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