includes/DifferenceEngine.php

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