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