includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @addtogroup DifferenceEngine
   5  */
   6
   7 /**
   8  * @todo document
   9  * @public
  10  * @addtogroup DifferenceEngine
  11  */
  12 class DifferenceEngine {
  13         /**#@+
  14          * @private
  15          */
  16         var $mOldid, $mNewid, $mTitle;
  17         var $mOldtitle, $mNewtitle, $mPagetitle;
  18         var $mOldtext, $mNewtext;
  19         var $mOldPage, $mNewPage;
  20         var $mRcidMarkPatrolled;
  21         var $mOldRev, $mNewRev;
  22         var $mRevisionsLoaded = false; // Have the revisions been loaded
  23         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  24         /**#@-*/
  25
  26         /**
  27          * Constructor
  28          * @param $titleObj Title object that the diff is associated with
  29          * @param $old Integer: old ID we want to show and diff with.
  30          * @param $new String: either 'prev' or 'next'.
  31          * @param $rcid Integer: ??? FIXME (default 0)
  32          */
  33         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0 ) {
  34                 $this->mTitle = $titleObj;
  35                 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
  36
  37                 if ( 'prev' === $new ) {
  38                         # Show diff between revision $old and the previous one.
  39                         # Get previous one from DB.
  40                         #
  41                         $this->mNewid = intval($old);
  42
  43                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  44
  45                 } elseif ( 'next' === $new ) {
  46                         # Show diff between revision $old and the previous one.
  47                         # Get previous one from DB.
  48                         #
  49                         $this->mOldid = intval($old);
  50                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  51                         if ( false === $this->mNewid ) {
  52                                 # if no result, NewId points to the newest old revision. The only newer
  53                                 # revision is cur, which is "0".
  54                                 $this->mNewid = 0;
  55                         }
  56
  57                 } else {
  58                         $this->mOldid = intval($old);
  59                         $this->mNewid = intval($new);
  60                 }
  61                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  62         }
  63
  64         function showDiffPage( $diffOnly = false ) {
  65                 global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
  66                 $fname = 'DifferenceEngine::showDiffPage';
  67                 wfProfileIn( $fname );
  68
  69                 # If external diffs are enabled both globally and for the user,
  70                 # we'll use the application/x-external-editor interface to call
  71                 # an external diff tool like kompare, kdiff3, etc.
  72                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  73                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  74                         $wgOut->disable();
  75                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  76                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  77                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  78                         $special=$wgLang->getNsText(NS_SPECIAL);
  79                         $control=<<<CONTROL
  80 [Process]
  81 Type=Diff text
  82 Engine=MediaWiki
  83 Script={$wgServer}{$wgScript}
  84 Special namespace={$special}
  85
  86 [File]
  87 Extension=wiki
  88 URL=$url1
  89
  90 [File 2]
  91 Extension=wiki
  92 URL=$url2
  93 CONTROL;
  94                         echo($control);
  95                         return;
  96                 }
  97
  98                 $wgOut->setArticleFlag( false );
  99                 if ( ! $this->loadRevisionData() ) {
 100                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, {$this->mNewid})";
 101                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 102                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 103                         $wgOut->addWikitext( $mtext );
 104                         wfProfileOut( $fname );
 105                         return;
 106                 }
 107
 108                 wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
 109
 110                 if ( $this->mNewRev->isCurrent() ) {
 111                         $wgOut->setArticleFlag( true );
 112                 }
 113
 114                 # mOldid is false if the difference engine is called with a "vague" query for
 115                 # a diff between a version V and its previous version V' AND the version V
 116                 # is the first version of that article. In that case, V' does not exist.
 117                 if ( $this->mOldid === false ) {
 118                         $this->showFirstRevision();
 119                         $this->renderNewRevision();  // should we respect $diffOnly here or not?
 120                         wfProfileOut( $fname );
 121                         return;
 122                 }
 123
 124                 $wgOut->suppressQuickbar();
 125
 126                 $oldTitle = $this->mOldPage->getPrefixedText();
 127                 $newTitle = $this->mNewPage->getPrefixedText();
 128                 if( $oldTitle == $newTitle ) {
 129                         $wgOut->setPageTitle( $newTitle );
 130                 } else {
 131                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 132                 }
 133                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 134                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 135
 136                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 137                         $wgOut->loginToUse();
 138                         $wgOut->output();
 139                         wfProfileOut( $fname );
 140                         exit;
 141                 }
 142
 143                 $sk = $wgUser->getSkin();
 144
 145                 if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 146                         $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
 147                 } else {
 148                         $rollback = '';
 149                 }
 150                 if( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isAllowed( 'patrol' ) ) {
 151                         $patrol = ' [' . $sk->makeKnownLinkObj( $this->mTitle, wfMsg( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 152                 } else {
 153                         $patrol = '';
 154                 }
 155
 156                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 157                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 158                 if ( $this->mNewRev->isCurrent() ) {
 159                         $nextlink = '&nbsp;';
 160                 } else {
 161                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 162                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 163                 }
 164
 165                 $oldminor = '';
 166                 $newminor = '';
 167
 168                 if ($this->mOldRev->mMinorEdit == 1) {
 169                         $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
 170                                 wfMsg( 'minoreditletter') ) . ' ';
 171                 }
 172
 173                 if ($this->mNewRev->mMinorEdit == 1) {
 174                         $newminor = wfElement( 'span', array( 'class' => 'minor' ),
 175                         wfMsg( 'minoreditletter') ) . ' ';
 176                 }
 177
 178                 $rdel = ''; $ldel = '';
 179                 if( $wgUser->isAllowed( 'deleterevision' ) ) {
 180                         $revdel = SpecialPage::getTitleFor( 'Revisiondelete' );
 181                         if( !$this->mOldRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 182                         // If revision was hidden from sysops
 183                                 $ldel = wfMsgHtml('rev-delundel');
 184                         } else {
 185                                 $ldel = $sk->makeKnownLinkObj( $revdel,
 186                                         wfMsgHtml('rev-delundel'),
 187                                         'target=' . urlencode( $this->mOldRev->mTitle->getPrefixedDbkey() ) .
 188                                         '&oldid=' . urlencode( $this->mOldRev->getId() ) );
 189                                 // Bolden oversighted content
 190                                 if( $this->mOldRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 191                                         $ldel = "<strong>$ldel</strong>";
 192                         }
 193                         $ldel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$ldel</small>)</tt> ";
 194                         // We don't currently handle well changing the top revision's settings
 195                         if( $this->mNewRev->isCurrent() ) {
 196                         // If revision was hidden from sysops
 197                                 $rdel = wfMsgHtml('rev-delundel');
 198                         } else if( !$this->mNewRev->userCan( Revision::DELETED_RESTRICTED ) ) {
 199                         // If revision was hidden from sysops
 200                                 $rdel = wfMsgHtml('rev-delundel');
 201                         } else {
 202                                 $rdel = $sk->makeKnownLinkObj( $revdel,
 203                                         wfMsgHtml('rev-delundel'),
 204                                         'target=' . urlencode( $this->mNewRev->mTitle->getPrefixedDbkey() ) .
 205                                         '&oldid=' . urlencode( $this->mNewRev->getId() ) );
 206                                 // Bolden oversighted content
 207                                 if( $this->mNewRev->isDeleted( Revision::DELETED_RESTRICTED ) )
 208                                         $rdel = "<strong>$rdel</strong>";
 209                         }
 210                         $rdel = "&nbsp;&nbsp;&nbsp;<tt>(<small>$rdel</small>)</tt> ";
 211                 }
 212
 213                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />" .
 214                         $sk->revUserTools( $this->mOldRev, true ) . "<br />" .
 215                         $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly, true ) . "$ldel<br />" .
 216                         $prevlink;
 217                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />" .
 218                         $sk->revUserTools( $this->mNewRev, true ) . " $rollback<br />" .
 219                         $newminor . $sk->revComment( $this->mNewRev, !$diffOnly, true ) . "$rdel<br />" .
 220                         $nextlink . $patrol;
 221
 222                 $this->showDiff( $oldHeader, $newHeader );
 223
 224                 if ( !$diffOnly )
 225                         $this->renderNewRevision();
 226
 227                 wfProfileOut( $fname );
 228         }
 229
 230         /**
 231          * Show the new revision of the page.
 232          */
 233         function renderNewRevision() {
 234                 global $wgOut;
 235                 $fname = 'DifferenceEngine::renderNewRevision';
 236                 wfProfileIn( $fname );
 237
 238                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 239                 #add deleted rev tag if needed
 240                 if( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 241                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-permission' ) );
 242                 } else if( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 243                         $wgOut->addWikiText( wfMsg( 'rev-deleted-text-view' ) );
 244                 }
 245
 246                 if( !$this->mNewRev->isCurrent() ) {
 247                         $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
 248                 }
 249
 250                 $this->loadNewText();
 251                 if( is_object( $this->mNewRev ) ) {
 252                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 253                 }
 254
 255                 $wgOut->addWikiTextTidy( $this->mNewtext );
 256
 257                 if( !$this->mNewRev->isCurrent() ) {
 258                         $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
 259                 }
 260
 261                 wfProfileOut( $fname );
 262         }
 263
 264         /**
 265          * Show the first revision of an article. Uses normal diff headers in
 266          * contrast to normal "old revision" display style.
 267          */
 268         function showFirstRevision() {
 269                 global $wgOut, $wgUser;
 270
 271                 $fname = 'DifferenceEngine::showFirstRevision';
 272                 wfProfileIn( $fname );
 273
 274                 # Get article text from the DB
 275                 #
 276                 if ( ! $this->loadNewText() ) {
 277                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 278                           "{$this->mNewid})";
 279                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 280                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 281                         $wgOut->addWikitext( $mtext );
 282                         wfProfileOut( $fname );
 283                         return;
 284                 }
 285                 if ( $this->mNewRev->isCurrent() ) {
 286                         $wgOut->setArticleFlag( true );
 287                 }
 288
 289                 # Check if user is allowed to look at this page. If not, bail out.
 290                 #
 291                 if ( !( $this->mTitle->userCanRead() ) ) {
 292                         $wgOut->loginToUse();
 293                         $wgOut->output();
 294                         wfProfileOut( $fname );
 295                         exit;
 296                 }
 297
 298                 # Prepare the header box
 299                 #
 300                 $sk = $wgUser->getSkin();
 301
 302                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 303                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
 304                         $sk->revUserTools( $this->mNewRev ) . "<br />" .
 305                         $sk->revComment( $this->mNewRev ) . "<br />" .
 306                         $nextlink . "</div>\n";
 307
 308                 $wgOut->addHTML( $header );
 309
 310                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 311                 $wgOut->setRobotpolicy( 'noindex,nofollow' );
 312
 313                 wfProfileOut( $fname );
 314         }
 315
 316         /**
 317          * Get the diff text, send it to $wgOut
 318          * Returns false if the diff could not be generated, otherwise returns true
 319          */
 320         function showDiff( $otitle, $ntitle ) {
 321                 global $wgOut;
 322                 $diff = $this->getDiff( $otitle, $ntitle );
 323                 if ( $diff === false ) {
 324                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
 325                         return false;
 326                 } else {
 327                         $wgOut->addHTML( $diff );
 328                         return true;
 329                 }
 330         }
 331
 332         /**
 333          * Get diff table, including header
 334          * Note that the interface has changed, it's no longer static.
 335          * Returns false on error
 336          */
 337         function getDiff( $otitle, $ntitle ) {
 338                 $body = $this->getDiffBody();
 339                 if ( $body === false ) {
 340                         return false;
 341                 } else {
 342                         $multi = $this->getMultiNotice();
 343                         return $this->addHeader( $body, $otitle, $ntitle, $multi );
 344                 }
 345         }
 346
 347         /**
 348          * Get the diff table body, without header
 349          * Results are cached
 350          * Returns false on error
 351          */
 352         function getDiffBody() {
 353                 global $wgMemc;
 354                 $fname = 'DifferenceEngine::getDiffBody';
 355                 wfProfileIn( $fname );
 356
 357                 // Cacheable?
 358                 $key = false;
 359                 if ( $this->mOldid && $this->mNewid ) {
 360                         // Try cache
 361                         $key = wfMemcKey( 'diff', 'oldid', $this->mOldid, 'newid', $this->mNewid );
 362                         $difftext = $wgMemc->get( $key );
 363                         if ( $difftext ) {
 364                                 wfIncrStats( 'diff_cache_hit' );
 365                                 $difftext = $this->localiseLineNumbers( $difftext );
 366                                 $difftext .= "\n<!-- diff cache key $key -->\n";
 367                                 wfProfileOut( $fname );
 368                                 return $difftext;
 369                         }
 370                 }
 371
 372                 // Loadtext is permission safe, this just clears out the diff
 373                 if ( !$this->loadText() ) {
 374                         wfProfileOut( $fname );
 375                         return false;
 376                 } else if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 377                         return '';
 378                 } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 379                         return '';
 380                 }
 381
 382                 $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
 383
 384                 // Save to cache for 7 days
 385                 // Only do this for public revs, otherwise an admin can view the diff and a non-admin can nab it!
 386                 if ( $this->mOldRev && $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 387                         wfIncrStats( 'diff_uncacheable' );
 388                 } else if ( $this->mNewRev && $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 389                         wfIncrStats( 'diff_uncacheable' );
 390                 } else if ( $key !== false && $difftext !== false ) {
 391                         wfIncrStats( 'diff_cache_miss' );
 392                         $wgMemc->set( $key, $difftext, 7*86400 );
 393                 } else {
 394                         wfIncrStats( 'diff_uncacheable' );
 395                 }
 396                 // Replace line numbers with the text in the user's language
 397                 if ( $difftext !== false ) {
 398                         $difftext = $this->localiseLineNumbers( $difftext );
 399                 }
 400                 wfProfileOut( $fname );
 401                 return $difftext;
 402         }
 403
 404         /**
 405          * Generate a diff, no caching
 406          * $otext and $ntext must be already segmented
 407          */
 408         function generateDiffBody( $otext, $ntext ) {
 409                 global $wgExternalDiffEngine, $wgContLang;
 410                 $fname = 'DifferenceEngine::generateDiffBody';
 411
 412                 $otext = str_replace( "\r\n", "\n", $otext );
 413                 $ntext = str_replace( "\r\n", "\n", $ntext );
 414
 415                 if ( $wgExternalDiffEngine == 'wikidiff' ) {
 416                         # For historical reasons, external diff engine expects
 417                         # input text to be HTML-escaped already
 418                         $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
 419                         $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
 420                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 421                                 dl('php_wikidiff.so');
 422                         }
 423                         return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
 424                 }
 425
 426                 if ( $wgExternalDiffEngine == 'wikidiff2' ) {
 427                         # Better external diff engine, the 2 may some day be dropped
 428                         # This one does the escaping and segmenting itself
 429                         if ( !function_exists( 'wikidiff2_do_diff' ) ) {
 430                                 wfProfileIn( "$fname-dl" );
 431                                 @dl('php_wikidiff2.so');
 432                                 wfProfileOut( "$fname-dl" );
 433                         }
 434                         if ( function_exists( 'wikidiff2_do_diff' ) ) {
 435                                 wfProfileIn( 'wikidiff2_do_diff' );
 436                                 $text = wikidiff2_do_diff( $otext, $ntext, 2 );
 437                                 wfProfileOut( 'wikidiff2_do_diff' );
 438                                 return $text;
 439                         }
 440                 }
 441                 if ( $wgExternalDiffEngine !== false ) {
 442                         # Diff via the shell
 443                         global $wgTmpDirectory;
 444                         $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
 445                         $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
 446
 447                         $tempFile1 = fopen( $tempName1, "w" );
 448                         if ( !$tempFile1 ) {
 449                                 wfProfileOut( $fname );
 450                                 return false;
 451                         }
 452                         $tempFile2 = fopen( $tempName2, "w" );
 453                         if ( !$tempFile2 ) {
 454                                 wfProfileOut( $fname );
 455                                 return false;
 456                         }
 457                         fwrite( $tempFile1, $otext );
 458                         fwrite( $tempFile2, $ntext );
 459                         fclose( $tempFile1 );
 460                         fclose( $tempFile2 );
 461                         $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
 462                         wfProfileIn( "$fname-shellexec" );
 463                         $difftext = wfShellExec( $cmd );
 464                         wfProfileOut( "$fname-shellexec" );
 465                         unlink( $tempName1 );
 466                         unlink( $tempName2 );
 467                         return $difftext;
 468                 }
 469
 470                 # Native PHP diff
 471                 $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
 472                 $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
 473                 $diffs = new Diff( $ota, $nta );
 474                 $formatter = new TableDiffFormatter();
 475                 return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
 476         }
 477
 478
 479         /**
 480          * Replace line numbers with the text in the user's language
 481          */
 482         function localiseLineNumbers( $text ) {
 483                 return preg_replace_callback( '/<!--LINE (\d+)-->/',
 484                         array( &$this, 'localiseLineNumbersCb' ), $text );
 485         }
 486
 487         function localiseLineNumbersCb( $matches ) {
 488                 global $wgLang;
 489                 return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
 490         }
 491
 492
 493         /**
 494          * If there are revisions between the ones being compared, return a note saying so.
 495          */
 496         function getMultiNotice() {
 497                 if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
 498                         return '';
 499
 500                 if( !$this->mOldPage->equals( $this->mNewPage ) ) {
 501                         // Comparing two different pages? Count would be meaningless.
 502                         return '';
 503                 }
 504
 505                 $oldid = $this->mOldRev->getId();
 506                 $newid = $this->mNewRev->getId();
 507                 if ( $oldid > $newid ) {
 508                         $tmp = $oldid; $oldid = $newid; $newid = $tmp;
 509                 }
 510
 511                 $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
 512                 if ( !$n )
 513                         return '';
 514
 515                 return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
 516         }
 517
 518
 519         /**
 520          * Add the header to a diff body
 521          */
 522         function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
 523                 global $wgOut;
 524
 525                 $header = "
 526                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 527                         <tr>
 528                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 529                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 530                         </tr>
 531                 ";
 532
 533                 if ( $multi != '' )
 534                         $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
 535
 536                 return $header . $diff . "</table>";
 537         }
 538
 539         /**
 540          * Use specified text instead of loading from the database
 541          */
 542         function setText( $oldText, $newText ) {
 543                 $this->mOldtext = $oldText;
 544                 $this->mNewtext = $newText;
 545                 $this->mTextLoaded = 2;
 546         }
 547
 548         /**
 549          * Load revision metadata for the specified articles. If newid is 0, then compare
 550          * the old article in oldid to the current article; if oldid is 0, then
 551          * compare the current article to the immediately previous one (ignoring the
 552          * value of newid).
 553          *
 554          * If oldid is false, leave the corresponding revision object set
 555          * to false. This is impossible via ordinary user input, and is provided for
 556          * API convenience.
 557          */
 558         function loadRevisionData() {
 559                 global $wgLang;
 560                 if ( $this->mRevisionsLoaded ) {
 561                         return true;
 562                 } else {
 563                         // Whether it succeeds or fails, we don't want to try again
 564                         $this->mRevisionsLoaded = true;
 565                 }
 566
 567                 // Load the new revision object
 568                 if( $this->mNewid ) {
 569                         $this->mNewRev = Revision::newFromId( $this->mNewid );
 570                 } else {
 571                         $this->mNewRev = Revision::newFromTitle( $this->mTitle );
 572                 }
 573
 574                 if( is_null( $this->mNewRev ) ) {
 575                         return false;
 576                 }
 577
 578                 // Set assorted variables
 579                 $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 580                 $this->mNewPage = $this->mNewRev->getTitle();
 581                 if( $this->mNewRev->isCurrent() ) {
 582                         $newLink = $this->mNewPage->escapeLocalUrl();
 583                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 584                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
 585
 586                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
 587                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 588
 589                 } else {
 590                         $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
 591                         $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
 592                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
 593
 594                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
 595                                 . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 596                 }
 597                 if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
 598                         $this->mNewtitle = "<span class='history-deleted'>{$this->mPagetitle}</span>";
 599                 } else if ( $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
 600                         $this->mNewtitle = '<span class="history-deleted">'.$this->mNewtitle.'</span>';
 601                 }
 602
 603                 // Load the old revision object
 604                 $this->mOldRev = false;
 605                 if( $this->mOldid ) {
 606                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 607                 } elseif ( $this->mOldid === 0 ) {
 608                         $rev = $this->mNewRev->getPrevious();
 609                         if( $rev ) {
 610                                 $this->mOldid = $rev->getId();
 611                                 $this->mOldRev = $rev;
 612                         } else {
 613                                 // No previous revision; mark to show as first-version only.
 614                                 $this->mOldid = false;
 615                                 $this->mOldRev = false;
 616                         }
 617                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 618
 619                 if( is_null( $this->mOldRev ) ) {
 620                         return false;
 621                 }
 622
 623                 if ( $this->mOldRev ) {
 624                         $this->mOldPage = $this->mOldRev->getTitle();
 625
 626                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 627                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 628                         $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
 629                         $this->mOldPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 630
 631                         $this->mOldtitle = "<a href='$oldLink'>{$this->mOldPagetitle}</a>"
 632                                 . "(<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
 633                         // Now that we considered old rev, we can make undo link (bug 8133, multi-edit undo)
 634                         $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undoto=' . $this->mNewid);
 635                         if ( $this->mNewRev->userCan(Revision::DELETED_TEXT) )
 636                                 $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
 637
 638                         if ( !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
 639                                 $this->mOldtitle = "<span class='history-deleted'>{$this->mOldPagetitle}</span>";
 640                         } else if ( $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
 641                                 $this->mOldtitle = '<span class="history-deleted">'.$this->mOldtitle.'</span>';
 642                         }
 643                 }
 644
 645                 return true;
 646         }
 647
 648         /**
 649          * Load the text of the revisions, as well as revision data.
 650          */
 651         function loadText() {
 652                 if ( $this->mTextLoaded == 2 ) {
 653                         return true;
 654                 } else {
 655                         // Whether it succeeds or fails, we don't want to try again
 656                         $this->mTextLoaded = 2;
 657                 }
 658
 659                 if ( !$this->loadRevisionData() ) {
 660                         return false;
 661                 }
 662                 if ( $this->mOldRev ) {
 663                         $this->mOldtext = $this->mOldRev->revText();
 664                         if ( $this->mOldtext === false ) {
 665                                 return false;
 666                         }
 667                 }
 668                 if ( $this->mNewRev ) {
 669                         $this->mNewtext = $this->mNewRev->revText();
 670                         if ( $this->mNewtext === false ) {
 671                                 return false;
 672                         }
 673                 }
 674                 return true;
 675         }
 676
 677         /**
 678          * Load the text of the new revision, not the old one
 679          */
 680         function loadNewText() {
 681                 if ( $this->mTextLoaded >= 1 ) {
 682                         return true;
 683                 } else {
 684                         $this->mTextLoaded = 1;
 685                 }
 686                 if ( !$this->loadRevisionData() ) {
 687                         return false;
 688                 }
 689                 $this->mNewtext = $this->mNewRev->getText();
 690                 return true;
 691         }
 692
 693
 694 }
 695
 696 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 697 //
 698 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 699 // You may copy this code freely under the conditions of the GPL.
 700 //
 701
 702 define('USE_ASSERTS', function_exists('assert'));
 703
 704 /**
 705  * @todo document
 706  * @private
 707  * @addtogroup DifferenceEngine
 708  */
 709 class _DiffOp {
 710         var $type;
 711         var $orig;
 712         var $closing;
 713
 714         function reverse() {
 715                 trigger_error('pure virtual', E_USER_ERROR);
 716         }
 717
 718         function norig() {
 719                 return $this->orig ? sizeof($this->orig) : 0;
 720         }
 721
 722         function nclosing() {
 723                 return $this->closing ? sizeof($this->closing) : 0;
 724         }
 725 }
 726
 727 /**
 728  * @todo document
 729  * @private
 730  * @addtogroup DifferenceEngine
 731  */
 732 class _DiffOp_Copy extends _DiffOp {
 733         var $type = 'copy';
 734
 735         function _DiffOp_Copy ($orig, $closing = false) {
 736                 if (!is_array($closing))
 737                         $closing = $orig;
 738                 $this->orig = $orig;
 739                 $this->closing = $closing;
 740         }
 741
 742         function reverse() {
 743                 return new _DiffOp_Copy($this->closing, $this->orig);
 744         }
 745 }
 746
 747 /**
 748  * @todo document
 749  * @private
 750  * @addtogroup DifferenceEngine
 751  */
 752 class _DiffOp_Delete extends _DiffOp {
 753         var $type = 'delete';
 754
 755         function _DiffOp_Delete ($lines) {
 756                 $this->orig = $lines;
 757                 $this->closing = false;
 758         }
 759
 760         function reverse() {
 761                 return new _DiffOp_Add($this->orig);
 762         }
 763 }
 764
 765 /**
 766  * @todo document
 767  * @private
 768  * @addtogroup DifferenceEngine
 769  */
 770 class _DiffOp_Add extends _DiffOp {
 771         var $type = 'add';
 772
 773         function _DiffOp_Add ($lines) {
 774                 $this->closing = $lines;
 775                 $this->orig = false;
 776         }
 777
 778         function reverse() {
 779                 return new _DiffOp_Delete($this->closing);
 780         }
 781 }
 782
 783 /**
 784  * @todo document
 785  * @private
 786  * @addtogroup DifferenceEngine
 787  */
 788 class _DiffOp_Change extends _DiffOp {
 789         var $type = 'change';
 790
 791         function _DiffOp_Change ($orig, $closing) {
 792                 $this->orig = $orig;
 793                 $this->closing = $closing;
 794         }
 795
 796         function reverse() {
 797                 return new _DiffOp_Change($this->closing, $this->orig);
 798         }
 799 }
 800
 801
 802 /**
 803  * Class used internally by Diff to actually compute the diffs.
 804  *
 805  * The algorithm used here is mostly lifted from the perl module
 806  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 807  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 808  *
 809  * More ideas are taken from:
 810  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 811  *
 812  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 813  * diffutils-2.7, which can be found at:
 814  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 815  *
 816  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 817  * are my own.
 818  *
 819  * Line length limits for robustness added by Tim Starling, 2005-08-31
 820  *
 821  * @author Geoffrey T. Dairiki, Tim Starling
 822  * @private
 823  * @addtogroup DifferenceEngine
 824  */
 825 class _DiffEngine
 826 {
 827         const MAX_XREF_LENGTH =  10000;
 828
 829         function diff ($from_lines, $to_lines) {
 830                 $fname = '_DiffEngine::diff';
 831                 wfProfileIn( $fname );
 832
 833                 $n_from = sizeof($from_lines);
 834                 $n_to = sizeof($to_lines);
 835
 836                 $this->xchanged = $this->ychanged = array();
 837                 $this->xv = $this->yv = array();
 838                 $this->xind = $this->yind = array();
 839                 unset($this->seq);
 840                 unset($this->in_seq);
 841                 unset($this->lcs);
 842
 843                 // Skip leading common lines.
 844                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 845                         if ($from_lines[$skip] !== $to_lines[$skip])
 846                                 break;
 847                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 848                 }
 849                 // Skip trailing common lines.
 850                 $xi = $n_from; $yi = $n_to;
 851                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 852                         if ($from_lines[$xi] !== $to_lines[$yi])
 853                                 break;
 854                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 855                 }
 856
 857                 // Ignore lines which do not exist in both files.
 858                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 859                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 860                 }
 861
 862                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 863                         $line = $to_lines[$yi];
 864                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 865                                 continue;
 866                         $yhash[$this->_line_hash($line)] = 1;
 867                         $this->yv[] = $line;
 868                         $this->yind[] = $yi;
 869                 }
 870                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 871                         $line = $from_lines[$xi];
 872                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 873                                 continue;
 874                         $this->xv[] = $line;
 875                         $this->xind[] = $xi;
 876                 }
 877
 878                 // Find the LCS.
 879                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 880
 881                 // Merge edits when possible
 882                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 883                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 884
 885                 // Compute the edit operations.
 886                 $edits = array();
 887                 $xi = $yi = 0;
 888                 while ($xi < $n_from || $yi < $n_to) {
 889                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 890                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 891
 892                         // Skip matching "snake".
 893                         $copy = array();
 894                         while ( $xi < $n_from && $yi < $n_to
 895                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 896                                 $copy[] = $from_lines[$xi++];
 897                                 ++$yi;
 898                         }
 899                         if ($copy)
 900                                 $edits[] = new _DiffOp_Copy($copy);
 901
 902                         // Find deletes & adds.
 903                         $delete = array();
 904                         while ($xi < $n_from && $this->xchanged[$xi])
 905                                 $delete[] = $from_lines[$xi++];
 906
 907                         $add = array();
 908                         while ($yi < $n_to && $this->ychanged[$yi])
 909                                 $add[] = $to_lines[$yi++];
 910
 911                         if ($delete && $add)
 912                                 $edits[] = new _DiffOp_Change($delete, $add);
 913                         elseif ($delete)
 914                                 $edits[] = new _DiffOp_Delete($delete);
 915                         elseif ($add)
 916                                 $edits[] = new _DiffOp_Add($add);
 917                 }
 918                 wfProfileOut( $fname );
 919                 return $edits;
 920         }
 921
 922         /**
 923          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 924          */
 925         function _line_hash( $line ) {
 926                 if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
 927                         return md5( $line );
 928                 } else {
 929                         return $line;
 930                 }
 931         }
 932
 933
 934         /* Divide the Largest Common Subsequence (LCS) of the sequences
 935          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 936          * sized segments.
 937          *
 938          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 939          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 940          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 941          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 942          * that (X0, Y0) == (XOFF, YOFF) and
 943          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 944          *
 945          * This function assumes that the first lines of the specified portions
 946          * of the two files do not match, and likewise that the last lines do not
 947          * match.  The caller must trim matching lines from the beginning and end
 948          * of the portions it is going to specify.
 949          */
 950         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 951                 $fname = '_DiffEngine::_diag';
 952                 wfProfileIn( $fname );
 953                 $flip = false;
 954
 955                 if ($xlim - $xoff > $ylim - $yoff) {
 956                         // Things seems faster (I'm not sure I understand why)
 957                                 // when the shortest sequence in X.
 958                                 $flip = true;
 959                         list ($xoff, $xlim, $yoff, $ylim)
 960                         = array( $yoff, $ylim, $xoff, $xlim);
 961                 }
 962
 963                 if ($flip)
 964                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 965                                 $ymatches[$this->xv[$i]][] = $i;
 966                 else
 967                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 968                                 $ymatches[$this->yv[$i]][] = $i;
 969
 970                 $this->lcs = 0;
 971                 $this->seq[0]= $yoff - 1;
 972                 $this->in_seq = array();
 973                 $ymids[0] = array();
 974
 975                 $numer = $xlim - $xoff + $nchunks - 1;
 976                 $x = $xoff;
 977                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 978                         wfProfileIn( "$fname-chunk" );
 979                         if ($chunk > 0)
 980                                 for ($i = 0; $i <= $this->lcs; $i++)
 981                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 982
 983                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 984                         for ( ; $x < $x1; $x++) {
 985                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 986                                         if (empty($ymatches[$line]))
 987                                                 continue;
 988                                 $matches = $ymatches[$line];
 989                                 reset($matches);
 990                                 while (list ($junk, $y) = each($matches))
 991                                         if (empty($this->in_seq[$y])) {
 992                                                 $k = $this->_lcs_pos($y);
 993                                                 USE_ASSERTS && assert($k > 0);
 994                                                 $ymids[$k] = $ymids[$k-1];
 995                                                 break;
 996                                         }
 997                                 while (list ( /* $junk */, $y) = each($matches)) {
 998                                         if ($y > $this->seq[$k-1]) {
 999                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
1000                                                 // Optimization: this is a common case:
1001                                                 //      next match is just replacing previous match.
1002                                                 $this->in_seq[$this->seq[$k]] = false;
1003                                                 $this->seq[$k] = $y;
1004                                                 $this->in_seq[$y] = 1;
1005                                         } else if (empty($this->in_seq[$y])) {
1006                                                 $k = $this->_lcs_pos($y);
1007                                                 USE_ASSERTS && assert($k > 0);
1008                                                 $ymids[$k] = $ymids[$k-1];
1009                                         }
1010                                 }
1011                         }
1012                         wfProfileOut( "$fname-chunk" );
1013                 }
1014
1015                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
1016                 $ymid = $ymids[$this->lcs];
1017                 for ($n = 0; $n < $nchunks - 1; $n++) {
1018                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
1019                         $y1 = $ymid[$n] + 1;
1020                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
1021                 }
1022                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
1023
1024                 wfProfileOut( $fname );
1025                 return array($this->lcs, $seps);
1026         }
1027
1028         function _lcs_pos ($ypos) {
1029                 $fname = '_DiffEngine::_lcs_pos';
1030                 wfProfileIn( $fname );
1031
1032                 $end = $this->lcs;
1033                 if ($end == 0 || $ypos > $this->seq[$end]) {
1034                         $this->seq[++$this->lcs] = $ypos;
1035                         $this->in_seq[$ypos] = 1;
1036                         wfProfileOut( $fname );
1037                         return $this->lcs;
1038                 }
1039
1040                 $beg = 1;
1041                 while ($beg < $end) {
1042                         $mid = (int)(($beg + $end) / 2);
1043                         if ( $ypos > $this->seq[$mid] )
1044                                 $beg = $mid + 1;
1045                         else
1046                                 $end = $mid;
1047                 }
1048
1049                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
1050
1051                 $this->in_seq[$this->seq[$end]] = false;
1052                 $this->seq[$end] = $ypos;
1053                 $this->in_seq[$ypos] = 1;
1054                 wfProfileOut( $fname );
1055                 return $end;
1056         }
1057
1058         /* Find LCS of two sequences.
1059          *
1060          * The results are recorded in the vectors $this->{x,y}changed[], by
1061          * storing a 1 in the element for each line that is an insertion
1062          * or deletion (ie. is not in the LCS).
1063          *
1064          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1065          *
1066          * Note that XLIM, YLIM are exclusive bounds.
1067          * All line numbers are origin-0 and discarded lines are not counted.
1068          */
1069         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1070                 $fname = '_DiffEngine::_compareseq';
1071                 wfProfileIn( $fname );
1072
1073                 // Slide down the bottom initial diagonal.
1074                 while ($xoff < $xlim && $yoff < $ylim
1075                            && $this->xv[$xoff] == $this->yv[$yoff]) {
1076                         ++$xoff;
1077                         ++$yoff;
1078                 }
1079
1080                 // Slide up the top initial diagonal.
1081                 while ($xlim > $xoff && $ylim > $yoff
1082                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1083                         --$xlim;
1084                         --$ylim;
1085                 }
1086
1087                 if ($xoff == $xlim || $yoff == $ylim)
1088                         $lcs = 0;
1089                 else {
1090                         // This is ad hoc but seems to work well.
1091                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1092                         //$nchunks = max(2,min(8,(int)$nchunks));
1093                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1094                         list ($lcs, $seps)
1095                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1096                 }
1097
1098                 if ($lcs == 0) {
1099                         // X and Y sequences have no common subsequence:
1100                         // mark all changed.
1101                         while ($yoff < $ylim)
1102                                 $this->ychanged[$this->yind[$yoff++]] = 1;
1103                         while ($xoff < $xlim)
1104                                 $this->xchanged[$this->xind[$xoff++]] = 1;
1105                 } else {
1106                         // Use the partitions to split this problem into subproblems.
1107                         reset($seps);
1108                         $pt1 = $seps[0];
1109                         while ($pt2 = next($seps)) {
1110                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1111                                 $pt1 = $pt2;
1112                         }
1113                 }
1114                 wfProfileOut( $fname );
1115         }
1116
1117         /* Adjust inserts/deletes of identical lines to join changes
1118          * as much as possible.
1119          *
1120          * We do something when a run of changed lines include a
1121          * line at one end and has an excluded, identical line at the other.
1122          * We are free to choose which identical line is included.
1123          * `compareseq' usually chooses the one at the beginning,
1124          * but usually it is cleaner to consider the following identical line
1125          * to be the "change".
1126          *
1127          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1128          */
1129         function _shift_boundaries ($lines, &$changed, $other_changed) {
1130                 $fname = '_DiffEngine::_shift_boundaries';
1131                 wfProfileIn( $fname );
1132                 $i = 0;
1133                 $j = 0;
1134
1135                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1136                 $len = sizeof($lines);
1137                 $other_len = sizeof($other_changed);
1138
1139                 while (1) {
1140                         /*
1141                          * Scan forwards to find beginning of another run of changes.
1142                          * Also keep track of the corresponding point in the other file.
1143                          *
1144                          * Throughout this code, $i and $j are adjusted together so that
1145                          * the first $i elements of $changed and the first $j elements
1146                          * of $other_changed both contain the same number of zeros
1147                          * (unchanged lines).
1148                          * Furthermore, $j is always kept so that $j == $other_len or
1149                          * $other_changed[$j] == false.
1150                          */
1151                         while ($j < $other_len && $other_changed[$j])
1152                                 $j++;
1153
1154                         while ($i < $len && ! $changed[$i]) {
1155                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1156                                 $i++; $j++;
1157                                 while ($j < $other_len && $other_changed[$j])
1158                                         $j++;
1159                         }
1160
1161                         if ($i == $len)
1162                                 break;
1163
1164                         $start = $i;
1165
1166                         // Find the end of this run of changes.
1167                         while (++$i < $len && $changed[$i])
1168                                 continue;
1169
1170                         do {
1171                                 /*
1172                                  * Record the length of this run of changes, so that
1173                                  * we can later determine whether the run has grown.
1174                                  */
1175                                 $runlength = $i - $start;
1176
1177                                 /*
1178                                  * Move the changed region back, so long as the
1179                                  * previous unchanged line matches the last changed one.
1180                                  * This merges with previous changed regions.
1181                                  */
1182                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1183                                         $changed[--$start] = 1;
1184                                         $changed[--$i] = false;
1185                                         while ($start > 0 && $changed[$start - 1])
1186                                                 $start--;
1187                                         USE_ASSERTS && assert('$j > 0');
1188                                         while ($other_changed[--$j])
1189                                                 continue;
1190                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1191                                 }
1192
1193                                 /*
1194                                  * Set CORRESPONDING to the end of the changed run, at the last
1195                                  * point where it corresponds to a changed run in the other file.
1196                                  * CORRESPONDING == LEN means no such point has been found.
1197                                  */
1198                                 $corresponding = $j < $other_len ? $i : $len;
1199
1200                                 /*
1201                                  * Move the changed region forward, so long as the
1202                                  * first changed line matches the following unchanged one.
1203                                  * This merges with following changed regions.
1204                                  * Do this second, so that if there are no merges,
1205                                  * the changed region is moved forward as far as possible.
1206                                  */
1207                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1208                                         $changed[$start++] = false;
1209                                         $changed[$i++] = 1;
1210                                         while ($i < $len && $changed[$i])
1211                                                 $i++;
1212
1213                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1214                                         $j++;
1215                                         if ($j < $other_len && $other_changed[$j]) {
1216                                                 $corresponding = $i;
1217                                                 while ($j < $other_len && $other_changed[$j])
1218                                                         $j++;
1219                                         }
1220                                 }
1221                         } while ($runlength != $i - $start);
1222
1223                         /*
1224                          * If possible, move the fully-merged run of changes
1225                          * back to a corresponding run in the other file.
1226                          */
1227                         while ($corresponding < $i) {
1228                                 $changed[--$start] = 1;
1229                                 $changed[--$i] = 0;
1230                                 USE_ASSERTS && assert('$j > 0');
1231                                 while ($other_changed[--$j])
1232                                         continue;
1233                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1234                         }
1235                 }
1236                 wfProfileOut( $fname );
1237         }
1238 }
1239
1240 /**
1241  * Class representing a 'diff' between two sequences of strings.
1242  * @todo document
1243  * @private
1244  * @addtogroup DifferenceEngine
1245  */
1246 class Diff
1247 {
1248         var $edits;
1249
1250         /**
1251          * Constructor.
1252          * Computes diff between sequences of strings.
1253          *
1254          * @param $from_lines array An array of strings.
1255          *                (Typically these are lines from a file.)
1256          * @param $to_lines array An array of strings.
1257          */
1258         function Diff($from_lines, $to_lines) {
1259                 $eng = new _DiffEngine;
1260                 $this->edits = $eng->diff($from_lines, $to_lines);
1261                 //$this->_check($from_lines, $to_lines);
1262         }
1263
1264         /**
1265          * Compute reversed Diff.
1266          *
1267          * SYNOPSIS:
1268          *
1269          *      $diff = new Diff($lines1, $lines2);
1270          *      $rev = $diff->reverse();
1271          * @return object A Diff object representing the inverse of the
1272          *                                original diff.
1273          */
1274         function reverse () {
1275                 $rev = $this;
1276                 $rev->edits = array();
1277                 foreach ($this->edits as $edit) {
1278                         $rev->edits[] = $edit->reverse();
1279                 }
1280                 return $rev;
1281         }
1282
1283         /**
1284          * Check for empty diff.
1285          *
1286          * @return bool True iff two sequences were identical.
1287          */
1288         function isEmpty () {
1289                 foreach ($this->edits as $edit) {
1290                         if ($edit->type != 'copy')
1291                                 return false;
1292                 }
1293                 return true;
1294         }
1295
1296         /**
1297          * Compute the length of the Longest Common Subsequence (LCS).
1298          *
1299          * This is mostly for diagnostic purposed.
1300          *
1301          * @return int The length of the LCS.
1302          */
1303         function lcs () {
1304                 $lcs = 0;
1305                 foreach ($this->edits as $edit) {
1306                         if ($edit->type == 'copy')
1307                                 $lcs += sizeof($edit->orig);
1308                 }
1309                 return $lcs;
1310         }
1311
1312         /**
1313          * Get the original set of lines.
1314          *
1315          * This reconstructs the $from_lines parameter passed to the
1316          * constructor.
1317          *
1318          * @return array The original sequence of strings.
1319          */
1320         function orig() {
1321                 $lines = array();
1322
1323                 foreach ($this->edits as $edit) {
1324                         if ($edit->orig)
1325                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1326                 }
1327                 return $lines;
1328         }
1329
1330         /**
1331          * Get the closing set of lines.
1332          *
1333          * This reconstructs the $to_lines parameter passed to the
1334          * constructor.
1335          *
1336          * @return array The sequence of strings.
1337          */
1338         function closing() {
1339                 $lines = array();
1340
1341                 foreach ($this->edits as $edit) {
1342                         if ($edit->closing)
1343                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1344                 }
1345                 return $lines;
1346         }
1347
1348         /**
1349          * Check a Diff for validity.
1350          *
1351          * This is here only for debugging purposes.
1352          */
1353         function _check ($from_lines, $to_lines) {
1354                 $fname = 'Diff::_check';
1355                 wfProfileIn( $fname );
1356                 if (serialize($from_lines) != serialize($this->orig()))
1357                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1358                 if (serialize($to_lines) != serialize($this->closing()))
1359                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1360
1361                 $rev = $this->reverse();
1362                 if (serialize($to_lines) != serialize($rev->orig()))
1363                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1364                 if (serialize($from_lines) != serialize($rev->closing()))
1365                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1366
1367
1368                 $prevtype = 'none';
1369                 foreach ($this->edits as $edit) {
1370                         if ( $prevtype == $edit->type )
1371                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1372                         $prevtype = $edit->type;
1373                 }
1374
1375                 $lcs = $this->lcs();
1376                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1377                 wfProfileOut( $fname );
1378         }
1379 }
1380
1381 /**
1382  * FIXME: bad name.
1383  * @todo document
1384  * @private
1385  * @addtogroup DifferenceEngine
1386  */
1387 class MappedDiff extends Diff
1388 {
1389         /**
1390          * Constructor.
1391          *
1392          * Computes diff between sequences of strings.
1393          *
1394          * This can be used to compute things like
1395          * case-insensitve diffs, or diffs which ignore
1396          * changes in white-space.
1397          *
1398          * @param $from_lines array An array of strings.
1399          *      (Typically these are lines from a file.)
1400          *
1401          * @param $to_lines array An array of strings.
1402          *
1403          * @param $mapped_from_lines array This array should
1404          *      have the same size number of elements as $from_lines.
1405          *      The elements in $mapped_from_lines and
1406          *      $mapped_to_lines are what is actually compared
1407          *      when computing the diff.
1408          *
1409          * @param $mapped_to_lines array This array should
1410          *      have the same number of elements as $to_lines.
1411          */
1412         function MappedDiff($from_lines, $to_lines,
1413                                                 $mapped_from_lines, $mapped_to_lines) {
1414                 $fname = 'MappedDiff::MappedDiff';
1415                 wfProfileIn( $fname );
1416
1417                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1418                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1419
1420                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1421
1422                 $xi = $yi = 0;
1423                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1424                         $orig = &$this->edits[$i]->orig;
1425                         if (is_array($orig)) {
1426                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1427                                 $xi += sizeof($orig);
1428                         }
1429
1430                         $closing = &$this->edits[$i]->closing;
1431                         if (is_array($closing)) {
1432                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1433                                 $yi += sizeof($closing);
1434                         }
1435                 }
1436                 wfProfileOut( $fname );
1437         }
1438 }
1439
1440 /**
1441  * A class to format Diffs
1442  *
1443  * This class formats the diff in classic diff format.
1444  * It is intended that this class be customized via inheritance,
1445  * to obtain fancier outputs.
1446  * @todo document
1447  * @private
1448  * @addtogroup DifferenceEngine
1449  */
1450 class DiffFormatter
1451 {
1452         /**
1453          * Number of leading context "lines" to preserve.
1454          *
1455          * This should be left at zero for this class, but subclasses
1456          * may want to set this to other values.
1457          */
1458         var $leading_context_lines = 0;
1459
1460         /**
1461          * Number of trailing context "lines" to preserve.
1462          *
1463          * This should be left at zero for this class, but subclasses
1464          * may want to set this to other values.
1465          */
1466         var $trailing_context_lines = 0;
1467
1468         /**
1469          * Format a diff.
1470          *
1471          * @param $diff object A Diff object.
1472          * @return string The formatted output.
1473          */
1474         function format($diff) {
1475                 $fname = 'DiffFormatter::format';
1476                 wfProfileIn( $fname );
1477
1478                 $xi = $yi = 1;
1479                 $block = false;
1480                 $context = array();
1481
1482                 $nlead = $this->leading_context_lines;
1483                 $ntrail = $this->trailing_context_lines;
1484
1485                 $this->_start_diff();
1486
1487                 foreach ($diff->edits as $edit) {
1488                         if ($edit->type == 'copy') {
1489                                 if (is_array($block)) {
1490                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1491                                                 $block[] = $edit;
1492                                         }
1493                                         else{
1494                                                 if ($ntrail) {
1495                                                         $context = array_slice($edit->orig, 0, $ntrail);
1496                                                         $block[] = new _DiffOp_Copy($context);
1497                                                 }
1498                                                 $this->_block($x0, $ntrail + $xi - $x0,
1499                                                                           $y0, $ntrail + $yi - $y0,
1500                                                                           $block);
1501                                                 $block = false;
1502                                         }
1503                                 }
1504                                 $context = $edit->orig;
1505                         }
1506                         else {
1507                                 if (! is_array($block)) {
1508                                         $context = array_slice($context, sizeof($context) - $nlead);
1509                                         $x0 = $xi - sizeof($context);
1510                                         $y0 = $yi - sizeof($context);
1511                                         $block = array();
1512                                         if ($context)
1513                                                 $block[] = new _DiffOp_Copy($context);
1514                                 }
1515                                 $block[] = $edit;
1516                         }
1517
1518                         if ($edit->orig)
1519                                 $xi += sizeof($edit->orig);
1520                         if ($edit->closing)
1521                                 $yi += sizeof($edit->closing);
1522                 }
1523
1524                 if (is_array($block))
1525                         $this->_block($x0, $xi - $x0,
1526                                                   $y0, $yi - $y0,
1527                                                   $block);
1528
1529                 $end = $this->_end_diff();
1530                 wfProfileOut( $fname );
1531                 return $end;
1532         }
1533
1534         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1535                 $fname = 'DiffFormatter::_block';
1536                 wfProfileIn( $fname );
1537                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1538                 foreach ($edits as $edit) {
1539                         if ($edit->type == 'copy')
1540                                 $this->_context($edit->orig);
1541                         elseif ($edit->type == 'add')
1542                                 $this->_added($edit->closing);
1543                         elseif ($edit->type == 'delete')
1544                                 $this->_deleted($edit->orig);
1545                         elseif ($edit->type == 'change')
1546                                 $this->_changed($edit->orig, $edit->closing);
1547                         else
1548                                 trigger_error('Unknown edit type', E_USER_ERROR);
1549                 }
1550                 $this->_end_block();
1551                 wfProfileOut( $fname );
1552         }
1553
1554         function _start_diff() {
1555                 ob_start();
1556         }
1557
1558         function _end_diff() {
1559                 $val = ob_get_contents();
1560                 ob_end_clean();
1561                 return $val;
1562         }
1563
1564         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1565                 if ($xlen > 1)
1566                         $xbeg .= "," . ($xbeg + $xlen - 1);
1567                 if ($ylen > 1)
1568                         $ybeg .= "," . ($ybeg + $ylen - 1);
1569
1570                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1571         }
1572
1573         function _start_block($header) {
1574                 echo $header;
1575         }
1576
1577         function _end_block() {
1578         }
1579
1580         function _lines($lines, $prefix = ' ') {
1581                 foreach ($lines as $line)
1582                         echo "$prefix $line\n";
1583         }
1584
1585         function _context($lines) {
1586                 $this->_lines($lines);
1587         }
1588
1589         function _added($lines) {
1590                 $this->_lines($lines, '>');
1591         }
1592         function _deleted($lines) {
1593                 $this->_lines($lines, '<');
1594         }
1595
1596         function _changed($orig, $closing) {
1597                 $this->_deleted($orig);
1598                 echo "---\n";
1599                 $this->_added($closing);
1600         }
1601 }
1602
1603
1604 /**
1605  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1606  *
1607  */
1608
1609 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1610
1611 /**
1612  * @todo document
1613  * @private
1614  * @addtogroup DifferenceEngine
1615  */
1616 class _HWLDF_WordAccumulator {
1617         function _HWLDF_WordAccumulator () {
1618                 $this->_lines = array();
1619                 $this->_line = '';
1620                 $this->_group = '';
1621                 $this->_tag = '';
1622         }
1623
1624         function _flushGroup ($new_tag) {
1625                 if ($this->_group !== '') {
1626                         if ($this->_tag == 'ins')
1627                                 $this->_line .= '<ins class="diffchange">' .
1628                                         htmlspecialchars ( $this->_group ) . '</ins>';
1629                         elseif ($this->_tag == 'del')
1630                                 $this->_line .= '<del class="diffchange">' .
1631                                         htmlspecialchars ( $this->_group ) . '</del>';
1632                         else
1633                                 $this->_line .= htmlspecialchars ( $this->_group );
1634                 }
1635                 $this->_group = '';
1636                 $this->_tag = $new_tag;
1637         }
1638
1639         function _flushLine ($new_tag) {
1640                 $this->_flushGroup($new_tag);
1641                 if ($this->_line != '')
1642                         array_push ( $this->_lines, $this->_line );
1643                 else
1644                         # make empty lines visible by inserting an NBSP
1645                         array_push ( $this->_lines, NBSP );
1646                 $this->_line = '';
1647         }
1648
1649         function addWords ($words, $tag = '') {
1650                 if ($tag != $this->_tag)
1651                         $this->_flushGroup($tag);
1652
1653                 foreach ($words as $word) {
1654                         // new-line should only come as first char of word.
1655                         if ($word == '')
1656                                 continue;
1657                         if ($word[0] == "\n") {
1658                                 $this->_flushLine($tag);
1659                                 $word = substr($word, 1);
1660                         }
1661                         assert(!strstr($word, "\n"));
1662                         $this->_group .= $word;
1663                 }
1664         }
1665
1666         function getLines() {
1667                 $this->_flushLine('~done');
1668                 return $this->_lines;
1669         }
1670 }
1671
1672 /**
1673  * @todo document
1674  * @private
1675  * @addtogroup DifferenceEngine
1676  */
1677 class WordLevelDiff extends MappedDiff
1678 {
1679         const MAX_LINE_LENGTH = 10000;
1680
1681         function WordLevelDiff ($orig_lines, $closing_lines) {
1682                 $fname = 'WordLevelDiff::WordLevelDiff';
1683                 wfProfileIn( $fname );
1684
1685                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1686                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1687
1688                 $this->MappedDiff($orig_words, $closing_words,
1689                                                   $orig_stripped, $closing_stripped);
1690                 wfProfileOut( $fname );
1691         }
1692
1693         function _split($lines) {
1694                 $fname = 'WordLevelDiff::_split';
1695                 wfProfileIn( $fname );
1696
1697                 $words = array();
1698                 $stripped = array();
1699                 $first = true;
1700                 foreach ( $lines as $line ) {
1701                         # If the line is too long, just pretend the entire line is one big word
1702                         # This prevents resource exhaustion problems
1703                         if ( $first ) {
1704                                 $first = false;
1705                         } else {
1706                                 $words[] = "\n";
1707                                 $stripped[] = "\n";
1708                         }
1709                         if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1710                                 $words[] = $line;
1711                                 $stripped[] = $line;
1712                         } else {
1713                                 $m = array();
1714                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1715                                         $line, $m))
1716                                 {
1717                                         $words = array_merge( $words, $m[0] );
1718                                         $stripped = array_merge( $stripped, $m[1] );
1719                                 }
1720                         }
1721                 }
1722                 wfProfileOut( $fname );
1723                 return array($words, $stripped);
1724         }
1725
1726         function orig () {
1727                 $fname = 'WordLevelDiff::orig';
1728                 wfProfileIn( $fname );
1729                 $orig = new _HWLDF_WordAccumulator;
1730
1731                 foreach ($this->edits as $edit) {
1732                         if ($edit->type == 'copy')
1733                                 $orig->addWords($edit->orig);
1734                         elseif ($edit->orig)
1735                                 $orig->addWords($edit->orig, 'del');
1736                 }
1737                 $lines = $orig->getLines();
1738                 wfProfileOut( $fname );
1739                 return $lines;
1740         }
1741
1742         function closing () {
1743                 $fname = 'WordLevelDiff::closing';
1744                 wfProfileIn( $fname );
1745                 $closing = new _HWLDF_WordAccumulator;
1746
1747                 foreach ($this->edits as $edit) {
1748                         if ($edit->type == 'copy')
1749                                 $closing->addWords($edit->closing);
1750                         elseif ($edit->closing)
1751                                 $closing->addWords($edit->closing, 'ins');
1752                 }
1753                 $lines = $closing->getLines();
1754                 wfProfileOut( $fname );
1755                 return $lines;
1756         }
1757 }
1758
1759 /**
1760  *      Wikipedia Table style diff formatter.
1761  * @todo document
1762  * @private
1763  * @addtogroup DifferenceEngine
1764  */
1765 class TableDiffFormatter extends DiffFormatter
1766 {
1767         function TableDiffFormatter() {
1768                 $this->leading_context_lines = 2;
1769                 $this->trailing_context_lines = 2;
1770         }
1771
1772         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1773                 $r = '<tr><td colspan="2" align="left"><strong><!--LINE '.$xbeg."--></strong></td>\n" .
1774                   '<td colspan="2" align="left"><strong><!--LINE '.$ybeg."--></strong></td></tr>\n";
1775                 return $r;
1776         }
1777
1778         function _start_block( $header ) {
1779                 echo $header;
1780         }
1781
1782         function _end_block() {
1783         }
1784
1785         function _lines( $lines, $prefix=' ', $color='white' ) {
1786         }
1787
1788         # HTML-escape parameter before calling this
1789         function addedLine( $line ) {
1790                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1791         }
1792
1793         # HTML-escape parameter before calling this
1794         function deletedLine( $line ) {
1795                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1796         }
1797
1798         # HTML-escape parameter before calling this
1799         function contextLine( $line ) {
1800                 return "<td> </td><td class='diff-context'>{$line}</td>";
1801         }
1802
1803         function emptyLine() {
1804                 return '<td colspan="2">&nbsp;</td>';
1805         }
1806
1807         function _added( $lines ) {
1808                 foreach ($lines as $line) {
1809                         echo '<tr>' . $this->emptyLine() .
1810                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1811                 }
1812         }
1813
1814         function _deleted($lines) {
1815                 foreach ($lines as $line) {
1816                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1817                           $this->emptyLine() . "</tr>\n";
1818                 }
1819         }
1820
1821         function _context( $lines ) {
1822                 foreach ($lines as $line) {
1823                         echo '<tr>' .
1824                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1825                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1826                 }
1827         }
1828
1829         function _changed( $orig, $closing ) {
1830                 $fname = 'TableDiffFormatter::_changed';
1831                 wfProfileIn( $fname );
1832
1833                 $diff = new WordLevelDiff( $orig, $closing );
1834                 $del = $diff->orig();
1835                 $add = $diff->closing();
1836
1837                 # Notice that WordLevelDiff returns HTML-escaped output.
1838                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1839
1840                 while ( $line = array_shift( $del ) ) {
1841                         $aline = array_shift( $add );
1842                         echo '<tr>' . $this->deletedLine( $line ) .
1843                                 $this->addedLine( $aline ) . "</tr>\n";
1844                 }
1845                 foreach ($add as $line) {       # If any leftovers
1846                         echo '<tr>' . $this->emptyLine() .
1847                                 $this->addedLine( $line ) . "</tr>\n";
1848                 }
1849                 wfProfileOut( $fname );
1850         }
1851 }
1852
1853 ?>