid = intval($old);
  61                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  62                         if ( false === $this->mNewid ) {
  63                                 # if no result, NewId points to the newest old revision. The only newer
  64                                 # revision is cur, which is "0".
  65                                 $this->mNewid = 0;
  66                         }
  67                 } else {
  68                         $this->mOldid = intval($old);
  69                         $this->mNewid = intval($new);
  70                         wfRunHooks( 'NewDifferenceEngine', array(&$titleObj, &$this->mOldid, &$this->mNewid, $old, $new) );
  71                 }
  72                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  73                 $this->mRefreshCache = $refreshCache;
  74                 $this->htmldiff = $htmldiff;
  75                 $this->unhide = $unhide;
  76         }
  77
  78         function getTitle() {
  79                 return $this->mTitle;
  80         }
  81
  82         function wasCacheHit() {
  83                 return $this->mCacheHit;
  84         }
  85
  86         function getOldid() {
  87                 return $this->mOldid;
  88         }
  89
  90         function getNewid() {
  91                 return $this->mNewid;
  92         }
  93
  94         function showDiffPage( $diffOnly = false ) {
  95                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol, $wgEnableHtmlDiff;
  96                 wfProfileIn( __METHOD__ );
  97
  98
  99                 # If external diffs are enabled both globally and for the user,
 100                 # we'll use the application/x-external-editor interface to call
 101                 # an external diff tool like kompare, kdiff3, etc.
 102                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
 103                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
 104                         $wgOut->disable();
 105                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
 106                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
 107                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
 108                         $special=$wgLang->getNsText(NS_SPECIAL);
 109                         $control=<<<CONTROL
 110                         [Process]
 111                         Type=Diff text
 112                         Engine=MediaWiki
 113                         Script={$wgServer}{$wgScript}
 114                         Special namespace={$special}
 115
 116                         [File]
 117                         Extension=wiki
 118                         URL=$url1
 119
 120                         [File 2]
 121                         Extension=wiki
 122                         URL=$url2
 123 CONTROL;
 124                         echo($control);
 125                         return;
 126                 }
 127
 128                 $wgOut->setArticleFlag( false );
 129                 if ( !$this->loadRevisionData() ) {
 130                         $t = $this->mTitle->getPrefixedText();
 131                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 132                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 133                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 134                         wfProfileOut( __METHOD__ );
 135                         return;
 136                 }
 137
 138                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 139
 140                 if ( $this->mNewRev->isCurrent() ) {
 141                         $wgOut->setArticleFlag( true );
 142                 }
 143
 144                 # mOldid is false if the difference engine is called with a "vague" query for
 145                 # a diff between a version V and its previous version V' AND the version V
 146                 # is the first version of that article. In that case, V' does not exist.
 147                 if ( $this->mOldid === false ) {
 148                         $this->showFirstRevision();
 149                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 150                         wfProfileOut( __METHOD__ );
 151                         return;
 152                 }
 153
 154                 $wgOut->suppressQuickbar();
 155
 156                 $oldTitle = $this->mOldPage->getPrefixedText();
 157                 $newTitle = $this->mNewPage->getPrefixedText();
 158                 if( $oldTitle == $newTitle ) {
 159                         $wgOut->setPageTitle( $newTitle );
 160                 } else {
 161                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 162                 }
 163                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 164                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 165
 166                 if ( !$this->mOldPage->userCanRead() || !$this->mNewPage->userCanRead() ) {
 167                         $wgOut->loginToUse();
 168                         $wgOut->output();
 169                         $wgOut->disable();
 170                         wfProfileOut( __METHOD__ );
 171                         return;
 172                 }
 173
 174                 $sk = $wgUser->getSkin();
 175
 176                 // Check if page is editable
 177                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 178                 if ( $editable && $this->mNewRev->isCurrent() && $wgUser->isAllowed( 'rollback' ) ) {
 179                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 180                 } else {
 181                         $rollback = '';
 182                 }
 183
 184                 // Prepare a change patrol link, if applicable
 185                 if( $wgUseRCPatrol && $this->mTitle->userCan('patrol') ) {
 186                         // If we've been given an explicit change identifier, use it; saves time
 187                         if( $this->mRcidMarkPatrolled ) {
 188                                 $rcid = $this->mRcidMarkPatrolled;
 189                                 $rc = RecentChange::newFromId( $rcid );
 190                                 // Already patrolled?
 191                                 $rcid = is_object($rc) && !$rc->getAttribute('rc_patrolled') ? $rcid : 0;
 192                         } else {
 193                                 // Look for an unpatrolled change corresponding to this diff
 194                                 $db = wfGetDB( DB_SLAVE );
 195                                 $change = RecentChange::newFromConds(
 196                                         array(
 197                                         // Redundant user,timestamp condition so we can use the existing index
 198                                                 'rc_user_text'  => $this->mNewRev->getRawUserText(),
 199                                                 'rc_timestamp'  => $db->timestamp( $this->mNewRev->getTimestamp() ),
 200                                                 'rc_this_oldid' => $this->mNewid,
 201                                                 'rc_last_oldid' => $this->mOldid,
 202                                                 'rc_patrolled'  => 0
 203                                         ),
 204                                         __METHOD__
 205                                 );
 206                                 if( $change instanceof RecentChange ) {
 207                                         $rcid = $change->mAttribs['rc_id'];
 208                                         $this->mRcidMarkPatrolled = $rcid;
 209                                 } else {
 210                                         // None found
 211                                         $rcid = 0;
 212                                 }
 213                         }
 214                         // Build the link
 215                         if( $rcid ) {
 216                                 $patrol = ' <span class="patrollink">[' . $sk->makeKnownLinkObj( $this->mTitle,
 217                                         wfMsgHtml( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$rcid}" ) . ']</span>';
 218                         } else {
 219                                 $patrol = '';
 220                         }
 221                 } else {
 222                         $patrol = '';
 223                 }
 224
 225                 $diffOnlyArg = '';
 226                 # Carry over 'diffonly' param via navigation links
 227                 if( $diffOnly != $wgUser->getBoolOption('diffonly') ) {
 228                         $diffOnlyArg = '&diffonly='.$diffOnly;
 229                 }
 230                 $htmldiffarg = $this->htmlDiffArgument();
 231                 # Make "previous revision link"
 232                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 233                         "diff=prev&oldid={$this->mOldid}{$htmldiffarg}{$diffOnlyArg}", '', '', 'id="differences-prevlink"' );
 234                 # Make "next revision link"
 235                 if( $this->mNewRev->isCurrent() ) {
 236                         $nextlink = '&nbsp;';
 237                 } else {
 238                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 239                                 "diff=next&oldid={$this->mNewid}{$htmldiffarg}{$diffOnlyArg}", '', '', 'id="differences-nextlink"' );
 240                 }
 241
 242                 $oldminor = '';
 243                 $newminor = '';
 244
 245                 if( $this->mOldRev->isMinor() ) {
 246                         $oldminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 247                 }
 248                 if( $this->mNewRev->isMinor() ) {
 249                         $newminor = Xml::span( wfMsg( 'minoreditletter' ), 'minor' ) . ' ';
 250                 }
 251
 252                 $rdel = ''; $ldel = '';
 253                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 254                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 255                                 // If revision was hidden from sysops
 256                                 $ldel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 257                         } else {
 258                                 $query = array(
 259                                         'type' => 'revision',
 260                                         'target' => $this->mOldRev->mTitle->getPrefixedDbkey(),
 261                                         'ids' => $this->mOldRev->getId()
 262                                 );
 263                                 $ldel = $sk->revDeleteLink( $query, $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 264                         }
 265                         $ldel = "&nbsp;&nbsp;&nbsp;$ldel ";
 266                         // We don't currently handle well changing the top revision's settings
 267                         if( $this->mNewRev->isCurrent() ) {
 268                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 269                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 270                                 // If revision was hidden from sysops
 271                                 $rdel = Xml::tags( 'span', array( 'class'=>'mw-revdelundel-link' ), '('.wfMsgHtml( 'rev-delundel' ).')' );
 272                         } else {
 273                                 $query = array(
 274                                         'type' => 'revision',
 275                                         'target' =>  $this->mNewRev->mTitle->getPrefixedDbkey(),
 276                                         'ids' => $this->mNewRev->getId()
 277                                 );
 278                                 $rdel = $sk->revDeleteLink( $query, $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) );
 279                         }
 280                         $rdel = "&nbsp;&nbsp;&nbsp;$rdel ";
 281                 }
 282
 283                 $oldHeader = '<div id="mw-diff-otitle1"><strong>'.$this->mOldtitle.'</strong></div>' .
 284                         '<div id="mw-diff-otitle2">' . $sk->revUserTools( $this->mOldRev, !$this->unhide ) . "</div>" .
 285                         '<div id="mw-diff-otitle3">' . $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, !$this->unhide ).$ldel."</div>" .
 286                         '<div id="mw-diff-otitle4">' . $prevlink .'</div>';
 287                 $newHeader = '<div id="mw-diff-ntitle1"><strong>'.$this->mNewtitle.'</strong></div>' .
 288                         '<div id="mw-diff-ntitle2">' . $sk->revUserTools( $this->mNewRev, !$this->unhide ) . " $rollback</div>" .
 289                         '<div id="mw-diff-ntitle3">' . $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, !$this->unhide ).$rdel."</div>" .
 290                         '<div id="mw-diff-ntitle4">' . $nextlink . $patrol . '</div>';
 291
 292                 # Check if this user can see the revisions
 293                 $allowed = $this->mOldRev->userCan(Revision::DELETED_TEXT)
 294                         && $this->mNewRev->userCan(Revision::DELETED_TEXT);
 295                 $deleted = $this->mOldRev->isDeleted(Revision::DELETED_TEXT)
 296                         || $this->mNewRev->isDeleted(Revision::DELETED_TEXT);
 297                 # Output the diff if allowed...
 298                 if( $deleted && (!$this->unhide || !$allowed) ) {
 299                         $this->showDiffStyle();
 300                         $multi = $this->getMultiNotice();
 301                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 302                         if( !$allowed ) {
 303                                 # Give explanation for why revision is not visible
 304                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 305                                         array( 'rev-deleted-no-diff' ) );
 306                         } else {
 307                                 # Give explanation and add a link to view the diff...
 308                                 $link = $this->mTitle->getFullUrl( "diff={$this->mNewid}&oldid={$this->mOldid}&unhide=1" );
 309                                 $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n",
 310                                         array( 'rev-deleted-unhide-diff', $link ) );
 311                         }
 312                 } else if( $wgEnableHtmlDiff && $this->htmldiff ) {
 313                         $multi = $this->getMultiNotice();
 314                         $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'wikicodecomparison' ),
 315                                 'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=0', '', '', 'id="differences-switchtype"' ).'</div>');
 316                         $wgOut->addHTML( $this->addHeader( '', $oldHeader, $newHeader, $multi ) );
 317                         $this->renderHtmlDiff();
 318                 } else {
 319                         if( $wgEnableHtmlDiff ) {
 320                                 $wgOut->addHTML('<div class="diff-switchtype">'.$sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'visualcomparison' ),
 321                                         'diff='.$this->mNewid.'&oldid='.$this->mOldid.'&htmldiff=1', '', '', 'id="differences-switchtype"' ).'</div>');
 322                         }
 323                         $this->showDiff( $oldHeader, $newHeader );
 324                         if( !$diffOnly ) {
 325                                 $this->renderNewRevision();
 326                         }
 327                 }
 328                 wfProfileOut( __METHOD__ );
 329         }
 330
 331         /**
 332          * Show the new revision of the page.
 333          */
 334         function renderNewRevision() {
 335                 global $wgOut, $wgUser;
 336                 wfProfileIn( __METHOD__ );
 337
 338                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 339                 # Add deleted rev tag if needed
 340                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 341                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 342                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 343                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 344                 }
 345
 346                 if( !$this->mNewRev->isCurrent() ) {
 347                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 348                 }
 349
 350                 $this->loadNewText();
 351                 if( is_object( $this->mNewRev ) ) {
 352                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 353                 }
 354
 355                 if( $this->mTitle->isCssJsSubpage() || $this->mTitle->isCssOrJsPage() ) {
 356                         // Stolen from Article::view --AG 2007-10-11
 357                         // Give hooks a chance to customise the output
 358                         if( wfRunHooks( 'ShowRawCssJs', array( $this->mNewtext, $this->mTitle, $wgOut ) ) ) {
 359                                 // Wrap the whole lot in a <pre> and don't parse
 360                                 $m = array();
 361                                 preg_match( '!\.(css|js)$!u', $this->mTitle->getText(), $m );
 362                                 $wgOut->addHTML( "<pre class=\"mw-code mw-{$m[1]}\" dir=\"ltr\">\n" );
 363                                 $wgOut->addHTML( htmlspecialchars( $this->mNewtext ) );
 364                                 $wgOut->addHTML( "\n</pre>\n" );
 365                         }
 366                 } else {
 367                         $wgOut->addWikiTextTidy( $this->mNewtext );
 368                 }
 369
 370                 if( is_object( $this->mNewRev ) && !$this->mNewRev->isCurrent() ) {
 371                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 372                 }
 373                 # Add redundant patrol link on bottom...
 374                 if( $this->mRcidMarkPatrolled && $this->mTitle->quickUserCan('patrol') ) {
 375                         $sk = $wgUser->getSkin();
 376                         $wgOut->addHTML(
 377                                 "<div class='patrollink'>[" . $sk->makeKnownLinkObj( $this->mTitle,
 378                                         wfMsgHtml( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) .
 379                                 ']</div>'
 380                          );
 381                 }
 382
 383                 wfProfileOut( __METHOD__ );
 384         }
 385
 386
 387         function renderHtmlDiff() {
 388                 global $wgOut, $wgParser, $wgDebugComments;
 389                 wfProfileIn( __METHOD__ );
 390
 391                 $this->showDiffStyle();
 392
 393                 $wgOut->addHTML( '<h2>'.wfMsgHtml( 'visual-comparison' )."</h2>\n" );
 394                 #add deleted rev tag if needed
 395                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 396                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-permission' );
 397                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 398                         $wgOut->wrapWikiMsg( "<div class='mw-warning plainlinks'>\n$1</div>\n", 'rev-deleted-text-view' );
 399                 }
 400
 401                 if( !$this->mNewRev->isCurrent() ) {
 402                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 403                 }
 404
 405                 $this->loadText();
 406
 407                 // Old revision
 408                 if( is_object( $this->mOldRev ) ) {
 409                         $wgOut->setRevisionId( $this->mOldRev->getId() );
 410                 }
 411
 412                 $popts = $wgOut->parserOptions();
 413                 $oldTidy = $popts->setTidy( true );
 414                 $popts->setEditSection( false );
 415
 416                 $parserOutput = $wgParser->parse( $this->mOldtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 417                 $popts->setTidy( $oldTidy );
 418
 419                 //only for new?
 420                 //$wgOut->addParserOutputNoText( $parserOutput );
 421                 $oldHtml = $parserOutput->getText();
 422                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$oldHtml ) );
 423
 424                 // New revision
 425                 if( is_object( $this->mNewRev ) ) {
 426                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 427                 }
 428
 429                 $popts = $wgOut->parserOptions();
 430                 $oldTidy = $popts->setTidy( true );
 431
 432                 $parserOutput = $wgParser->parse( $this->mNewtext, $this->getTitle(), $popts, true, true, $wgOut->getRevisionId() );
 433                 $popts->setTidy( $oldTidy );
 434
 435                 $wgOut->addParserOutputNoText( $parserOutput );
 436                 $newHtml = $parserOutput->getText();
 437                 wfRunHooks( 'OutputPageBeforeHTML', array( &$wgOut, &$newHtml ) );
 438
 439                 unset($parserOutput, $popts);
 440
 441                 $differ = new HTMLDiffer(new DelegatingContentHandler($wgOut));
 442                 $differ->htmlDiff($oldHtml, $newHtml);
 443                 if ( $wgDebugComments ) {
 444                         $wgOut->addHTML( "\n<!-- HtmlDiff Debug Output:\n" . HTMLDiffer::getDebugOutput() . " End Debug -->" );
 445                 }
 446
 447                 wfProfileOut( __METHOD__ );
 448         }
 449
 450         /**
 451          * Show the first revision of an article. Uses normal diff headers in
 452          * contrast to normal "old revision" display style.
 453          */
 454         function showFirstRevision() {
 455                 global $wgOut, $wgUser;
 456                 wfProfileIn( __METHOD__ );
 457
 458                 # Get article text from the DB
 459                 #
 460                 if ( ! $this->loadNewText() ) {
 461                         $t = $this->mTitle->getPrefixedText();
 462                         $d = wfMsgExt( 'missingarticle-diff', array( 'escape' ), $this->mOldid, $this->mNewid );
 463                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 464                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>$t</nowiki>", $d );
 465                         wfProfileOut( __METHOD__ );
 466                         return;
 467                 }
 468                 if ( $this->mNewRev->isCurrent() ) {
 469                         $wgOut->setArticleFlag( true );
 470                 }
 471
 472                 # Check if user is allowed to look at this page. If not, bail out.
 473                 #
 474                 if ( !$this->mTitle->userCanRead() ) {
 475                         $wgOut->loginToUse();
 476                         $wgOut->output();
 477                         wfProfileOut( __METHOD__ );
 478                         throw new MWException("Permission Error: you do not have access to view this page");
 479                 }
 480
 481                 # Prepare the header box
 482                 #
 483                 $sk = $wgUser->getSkin();
 484
 485                 $next = $this->mTitle->getNextRevisionID( $this->mNewid );
 486                 if( !$next ) {
 487                         $nextlink = '';
 488                 } else {
 489                         $nextlink = '<br/>' . $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 490                                 'diff=next&oldid=' . $this->mNewid.$this->htmlDiffArgument(), '', '', 'id="differences-nextlink"' );
 491                 }
 492                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\">" .
 493                         $sk->revUserTools( $this->mNewRev ) . "<br/>" . $sk->revComment( $this->mNewRev ) . $nextlink . "</div>\n";
 494
 495                 $wgOut->addHTML( $header );
 496
 497                 $wgOut->setSubtitle( wfMsgExt( 'difference', array( 'parseinline' ) ) );
 498                 $wgOut->setRobotPolicy( 'noindex,nofollow' );
 499
 500                 wfProfileOut( __METHOD__ );
 501         }
 502
 503         function htmlDiffArgument(){
 504                 global $wgEnableHtmlDiff;
 505                 if($wgEnableHtmlDiff){
 506                         if($this->htmldiff){
 507                                 return '&htmldiff=1';
 508                         }else{
 509                                 return '&htmldiff=0';
 510                         }
 511                 }else{
 512                         return '';
 513                 }
 514         }
 515
 516         /**
 517          * Get the diff text, send it to $wgOut
 518          * Returns false if the diff could not be generated, otherwise returns true
 519          */
 520         function showDiff( $otitle, $ntitle ) {
 521                 global $wgOut;
 522                 $diff = $this->getDiff( $otitle, $ntitle );
 523                 if ( $diff === false ) {
 524                         $wgOut->addWikiMsg( 'missing-article', "<nowiki>(fixme, bug)</nowiki>", '' );
 525                         return false;
 526                 } else {
 527                         $this->showDiffStyle();
 528                         $wgOut->addHTML( $diff );
 529                         return true;
 530                 }
 531         }
 532
 533         /**
 534          * Add style sheets and supporting JS for diff display.
 535          */
 536         function showDiffStyle() {
 537                 global $wgStylePath, $wgStyleVersion, $wgOut;
 538                 $wgOut->addStyle( 'common/diff.css' );
 539
 540                 // JS is needed to detect old versions of Mozilla to work around an annoyance bug.
 541                 $wgOut->addScript( "<script type=\"text/javascript\" src=\"$wgStylePath/common/diff.js?$wgStyleVersion\"></script>" );
 542         }
 543
 544         /**
 545          * Get complete diff table, including header
 546          *
 547          * @param Title $otitle Old title
 548          * @param Title $ntitle New title
 549          * @return mixed
 550          */
 551         function getDiff( $otitle, $ntitle ) {
 552                 $body = $this->getDiffBody();
 553                 if ( $body === false ) {
 554                         return false;
 555                 } else {
 556                         $multi = $this->getMultiNotice();
 557                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 558                 }
 559         }
 560
 561         /**
 562          * Get the diff table body, without header
 563          *
 564          * @return mixed
 565          */
 566         function getDiffBody() {
 567                 global $wgMemc;
 568                 wfProfileIn( __METHOD__ );
 569                 $this->mCacheHit = true;
 570                 // Check if the diff should be hidden from this user
 571                 if ( !$this->loadRevisionData() )
 572                         return '';
 573                 if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 574                         return '';
 575                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 576                         return '';
 577                 } else if ( $this->mOldRev && $this->mNewRev && $this->mOldRev->getID() == $this->mNewRev->getID() ) {
 578                         return '';
 579                 }
 580                 // Cacheable?
 581                 $key = false;
 582                 if ( $this->mOldid && $this->mNewid ) {
 583                         $key = wfMemcKey( 'diff', 'version', MW_DIFF_VERSION, 'oldid', $this->mOldid, 'newid', $this->mNewid );
 584                         // Try cache
 585                         if ( !$this->mRefreshCache ) {
 586                                 $difftext = $wgMemc->get( $key );
 587                                 if ( $difftext ) {
 588                                         wfIncrStats( 'diff_cache_hit' );
 589                                         $difftext = $this->localiseLineNumbers( $difftext );
 590                                         $difftext .= "\n<!-- diff cache key $key -->\n";
 591                                         wfProfileOut( __METHOD__ );
 592                                         return $difftext;
 593                                 }
 594                         } // don't try to load but save the result
 595                 }
 596                 $this->mCacheHit = false;
 597
 598                 // Loadtext is permission safe, this just clears out the diff
 599                 if ( !$this->loadText() ) {
 600                         wfProfileOut( __METHOD__ );
 601                         return false;
 602                 }
 603
 604                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 605
 606                 // Save to cache for 7 days
 607                 if ( !wfRunHooks( 'AbortDiffCache', array( &$this ) ) ) {
 608                         wfIncrStats( 'diff_uncacheable' );
 609                 } else if ( $key !== false && $difftext !== false ) {
 610                         wfIncrStats( 'diff_cache_miss' );
 611                         $wgMemc->set( $key, $difftext, 7*86400 );
 612                 } else {
 613                         wfIncrStats( 'diff_uncacheable' );
 614                 }
 615                 // Replace line numbers with the text in the user's language
 616                 if ( $difftext !== false ) {
 617                         $difftext = $this->localiseLineNumbers( $difftext );
 618                 }
 619                 wfProfileOut( __METHOD__ );
 620                 return $difftext;
 621         }
 622
 623         /**
 624          * Make sure the proper modules are loaded before we try to
 625          * make the diff
 626          */
 627         private function initDiffEngines() {
 628                 global $wgExternalDiffEngine;
 629                 if ( $wgExternalDiffEngine == 'wikidiff' && !function_exists( 'wikidiff_do_diff' ) ) {
 630                         wfProfileIn( __METHOD__ . '-php_wikidiff.so' );
 631                         wfSuppressWarnings();
 632                         dl( 'php_wikidiff.so' );
 633                         wfRestoreWarnings();
 634                         wfProfileOut( __METHOD__ . '-php_wikidiff.so' );
 635                 }
 636                 else if ( $wgExternalDiffEngine == 'wikidiff2' && !function_exists( 'wikidiff2_do_diff' ) ) {
 637                         wfProfileIn( __METHOD__ . '-php_wikidiff2.so' );
 638                         wfSuppressWarnings();
 639                         dl( 'php_wikidiff2.so' );
 640                         wfRestoreWarnings();
 641                         wfProfileOut( __METHOD__ . '-php_wikidiff2.so' );
 642                 }
 643         }
 644
 645         /**
 646          * Generate a diff, no caching
 647          * $otext and $ntext must be already segmented
 648          */
 649         function generateDiffBody( $otext, $ntext ) {
 650                 global $wgExternalDiffEngine, $wgContLang;
 651
 652                 $otext = str_replace( "\r\n", "\n", $otext );
 653                 $ntext = str_replace( "\r\n", "\n", $ntext );
 654
 655                 $this->initDiffEngines();
 656
 657                 if ( $wgExternalDiffEngine == 'wikidiff' && function_exists( 'wikidiff_do_diff' ) ) {
 658                         # For historical reasons, external diff engine expects
 659                         # input text to be HTML-escaped already
 660                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 661                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 662                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) ) .
 663                         $this->debug( 'wikidiff1' );
 664                 }
 665
 666                 if ( $wgExternalDiffEngine == 'wikidiff2' && function_exists( 'wikidiff2_do_diff' ) ) {
 667                         # Better external diff engine, the 2 may some day be dropped
 668                         # This one does the escaping and segmenting itself
 669                         wfProfileIn( 'wikidiff2_do_diff' );
 670                         $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 671                         $text .= $this->debug( 'wikidiff2' );
 672                         wfProfileOut( 'wikidiff2_do_diff' );
 673                         return $text;
 674                 }
 675                 if ( $wgExternalDiffEngine != 'wikidiff3' && $wgExternalDiffEngine !== false ) {
 676                         # Diff via the shell
 677                         global $wgTmpDirectory;
 678                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 679                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 680
 681                         $tempFile1 = fopen( $tempName1, "w" );
 682                         if ( !$tempFile1 ) {
 683                                 wfProfileOut( __METHOD__ );
 684                                 return false;
 685                         }
 686                         $tempFile2 = fopen( $tempName2, "w" );
 687                         if ( !$tempFile2 ) {
 688                                 wfProfileOut( __METHOD__ );
 689                                 return false;
 690                         }
 691                         fwrite( $tempFile1, $otext );
 692                         fwrite( $tempFile2, $ntext );
 693                         fclose( $tempFile1 );
 694                         fclose( $tempFile2 );
 695                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 696                         wfProfileIn( __METHOD__ . "-shellexec" );
 697                         $difftext = wfShellExec( $cmd );
 698                         $difftext .= $this->debug( "external $wgExternalDiffEngine" );
 699                         wfProfileOut( __METHOD__ . "-shellexec" );
 700                         unlink( $tempName1 );
 701                         unlink( $tempName2 );
 702                         return $difftext;
 703                 }
 704
 705                 # Native PHP diff
 706                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 707                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 708                 $diffs = new Diff( $ota, $nta );
 709                 $formatter = new TableDiffFormatter();
 710                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) ) .
 711                 $this->debug();
 712         }
 713
 714         /**
 715          * Generate a debug comment indicating diff generating time,
 716          * server node, and generator backend.
 717          */
 718         protected function debug( $generator="internal" ) {
 719                 global $wgShowHostnames;
 720                 $data = array( $generator );
 721                 if( $wgShowHostnames ) {
 722                         $data[] = wfHostname();
 723                 }
 724                 $data[] = wfTimestamp( TS_DB );
 725                 return "<!-- diff generator: " .
 726                 implode( " ",
 727                 array_map(
 728                                         "htmlspecialchars",
 729                 $data ) ) .
 730                         " -->\n";
 731         }
 732
 733         /**
 734          * Replace line numbers with the text in the user's language
 735          */
 736         function localiseLineNumbers( $text ) {
 737                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 738                 array( &$this, 'localiseLineNumbersCb' ), $text );
 739         }
 740
 741         function localiseLineNumbersCb( $matches ) {
 742                 global $wgLang;
 743                 return wfMsgExt( 'lineno', 'escape', $wgLang->formatNum( $matches[1] ) );
 744         }
 745
 746
 747         /**
 748          * If there are revisions between the ones being compared, return a note saying so.
 749          */
 750         function getMultiNotice() {
 751                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 752                 return '';
 753
 754                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 755                         // Comparing two different pages? Count would be meaningless.
 756                         return '';
 757                 }
 758
 759                 $oldid = $this->mOldRev->getId();
 760                 $newid = $this->mNewRev->getId();
 761                 if ( $oldid > $newid ) {
 762                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 763                 }
 764
 765                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 766                 if ( !$n )
 767                 return '';
 768
 769                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 770         }
 771
 772
 773         /**
 774          * Add the header to a diff body
 775          */
 776         static function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 777                 $colspan = 1;
 778                 $header = "<table class='diff'>";
 779                 if( $diff ) { // Safari/Chrome show broken output if cols not used
 780                         $header .= "
 781                         <col class='diff-marker' />
 782                         <col class='diff-content' />
 783                         <col class='diff-marker' />
 784                         <col class='diff-content' />";
 785                         $colspan = 2;
 786                 }
 787                 $header .= "
 788                 <tr valign='top'>
 789                 <td colspan='$colspan' class='diff-otitle'>{$otitle}</td>
 790                 <td colspan='$colspan' class='diff-ntitle'>{$ntitle}</td>
 791                 </tr>";
 792
 793                 if ( $multi != '' )
 794                 $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 795
 796                 return $header . $diff . "</table>";
 797         }
 798
 799         /**
 800          * Use specified text instead of loading from the database
 801          */
 802         function setText( $oldText, $newText ) {
 803                 $this->mOldtext = $oldText;
 804                 $this->mNewtext = $newText;
 805                 $this->mTextLoaded = 2;
 806                 $this->mRevisionsLoaded = true;
 807         }
 808
 809         /**
 810          * Load revision metadata for the specified articles. If newid is 0, then compare
 811          * the old article in oldid to the current article; if oldid is 0, then
 812          * compare the current article to the immediately previous one (ignoring the
 813          * value of newid).
 814          *
 815          * If oldid is false, leave the corresponding revision object set
 816          * to false. This is impossible via ordinary user input, and is provided for
 817          * API convenience.
 818          */
 819         function loadRevisionData() {
 820                 global $wgLang, $wgUser;
 821                 if ( $this->mRevisionsLoaded ) {
 822                         return true;
 823                 } else {
 824                         // Whether it succeeds or fails, we don't want to try again
 825                         $this->mRevisionsLoaded = true;
 826                 }
 827
 828                 // Load the new revision object
 829                 $this->mNewRev = $this->mNewid
 830                         ? Revision::newFromId( $this->mNewid )
 831                         : Revision::newFromTitle( $this->mTitle );
 832                 if( !$this->mNewRev instanceof Revision )
 833                         return false;
 834
 835                 // Update the new revision ID in case it was 0 (makes life easier doing UI stuff)
 836                 $this->mNewid = $this->mNewRev->getId();
 837
 838                 // Check if page is editable
 839                 $editable = $this->mNewRev->getTitle()->userCan( 'edit' );
 840
 841                 // Set assorted variables
 842                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 843                 $this->mNewPage = $this->mNewRev->getTitle();
 844                 if( $this->mNewRev->isCurrent() ) {
 845                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 846                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev-asof', $timestamp ) );
 847                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 848
 849                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 850                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 851                 } else {
 852                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 853                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 854                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 855
 856                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 857                         $this->mNewtitle .= " (<a href='$newEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 858                 }
 859                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 860                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 861                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 862                         $this->mNewtitle = "<span class='history-deleted'>{$this->mNewtitle}</span>";
 863                 }
 864
 865                 // Load the old revision object
 866                 $this->mOldRev = false;
 867                 if( $this->mOldid ) {
 868                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 869                 } elseif ( $this->mOldid === 0 ) {
 870                         $rev = $this->mNewRev->getPrevious();
 871                         if( $rev ) {
 872                                 $this->mOldid = $rev->getId();
 873                                 $this->mOldRev = $rev;
 874                         } else {
 875                                 // No previous revision; mark to show as first-version only.
 876                                 $this->mOldid = false;
 877                                 $this->mOldRev = false;
 878                         }
 879                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 880
 881                 if( is_null( $this->mOldRev ) ) {
 882                         return false;
 883                 }
 884
 885                 if ( $this->mOldRev ) {
 886                         $this->mOldPage = $this->mOldRev->getTitle();
 887
 888                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 889                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 890                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 891                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 892
 893                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 894                         . " (<a href='$oldEdit'>" . wfMsgHtml( $editable ? 'editold' : 'viewsourceold' ) . "</a>)";
 895                         // Add an "undo" link
 896                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
 897                         $htmlLink = htmlspecialchars( wfMsg( 'editundo' ) );
 898                         $htmlTitle = $wgUser->getSkin()->tooltip( 'undo' );
 899                         if( $editable && !$this->mOldRev->isDeleted( Revision::DELETED_TEXT ) && !$this->mNewRev->isDeleted( Revision::DELETED_TEXT ) ) {
 900                                 $this->mNewtitle .= " (<a href='$newUndo' $htmlTitle>" . $htmlLink . "</a>)";
 901                         }
 902
 903                         if( !$this->mOldRev->userCan( Revision::DELETED_TEXT ) ) {
 904                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldPagetitle . '</span>';
 905                         } else if( $this->mOldRev->isDeleted( Revision::DELETED_TEXT ) ) {
 906                                 $this->mOldtitle = '<span class="history-deleted">' . $this->mOldtitle . '</span>';
 907                         }
 908                 }
 909
 910                 return true;
 911         }
 912
 913         /**
 914          * Load the text of the revisions, as well as revision data.
 915          */
 916         function loadText() {
 917                 if ( $this->mTextLoaded == 2 ) {
 918                         return true;
 919                 } else {
 920                         // Whether it succeeds or fails, we don't want to try again
 921                         $this->mTextLoaded = 2;
 922                 }
 923
 924                 if ( !$this->loadRevisionData() ) {
 925                         return false;
 926                 }
 927                 if ( $this->mOldRev ) {
 928                         $this->mOldtext = $this->mOldRev->getText( Revision::FOR_THIS_USER );
 929                         if ( $this->mOldtext === false ) {
 930                                 return false;
 931                         }
 932                 }
 933                 if ( $this->mNewRev ) {
 934                         $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 935                         if ( $this->mNewtext === false ) {
 936                                 return false;
 937                         }
 938                 }
 939                 return true;
 940         }
 941
 942         /**
 943          * Load the text of the new revision, not the old one
 944          */
 945         function loadNewText() {
 946                 if ( $this->mTextLoaded >= 1 ) {
 947                         return true;
 948                 } else {
 949                         $this->mTextLoaded = 1;
 950                 }
 951                 if ( !$this->loadRevisionData() ) {
 952                         return false;
 953                 }
 954                 $this->mNewtext = $this->mNewRev->getText( Revision::FOR_THIS_USER );
 955                 return true;
 956         }
 957
 958
 959 }
 960
 961 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 962 //
 963 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 964 // You may copy this code freely under the conditions of the GPL.
 965 //
 966
 967 define('USE_ASSERTS', function_exists('assert'));
 968
 969 /**
 970  * @todo document
 971  * @private
 972  * @ingroup DifferenceEngine
 973  */
 974 class _DiffOp {
 975         var $type;
 976         var $orig;
 977         var $closing;
 978
 979         function reverse() {
 980                 trigger_error('pure virtual', E_USER_ERROR);
 981         }
 982
 983         function norig() {
 984                 return $this->orig ? sizeof($this->orig) : 0;
 985         }
 986
 987         function nclosing() {
 988                 return $this->closing ? sizeof($this->closing) : 0;
 989         }
 990 }
 991
 992 /**
 993  * @todo document
 994  * @private
 995  * @ingroup DifferenceEngine
 996  */
 997 class _DiffOp_Copy extends _DiffOp {
 998         var $type = 'copy';
 999
1000         function _DiffOp_Copy ($orig, $closing = false) {
1001                 if (!is_array($closing))
1002                 $closing = $orig;
1003                 $this->orig = $orig;
1004                 $this->closing = $closing;
1005         }
1006
1007         function reverse() {
1008                 return new _DiffOp_Copy($this->closing, $this->orig);
1009         }
1010 }
1011
1012 /**
1013  * @todo document
1014  * @private
1015  * @ingroup DifferenceEngine
1016  */
1017 class _DiffOp_Delete extends _DiffOp {
1018         var $type = 'delete';
1019
1020         function _DiffOp_Delete ($lines) {
1021                 $this->orig = $lines;
1022                 $this->closing = false;
1023         }
1024
1025         function reverse() {
1026                 return new _DiffOp_Add($this->orig);
1027         }
1028 }
1029
1030 /**
1031  * @todo document
1032  * @private
1033  * @ingroup DifferenceEngine
1034  */
1035 class _DiffOp_Add extends _DiffOp {
1036         var $type = 'add';
1037
1038         function _DiffOp_Add ($lines) {
1039                 $this->closing = $lines;
1040                 $this->orig = false;
1041         }
1042
1043         function reverse() {
1044                 return new _DiffOp_Delete($this->closing);
1045         }
1046 }
1047
1048 /**
1049  * @todo document
1050  * @private
1051  * @ingroup DifferenceEngine
1052  */
1053 class _DiffOp_Change extends _DiffOp {
1054         var $type = 'change';
1055
1056         function _DiffOp_Change ($orig, $closing) {
1057                 $this->orig = $orig;
1058                 $this->closing = $closing;
1059         }
1060
1061         function reverse() {
1062                 return new _DiffOp_Change($this->closing, $this->orig);
1063         }
1064 }
1065
1066 /**
1067  * Class used internally by Diff to actually compute the diffs.
1068  *
1069  * The algorithm used here is mostly lifted from the perl module
1070  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
1071  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
1072  *
1073  * More ideas are taken from:
1074  *       http://www.ics.uci.edu/~eppstein/161/960229.html
1075  *
1076  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
1077  * diffutils-2.7, which can be found at:
1078  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
1079  *
1080  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
1081  * are my own.
1082  *
1083  * Line length limits for robustness added by Tim Starling, 2005-08-31
1084  * Alternative implementation added by Guy Van den Broeck, 2008-07-30
1085  *
1086  * @author Geoffrey T. Dairiki, Tim Starling, Guy Van den Broeck
1087  * @private
1088  * @ingroup DifferenceEngine
1089  */
1090 class _DiffEngine {
1091
1092         const MAX_XREF_LENGTH =  10000;
1093
1094         function diff ($from_lines, $to_lines){
1095                 wfProfileIn( __METHOD__ );
1096
1097                 // Diff and store locally
1098                 $this->diff_local($from_lines, $to_lines);
1099
1100                 // Merge edits when possible
1101                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
1102                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
1103
1104                 // Compute the edit operations.
1105                 $n_from = sizeof($from_lines);
1106                 $n_to = sizeof($to_lines);
1107
1108                 $edits = array();
1109                 $xi = $yi = 0;
1110                 while ($xi < $n_from || $yi < $n_to) {
1111                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
1112                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
1113
1114                         // Skip matching "snake".
1115                         $copy = array();
1116                         while ( $xi < $n_from && $yi < $n_to
1117                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
1118                                 $copy[] = $from_lines[$xi++];
1119                                 ++$yi;
1120                         }
1121                         if ($copy)
1122                         $edits[] = new _DiffOp_Copy($copy);
1123
1124                         // Find deletes & adds.
1125                         $delete = array();
1126                         while ($xi < $n_from && $this->xchanged[$xi])
1127                         $delete[] = $from_lines[$xi++];
1128
1129                         $add = array();
1130                         while ($yi < $n_to && $this->ychanged[$yi])
1131                         $add[] = $to_lines[$yi++];
1132
1133                         if ($delete && $add)
1134                         $edits[] = new _DiffOp_Change($delete, $add);
1135                         elseif ($delete)
1136                         $edits[] = new _DiffOp_Delete($delete);
1137                         elseif ($add)
1138                         $edits[] = new _DiffOp_Add($add);
1139                 }
1140                 wfProfileOut( __METHOD__ );
1141                 return $edits;
1142         }
1143
1144         function diff_local ($from_lines, $to_lines) {
1145                 global $wgExternalDiffEngine;
1146                 wfProfileIn( __METHOD__);
1147
1148                 if($wgExternalDiffEngine == 'wikidiff3'){
1149                         // wikidiff3
1150                         $wikidiff3 = new WikiDiff3();
1151                         $wikidiff3->diff($from_lines, $to_lines);
1152                         $this->xchanged = $wikidiff3->removed;
1153                         $this->ychanged = $wikidiff3->added;
1154                         unset($wikidiff3);
1155                 }else{
1156                         // old diff
1157                         $n_from = sizeof($from_lines);
1158                         $n_to = sizeof($to_lines);
1159                         $this->xchanged = $this->ychanged = array();
1160                         $this->xv = $this->yv = array();
1161                         $this->xind = $this->yind = array();
1162                         unset($this->seq);
1163                         unset($this->in_seq);
1164                         unset($this->lcs);
1165
1166                         // Skip leading common lines.
1167                         for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
1168                                 if ($from_lines[$skip] !== $to_lines[$skip])
1169                                 break;
1170                                 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
1171                         }
1172                         // Skip trailing common lines.
1173                         $xi = $n_from; $yi = $n_to;
1174                         for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
1175                                 if ($from_lines[$xi] !== $to_lines[$yi])
1176                                 break;
1177                                 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
1178                         }
1179
1180                         // Ignore lines which do not exist in both files.
1181                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1182                                 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
1183                         }
1184
1185                         for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
1186                                 $line = $to_lines[$yi];
1187                                 if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
1188                                 continue;
1189                                 $yhash[$this->_line_hash($line)] = 1;
1190                                 $this->yv[] = $line;
1191                                 $this->yind[] = $yi;
1192                         }
1193                         for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
1194                                 $line = $from_lines[$xi];
1195                                 if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
1196                                 continue;
1197                                 $this->xv[] = $line;
1198                                 $this->xind[] = $xi;
1199                         }
1200
1201                         // Find the LCS.
1202                         $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
1203                 }
1204                 wfProfileOut( __METHOD__ );
1205         }
1206
1207         /**
1208          * Returns the whole line if it's small enough, or the MD5 hash otherwise
1209          */
1210         function _line_hash( $line ) {
1211                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
1212                         return md5( $line );
1213                 } else {
1214                         return $line;
1215                 }
1216         }
1217
1218         /* Divide the Largest Common Subsequence (LCS) of the sequences
1219          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
1220          * sized segments.
1221          *
1222          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
1223          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
1224          * sub sequences.  The first sub-sequence is contained in [X0, X1),
1225          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
1226          * that (X0, Y0) == (XOFF, YOFF) and
1227          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
1228          *
1229          * This function assumes that the first lines of the specified portions
1230          * of the two files do not match, and likewise that the last lines do not
1231          * match.  The caller must trim matching lines from the beginning and end
1232          * of the portions it is going to specify.
1233          */
1234         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
1235                 $flip = false;
1236
1237                 if ($xlim - $xoff > $ylim - $yoff) {
1238                         // Things seems faster (I'm not sure I understand why)
1239                         // when the shortest sequence in X.
1240                         $flip = true;
1241                         list ($xoff, $xlim, $yoff, $ylim)
1242                         = array( $yoff, $ylim, $xoff, $xlim);
1243                 }
1244
1245                 if ($flip)
1246                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1247                 $ymatches[$this->xv[$i]][] = $i;
1248                 else
1249                 for ($i = $ylim - 1; $i >= $yoff; $i--)
1250                 $ymatches[$this->yv[$i]][] = $i;
1251
1252                 $this->lcs = 0;
1253                 $this->seq[0]= $yoff - 1;
1254                 $this->in_seq = array();
1255                 $ymids[0] = array();
1256
1257                 $numer = $xlim - $xoff + $nchunks - 1;
1258                 $x = $xoff;
1259                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
1260                         if ($chunk > 0)
1261                         for ($i = 0; $i <= $this->lcs; $i++)
1262                         $ymids[$i][$chunk-1] = $this->seq[$i];
1263
1264                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
1265                         for ( ; $x < $x1; $x++) {
1266                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
1267                                 if (empty($ymatches[$line]))
1268                                 continue;
1269                                 $matches = $ymatches[$line];
1270                                 reset($matches);
1271                                 while (list ($junk, $y) = each($matches))
1272                                 if (empty($this->in_seq[$y])) {
1273                                         $k = $this->_lcs_pos($y);
1274                                         USE_ASSERTS && assert($k > 0);
1275                                         $ymids[$k] = $ymids[$k-1];
1276                                         break;
1277                                 }
1278                                 while (list ( /* $junk */, $y) = each($matches)) {
1279                                         if ($y > $this->seq[$k-1]) {
1280                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1281                                                 // Optimization: this is a common case:
1282                                                 //      next match is just replacing previous match.
1283                                                 $this->in_seq[$this->seq[$k]] = false;
1284                                                 $this->seq[$k] = $y;
1285                                                 $this->in_seq[$y] = 1;
1286                                         } else if (empty($this->in_seq[$y])) {
1287                                                 $k = $this->_lcs_pos($y);
1288                                                 USE_ASSERTS && assert($k > 0);
1289                                                 $ymids[$k] = $ymids[$k-1];
1290                                         }
1291                                 }
1292                         }
1293                 }
1294
1295                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1296                 $ymid = $ymids[$this->lcs];
1297                 for ($n = 0; $n < $nchunks - 1; $n++) {
1298                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1299                         $y1 = $ymid[$n] + 1;
1300                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1301                 }
1302                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1303
1304                 return array($this->lcs, $seps);
1305         }
1306
1307         function _lcs_pos ($ypos) {
1308                 $end = $this->lcs;
1309                 if ($end == 0 || $ypos > $this->seq[$end]) {
1310                         $this->seq[++$this->lcs] = $ypos;
1311                         $this->in_seq[$ypos] = 1;
1312                         return $this->lcs;
1313                 }
1314
1315                 $beg = 1;
1316                 while ($beg < $end) {
1317                         $mid = (int)(($beg + $end) / 2);
1318                         if ( $ypos > $this->seq[$mid] )
1319                         $beg = $mid + 1;
1320                         else
1321                         $end = $mid;
1322                 }
1323
1324                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1325
1326                 $this->in_seq[$this->seq[$end]] = false;
1327                 $this->seq[$end] = $ypos;
1328                 $this->in_seq[$ypos] = 1;
1329                 return $end;
1330         }
1331
1332         /* Find LCS of two sequences.
1333          *
1334          * The results are recorded in the vectors $this->{x,y}changed[], by
1335          * storing a 1 in the element for each line that is an insertion
1336          * or deletion (ie. is not in the LCS).
1337          *
1338          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1339          *
1340          * Note that XLIM, YLIM are exclusive bounds.
1341          * All line numbers are origin-0 and discarded lines are not counted.
1342          */
1343         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1344                 // Slide down the bottom initial diagonal.
1345                 while ($xoff < $xlim && $yoff < $ylim
1346                 && $this->xv[$xoff] == $this->yv[$yoff]) {
1347                         ++$xoff;
1348                         ++$yoff;
1349                 }
1350
1351                 // Slide up the top initial diagonal.
1352                 while ($xlim > $xoff && $ylim > $yoff
1353                 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1354                         --$xlim;
1355                         --$ylim;
1356                 }
1357
1358                 if ($xoff == $xlim || $yoff == $ylim)
1359                 $lcs = 0;
1360                 else {
1361                         // This is ad hoc but seems to work well.
1362                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1363                         //$nchunks = max(2,min(8,(int)$nchunks));
1364                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1365                         list ($lcs, $seps)
1366                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1367                 }
1368
1369                 if ($lcs == 0) {
1370                         // X and Y sequences have no common subsequence:
1371                         // mark all changed.
1372                         while ($yoff < $ylim)
1373                         $this->ychanged[$this->yind[$yoff++]] = 1;
1374                         while ($xoff < $xlim)
1375                         $this->xchanged[$this->xind[$xoff++]] = 1;
1376                 } else {
1377                         // Use the partitions to split this problem into subproblems.
1378                         reset($seps);
1379                         $pt1 = $seps[0];
1380                         while ($pt2 = next($seps)) {
1381                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1382                                 $pt1 = $pt2;
1383                         }
1384                 }
1385         }
1386
1387         /* Adjust inserts/deletes of identical lines to join changes
1388          * as much as possible.
1389          *
1390          * We do something when a run of changed lines include a
1391          * line at one end and has an excluded, identical line at the other.
1392          * We are free to choose which identical line is included.
1393          * `compareseq' usually chooses the one at the beginning,
1394          * but usually it is cleaner to consider the following identical line
1395          * to be the "change".
1396          *
1397          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1398          */
1399         function _shift_boundaries ($lines, &$changed, $other_changed) {
1400                 wfProfileIn( __METHOD__ );
1401                 $i = 0;
1402                 $j = 0;
1403
1404                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1405                 $len = sizeof($lines);
1406                 $other_len = sizeof($other_changed);
1407
1408                 while (1) {
1409                         /*
1410                          * Scan forwards to find beginning of another run of changes.
1411                          * Also keep track of the corresponding point in the other file.
1412                          *
1413                          * Throughout this code, $i and $j are adjusted together so that
1414                          * the first $i elements of $changed and the first $j elements
1415                          * of $other_changed both contain the same number of zeros
1416                          * (unchanged lines).
1417                          * Furthermore, $j is always kept so that $j == $other_len or
1418                          * $other_changed[$j] == false.
1419                          */
1420                         while ($j < $other_len && $other_changed[$j])
1421                         $j++;
1422
1423                         while ($i < $len && ! $changed[$i]) {
1424                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1425                                 $i++; $j++;
1426                                 while ($j < $other_len && $other_changed[$j])
1427                                 $j++;
1428                         }
1429
1430                         if ($i == $len)
1431                         break;
1432
1433                         $start = $i;
1434
1435                         // Find the end of this run of changes.
1436                         while (++$i < $len && $changed[$i])
1437                         continue;
1438
1439                         do {
1440                                 /*
1441                                  * Record the length of this run of changes, so that
1442                                  * we can later determine whether the run has grown.
1443                                  */
1444                                 $runlength = $i - $start;
1445
1446                                 /*
1447                                  * Move the changed region back, so long as the
1448                                  * previous unchanged line matches the last changed one.
1449                                  * This merges with previous changed regions.
1450                                  */
1451                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1452                                         $changed[--$start] = 1;
1453                                         $changed[--$i] = false;
1454                                         while ($start > 0 && $changed[$start - 1])
1455                                         $start--;
1456                                         USE_ASSERTS && assert('$j > 0');
1457                                         while ($other_changed[--$j])
1458                                         continue;
1459                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1460                                 }
1461
1462                                 /*
1463                                  * Set CORRESPONDING to the end of the changed run, at the last
1464                                  * point where it corresponds to a changed run in the other file.
1465                                  * CORRESPONDING == LEN means no such point has been found.
1466                                  */
1467                                 $corresponding = $j < $other_len ? $i : $len;
1468
1469                                 /*
1470                                  * Move the changed region forward, so long as the
1471                                  * first changed line matches the following unchanged one.
1472                                  * This merges with following changed regions.
1473                                  * Do this second, so that if there are no merges,
1474                                  * the changed region is moved forward as far as possible.
1475                                  */
1476                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1477                                         $changed[$start++] = false;
1478                                         $changed[$i++] = 1;
1479                                         while ($i < $len && $changed[$i])
1480                                         $i++;
1481
1482                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1483                                         $j++;
1484                                         if ($j < $other_len && $other_changed[$j]) {
1485                                                 $corresponding = $i;
1486                                                 while ($j < $other_len && $other_changed[$j])
1487                                                 $j++;
1488                                         }
1489                                 }
1490                         } while ($runlength != $i - $start);
1491
1492                         /*
1493                          * If possible, move the fully-merged run of changes
1494                          * back to a corresponding run in the other file.
1495                          */
1496                         while ($corresponding < $i) {
1497                                 $changed[--$start] = 1;
1498                                 $changed[--$i] = 0;
1499                                 USE_ASSERTS && assert('$j > 0');
1500                                 while ($other_changed[--$j])
1501                                 continue;
1502                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1503                         }
1504                 }
1505                 wfProfileOut( __METHOD__ );
1506         }
1507 }
1508
1509 /**
1510  * Class representing a 'diff' between two sequences of strings.
1511  * @todo document
1512  * @private
1513  * @ingroup DifferenceEngine
1514  */
1515 class Diff
1516 {
1517         var $edits;
1518
1519         /**
1520          * Constructor.
1521          * Computes diff between sequences of strings.
1522          *
1523          * @param $from_lines array An array of strings.
1524          *                (Typically these are lines from a file.)
1525          * @param $to_lines array An array of strings.
1526          */
1527         function Diff($from_lines, $to_lines) {
1528                 $eng = new _DiffEngine;
1529                 $this->edits = $eng->diff($from_lines, $to_lines);
1530                 //$this->_check($from_lines, $to_lines);
1531         }
1532
1533         /**
1534          * Compute reversed Diff.
1535          *
1536          * SYNOPSIS:
1537          *
1538          *      $diff = new Diff($lines1, $lines2);
1539          *      $rev = $diff->reverse();
1540          * @return object A Diff object representing the inverse of the
1541          *                                original diff.
1542          */
1543         function reverse () {
1544                 $rev = $this;
1545                 $rev->edits = array();
1546                 foreach ($this->edits as $edit) {
1547                         $rev->edits[] = $edit->reverse();
1548                 }
1549                 return $rev;
1550         }
1551
1552         /**
1553          * Check for empty diff.
1554          *
1555          * @return bool True iff two sequences were identical.
1556          */
1557         function isEmpty () {
1558                 foreach ($this->edits as $edit) {
1559                         if ($edit->type != 'copy')
1560                         return false;
1561                 }
1562                 return true;
1563         }
1564
1565         /**
1566          * Compute the length of the Longest Common Subsequence (LCS).
1567          *
1568          * This is mostly for diagnostic purposed.
1569          *
1570          * @return int The length of the LCS.
1571          */
1572         function lcs () {
1573                 $lcs = 0;
1574                 foreach ($this->edits as $edit) {
1575                         if ($edit->type == 'copy')
1576                         $lcs += sizeof($edit->orig);
1577                 }
1578                 return $lcs;
1579         }
1580
1581         /**
1582          * Get the original set of lines.
1583          *
1584          * This reconstructs the $from_lines parameter passed to the
1585          * constructor.
1586          *
1587          * @return array The original sequence of strings.
1588          */
1589         function orig() {
1590                 $lines = array();
1591
1592                 foreach ($this->edits as $edit) {
1593                         if ($edit->orig)
1594                         array_splice($lines, sizeof($lines), 0, $edit->orig);
1595                 }
1596                 return $lines;
1597         }
1598
1599         /**
1600          * Get the closing set of lines.
1601          *
1602          * This reconstructs the $to_lines parameter passed to the
1603          * constructor.
1604          *
1605          * @return array The sequence of strings.
1606          */
1607         function closing() {
1608                 $lines = array();
1609
1610                 foreach ($this->edits as $edit) {
1611                         if ($edit->closing)
1612                         array_splice($lines, sizeof($lines), 0, $edit->closing);
1613                 }
1614                 return $lines;
1615         }
1616
1617         /**
1618          * Check a Diff for validity.
1619          *
1620          * This is here only for debugging purposes.
1621          */
1622         function _check ($from_lines, $to_lines) {
1623                 wfProfileIn( __METHOD__ );
1624                 if (serialize($from_lines) != serialize($this->orig()))
1625                 trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1626                 if (serialize($to_lines) != serialize($this->closing()))
1627                 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1628
1629                 $rev = $this->reverse();
1630                 if (serialize($to_lines) != serialize($rev->orig()))
1631                 trigger_error("Reversed original doesn't match", E_USER_ERROR);
1632                 if (serialize($from_lines) != serialize($rev->closing()))
1633                 trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1634
1635
1636                 $prevtype = 'none';
1637                 foreach ($this->edits as $edit) {
1638                         if ( $prevtype == $edit->type )
1639                         trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1640                         $prevtype = $edit->type;
1641                 }
1642
1643                 $lcs = $this->lcs();
1644                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1645                 wfProfileOut( __METHOD__ );
1646         }
1647 }
1648
1649 /**
1650  * @todo document, bad name.
1651  * @private
1652  * @ingroup DifferenceEngine
1653  */
1654 class MappedDiff extends Diff
1655 {
1656         /**
1657          * Constructor.
1658          *
1659          * Computes diff between sequences of strings.
1660          *
1661          * This can be used to compute things like
1662          * case-insensitve diffs, or diffs which ignore
1663          * changes in white-space.
1664          *
1665          * @param $from_lines array An array of strings.
1666          *      (Typically these are lines from a file.)
1667          *
1668          * @param $to_lines array An array of strings.
1669          *
1670          * @param $mapped_from_lines array This array should
1671          *      have the same size number of elements as $from_lines.
1672          *      The elements in $mapped_from_lines and
1673          *      $mapped_to_lines are what is actually compared
1674          *      when computing the diff.
1675          *
1676          * @param $mapped_to_lines array This array should
1677          *      have the same number of elements as $to_lines.
1678          */
1679         function MappedDiff($from_lines, $to_lines,
1680         $mapped_from_lines, $mapped_to_lines) {
1681                 wfProfileIn( __METHOD__ );
1682
1683                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1684                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1685
1686                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1687
1688                 $xi = $yi = 0;
1689                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1690                         $orig = &$this->edits[$i]->orig;
1691                         if (is_array($orig)) {
1692                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1693                                 $xi += sizeof($orig);
1694                         }
1695
1696                         $closing = &$this->edits[$i]->closing;
1697                         if (is_array($closing)) {
1698                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1699                                 $yi += sizeof($closing);
1700                         }
1701                 }
1702                 wfProfileOut( __METHOD__ );
1703         }
1704 }
1705
1706 /**
1707  * A class to format Diffs
1708  *
1709  * This class formats the diff in classic diff format.
1710  * It is intended that this class be customized via inheritance,
1711  * to obtain fancier outputs.
1712  * @todo document
1713  * @private
1714  * @ingroup DifferenceEngine
1715  */
1716 class DiffFormatter {
1717         /**
1718          * Number of leading context "lines" to preserve.
1719          *
1720          * This should be left at zero for this class, but subclasses
1721          * may want to set this to other values.
1722          */
1723         var $leading_context_lines = 0;
1724
1725         /**
1726          * Number of trailing context "lines" to preserve.
1727          *
1728          * This should be left at zero for this class, but subclasses
1729          * may want to set this to other values.
1730          */
1731         var $trailing_context_lines = 0;
1732
1733         /**
1734          * Format a diff.
1735          *
1736          * @param $diff object A Diff object.
1737          * @return string The formatted output.
1738          */
1739         function format($diff) {
1740                 wfProfileIn( __METHOD__ );
1741
1742                 $xi = $yi = 1;
1743                 $block = false;
1744                 $context = array();
1745
1746                 $nlead = $this->leading_context_lines;
1747                 $ntrail = $this->trailing_context_lines;
1748
1749                 $this->_start_diff();
1750
1751                 foreach ($diff->edits as $edit) {
1752                         if ($edit->type == 'copy') {
1753                                 if (is_array($block)) {
1754                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1755                                                 $block[] = $edit;
1756                                         }
1757                                         else{
1758                                                 if ($ntrail) {
1759                                                         $context = array_slice($edit->orig, 0, $ntrail);
1760                                                         $block[] = new _DiffOp_Copy($context);
1761                                                 }
1762                                                 $this->_block($x0, $ntrail + $xi - $x0,
1763                                                 $y0, $ntrail + $yi - $y0,
1764                                                 $block);
1765                                                 $block = false;
1766                                         }
1767                                 }
1768                                 $context = $edit->orig;
1769                         }
1770                         else {
1771                                 if (! is_array($block)) {
1772                                         $context = array_slice($context, sizeof($context) - $nlead);
1773                                         $x0 = $xi - sizeof($context);
1774                                         $y0 = $yi - sizeof($context);
1775                                         $block = array();
1776                                         if ($context)
1777                                         $block[] = new _DiffOp_Copy($context);
1778                                 }
1779                                 $block[] = $edit;
1780                         }
1781
1782                         if ($edit->orig)
1783                         $xi += sizeof($edit->orig);
1784                         if ($edit->closing)
1785                         $yi += sizeof($edit->closing);
1786                 }
1787
1788                 if (is_array($block))
1789                 $this->_block($x0, $xi - $x0,
1790                 $y0, $yi - $y0,
1791                 $block);
1792
1793                 $end = $this->_end_diff();
1794                 wfProfileOut( __METHOD__ );
1795                 return $end;
1796         }
1797
1798         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1799                 wfProfileIn( __METHOD__ );
1800                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1801                 foreach ($edits as $edit) {
1802                         if ($edit->type == 'copy')
1803                         $this->_context($edit->orig);
1804                         elseif ($edit->type == 'add')
1805                         $this->_added($edit->closing);
1806                         elseif ($edit->type == 'delete')
1807                         $this->_deleted($edit->orig);
1808                         elseif ($edit->type == 'change')
1809                         $this->_changed($edit->orig, $edit->closing);
1810                         else
1811                         trigger_error('Unknown edit type', E_USER_ERROR);
1812                 }
1813                 $this->_end_block();
1814                 wfProfileOut( __METHOD__ );
1815         }
1816
1817         function _start_diff() {
1818                 ob_start();
1819         }
1820
1821         function _end_diff() {
1822                 $val = ob_get_contents();
1823                 ob_end_clean();
1824                 return $val;
1825         }
1826
1827         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1828                 if ($xlen > 1)
1829                 $xbeg .= "," . ($xbeg + $xlen - 1);
1830                 if ($ylen > 1)
1831                 $ybeg .= "," . ($ybeg + $ylen - 1);
1832
1833                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1834         }
1835
1836         function _start_block($header) {
1837                 echo $header . "\n";
1838         }
1839
1840         function _end_block() {
1841         }
1842
1843         function _lines($lines, $prefix = ' ') {
1844                 foreach ($lines as $line)
1845                 echo "$prefix $line\n";
1846         }
1847
1848         function _context($lines) {
1849                 $this->_lines($lines);
1850         }
1851
1852         function _added($lines) {
1853                 $this->_lines($lines, '>');
1854         }
1855         function _deleted($lines) {
1856                 $this->_lines($lines, '<');
1857         }
1858
1859         function _changed($orig, $closing) {
1860                 $this->_deleted($orig);
1861                 echo "---\n";
1862                 $this->_added($closing);
1863         }
1864 }
1865
1866 /**
1867  * A formatter that outputs unified diffs
1868  * @ingroup DifferenceEngine
1869  */
1870
1871 class UnifiedDiffFormatter extends DiffFormatter {
1872         var $leading_context_lines = 2;
1873         var $trailing_context_lines = 2;
1874
1875         function _added($lines) {
1876                 $this->_lines($lines, '+');
1877         }
1878         function _deleted($lines) {
1879                 $this->_lines($lines, '-');
1880         }
1881         function _changed($orig, $closing) {
1882                 $this->_deleted($orig);
1883                 $this->_added($closing);
1884         }
1885         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1886                 return "@@ -$xbeg,$xlen +$ybeg,$ylen @@";
1887         }
1888 }
1889
1890 /**
1891  * A pseudo-formatter that just passes along the Diff::$edits array
1892  * @ingroup DifferenceEngine
1893  */
1894 class ArrayDiffFormatter extends DiffFormatter {
1895         function format($diff) {
1896                 $oldline = 1;
1897                 $newline = 1;
1898                 $retval = array();
1899                 foreach($diff->edits as $edit)
1900                 switch($edit->type) {
1901                         case 'add':
1902                                 foreach($edit->closing as $l) {
1903                                         $retval[] = array(
1904                                                         'action' => 'add',
1905                                                         'new'=> $l,
1906                                                         'newline' => $newline++
1907                                         );
1908                                 }
1909                                 break;
1910                         case 'delete':
1911                                 foreach($edit->orig as $l) {
1912                                         $retval[] = array(
1913                                                         'action' => 'delete',
1914                                                         'old' => $l,
1915                                                         'oldline' => $oldline++,
1916                                         );
1917                                 }
1918                                 break;
1919                         case 'change':
1920                                 foreach($edit->orig as $i => $l) {
1921                                         $retval[] = array(
1922                                                         'action' => 'change',
1923                                                         'old' => $l,
1924                                                         'new' => @$edit->closing[$i],
1925                                                         'oldline' => $oldline++,
1926                                                         'newline' => $newline++,
1927                                         );
1928                                 }
1929                                 break;
1930                         case 'copy':
1931                                 $oldline += count($edit->orig);
1932                                 $newline += count($edit->orig);
1933                 }
1934                 return $retval;
1935         }
1936 }
1937
1938 /**
1939  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1940  *
1941  */
1942
1943 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1944
1945 /**
1946  * @todo document
1947  * @private
1948  * @ingroup DifferenceEngine
1949  */
1950 class _HWLDF_WordAccumulator {
1951         function _HWLDF_WordAccumulator () {
1952                 $this->_lines = array();
1953                 $this->_line = '';
1954                 $this->_group = '';
1955                 $this->_tag = '';
1956         }
1957
1958         function _flushGroup ($new_tag) {
1959                 if ($this->_group !== '') {
1960                         if ($this->_tag == 'ins')
1961                         $this->_line .= '<ins class="diffchange diffchange-inline">' .
1962                         htmlspecialchars ( $this->_group ) . '</ins>';
1963                         elseif ($this->_tag == 'del')
1964                         $this->_line .= '<del class="diffchange diffchange-inline">' .
1965                         htmlspecialchars ( $this->_group ) . '</del>';
1966                         else
1967                         $this->_line .= htmlspecialchars ( $this->_group );
1968                 }
1969                 $this->_group = '';
1970                 $this->_tag = $new_tag;
1971         }
1972
1973         function _flushLine ($new_tag) {
1974                 $this->_flushGroup($new_tag);
1975                 if ($this->_line != '')
1976                 array_push ( $this->_lines, $this->_line );
1977                 else
1978                 # make empty lines visible by inserting an NBSP
1979                 array_push ( $this->_lines, NBSP );
1980                 $this->_line = '';
1981         }
1982
1983         function addWords ($words, $tag = '') {
1984                 if ($tag != $this->_tag)
1985                 $this->_flushGroup($tag);
1986
1987                 foreach ($words as $word) {
1988                         // new-line should only come as first char of word.
1989                         if ($word == '')
1990                         continue;
1991                         if ($word[0] == "\n") {
1992                                 $this->_flushLine($tag);
1993                                 $word = substr($word, 1);
1994                         }
1995                         assert(!strstr($word, "\n"));
1996                         $this->_group .= $word;
1997                 }
1998         }
1999
2000         function getLines() {
2001                 $this->_flushLine('~done');
2002                 return $this->_lines;
2003         }
2004 }
2005
2006 /**
2007  * @todo document
2008  * @private
2009  * @ingroup DifferenceEngine
2010  */
2011 class WordLevelDiff extends MappedDiff {
2012         const MAX_LINE_LENGTH = 10000;
2013
2014         function WordLevelDiff ($orig_lines, $closing_lines) {
2015                 wfProfileIn( __METHOD__ );
2016
2017                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
2018                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
2019
2020                 $this->MappedDiff($orig_words, $closing_words,
2021                 $orig_stripped, $closing_stripped);
2022                 wfProfileOut( __METHOD__ );
2023         }
2024
2025         function _split($lines) {
2026                 wfProfileIn( __METHOD__ );
2027
2028                 $words = array();
2029                 $stripped = array();
2030                 $first = true;
2031                 foreach ( $lines as $line ) {
2032                         # If the line is too long, just pretend the entire line is one big word
2033                         # This prevents resource exhaustion problems
2034                         if ( $first ) {
2035                                 $first = false;
2036                         } else {
2037                                 $words[] = "\n";
2038                                 $stripped[] = "\n";
2039                         }
2040                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
2041                                 $words[] = $line;
2042                                 $stripped[] = $line;
2043                         } else {
2044                                 $m = array();
2045                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
2046                                 $line, $m))
2047                                 {
2048                                         $words = array_merge( $words, $m[0] );
2049                                         $stripped = array_merge( $stripped, $m[1] );
2050                                 }
2051                         }
2052                 }
2053                 wfProfileOut( __METHOD__ );
2054                 return array($words, $stripped);
2055         }
2056
2057         function orig () {
2058                 wfProfileIn( __METHOD__ );
2059                 $orig = new _HWLDF_WordAccumulator;
2060
2061                 foreach ($this->edits as $edit) {
2062                         if ($edit->type == 'copy')
2063                         $orig->addWords($edit->orig);
2064                         elseif ($edit->orig)
2065                         $orig->addWords($edit->orig, 'del');
2066                 }
2067                 $lines = $orig->getLines();
2068                 wfProfileOut( __METHOD__ );
2069                 return $lines;
2070         }
2071
2072         function closing () {
2073                 wfProfileIn( __METHOD__ );
2074                 $closing = new _HWLDF_WordAccumulator;
2075
2076                 foreach ($this->edits as $edit) {
2077                         if ($edit->type == 'copy')
2078                         $closing->addWords($edit->closing);
2079                         elseif ($edit->closing)
2080                         $closing->addWords($edit->closing, 'ins');
2081                 }
2082                 $lines = $closing->getLines();
2083                 wfProfileOut( __METHOD__ );
2084                 return $lines;
2085         }
2086 }
2087
2088 /**
2089  * Wikipedia Table style diff formatter.
2090  * @todo document
2091  * @private
2092  * @ingroup DifferenceEngine
2093  */
2094 class TableDiffFormatter extends DiffFormatter {
2095         function TableDiffFormatter() {
2096                 $this->leading_context_lines = 2;
2097                 $this->trailing_context_lines = 2;
2098         }
2099
2100         public static function escapeWhiteSpace( $msg ) {
2101                 $msg = preg_replace( '/^ /m', '&nbsp; ', $msg );
2102                 $msg = preg_replace( '/ $/m', ' &nbsp;', $msg );
2103                 $msg = preg_replace( '/  /', '&nbsp; ', $msg );
2104                 return $msg;
2105         }
2106
2107         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
2108                 $r = '<tr><td colspan="2" class="diff-lineno"><!--LINE '.$xbeg."--></td>\n" .
2109                   '<td colspan="2" class="diff-lineno"><!--LINE '.$ybeg."--></td></tr>\n";
2110                 return $r;
2111         }
2112
2113         function _start_block( $header ) {
2114                 echo $header;
2115         }
2116
2117         function _end_block() {
2118         }
2119
2120         function _lines( $lines, $prefix=' ', $color='white' ) {
2121         }
2122
2123         # HTML-escape parameter before calling this
2124         function addedLine( $line ) {
2125                 return $this->wrapLine( '+', 'diff-addedline', $line );
2126         }
2127
2128         # HTML-escape parameter before calling this
2129         function deletedLine( $line ) {
2130                 return $this->wrapLine( '-', 'diff-deletedline', $line );
2131         }
2132
2133         # HTML-escape parameter before calling this
2134         function contextLine( $line ) {
2135                 return $this->wrapLine( ' ', 'diff-context', $line );
2136         }
2137
2138         private function wrapLine( $marker, $class, $line ) {
2139                 if( $line !== '' ) {
2140                         // The <div> wrapper is needed for 'overflow: auto' style to scroll properly
2141                         $line = Xml::tags( 'div', null, $this->escapeWhiteSpace( $line ) );
2142                 }
2143                 return "<td class='diff-marker'>$marker</td><td class='$class'>$line</td>";
2144         }
2145
2146         function emptyLine() {
2147                 return '<td colspan="2">&nbsp;</td>';
2148         }
2149
2150         function _added( $lines ) {
2151                 foreach ($lines as $line) {
2152                         echo '<tr>' . $this->emptyLine() .
2153                         $this->addedLine( '<ins class="diffchange">' .
2154                         htmlspecialchars ( $line ) . '</ins>' ) . "</tr>\n";
2155                 }
2156         }
2157
2158         function _deleted($lines) {
2159                 foreach ($lines as $line) {
2160                         echo '<tr>' . $this->deletedLine( '<del class="diffchange">' .
2161                         htmlspecialchars ( $line ) . '</del>' ) .
2162                         $this->emptyLine() . "</tr>\n";
2163                 }
2164         }
2165
2166         function _context( $lines ) {
2167                 foreach ($lines as $line) {
2168                         echo '<tr>' .
2169                         $this->contextLine( htmlspecialchars ( $line ) ) .
2170                         $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
2171                 }
2172         }
2173
2174         function _changed( $orig, $closing ) {
2175                 wfProfileIn( __METHOD__ );
2176
2177                 $diff = new WordLevelDiff( $orig, $closing );
2178                 $del = $diff->orig();
2179                 $add = $diff->closing();
2180
2181                 # Notice that WordLevelDiff returns HTML-escaped output.
2182                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
2183
2184                 while ( $line = array_shift( $del ) ) {
2185                         $aline = array_shift( $add );
2186                         echo '<tr>' . $this->deletedLine( $line ) .
2187                         $this->addedLine( $aline ) . "</tr>\n";
2188                 }
2189                 foreach ($add as $line) {       # If any leftovers
2190                         echo '<tr>' . $this->emptyLine() .
2191                         $this->addedLine( $line ) . "</tr>\n";
2192                 }
2193                 wfProfileOut( __METHOD__ );
2194         }
2195 }