includes/DifferenceEngine.php

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