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