includes/DifferenceEngine.php

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