includes/DifferenceEngine.php

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