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