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